Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

New approaches for understanding urban greenspace using ecosystem services concepts and high spatial… Williams, David 2018

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2018_february_williams_david.pdf [ 3.14MB ]
Metadata
JSON: 24-1.0363005.json
JSON-LD: 24-1.0363005-ld.json
RDF/XML (Pretty): 24-1.0363005-rdf.xml
RDF/JSON: 24-1.0363005-rdf.json
Turtle: 24-1.0363005-turtle.txt
N-Triples: 24-1.0363005-rdf-ntriples.txt
Original Record: 24-1.0363005-source.json
Full Text
24-1.0363005-fulltext.txt
Citation
24-1.0363005.ris

Full Text

New approaches for understanding urban greenspace using ecosystem services concepts and high spatial resolution mapping by  David Williams  B.Sc., The University of British Columbia, 2015  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  MASTER OF SCIENCE in THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES (Forestry)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  December 2017  © David Williams, 2017 ii  Abstract Urban areas are where most people in the world directly benefit from ecosystem services (ES), yet there is evidence that ES are distributed inequitably with respect to socio-economic status which can lead to environmental injustices. There are a number of barriers to understanding UGS equity, some conceptual and some technical. One barrier is that few studies of environmental justice use similar quantitative methods, nor do they use concordant conceptual frameworks of UGS and ES equity. Another barrier is the fine scale information needed to accurately map greenspace can be difficult and expensive to obtain. In light of these barriers, this thesis seeks to contribute to UGS/ES equity studies in two fundamentally important ways.  First, I explore the concepts of equity in ecosystem services as applied to urban settings. I undertake a review of trans-disciplinary literature on urban systems to answer the question “How has environmental justice been considered and incorporated into urban ES research?” I characterize types of urban ES and measure the breadth of justice issues addressed in each article using a new environmental justice index (EJI). I also highlight the methods and results of key quantitative and qualitative papers that can inform future urban ES justice frameworks.   Second, I explore how new advances in remote sensing can better characterize UGS distributions via more accurate mapping of heterogeneous urban areas. I combine three-dimensional information from airborne Light Detection and Ranging (LiDAR) data with RapidEye high spatial resolution imagery in a Geographic Object-Based Image Analysis (GEOBIA) approach to classify urban landcover in a large metropolitan region. Though 5m RapidEye pixels were often mixed in urban areas, LiDAR data enabled accurate classification of fine spatial objects such as street trees and single-family dwellings.   Ultimately, I propose that mapping ES distributions among urban socio-demographic groups and assessing potential ES tradeoffs is not enough to avoid injustices. Because ES are socio-political constructs, gaining a comprehensive understanding of urban ES injustices is not merely a process of mapping greenspace, but also understanding how the groups in question ascribe value to the ES supply sources around them. iii  Lay Summary Chapter 2 is a review of justice issues of urban ecosystem services (ES) where I synthesize the state of the literature and develop an index measuring the breadth of justice issues in an article. I find the breadth of justice inquiry is not expanding with the growth in urban ES literature, and that ES justice issues must be considered when analyzing ES in cities. Chapter 3 presents a novel methodology for mapping urban landcover. The resultant map is created using high-resolution multispectral imagery and LiDAR data. Landcover class accuracies are high, and I find that there is a strong urban-rural greenness gradient. Finally, the conclusion synthesizes the previous chapters and sets the stage for future greenspace connectivity and greenspace equity analyses in Metro Vancouver.    iv  Preface My supervisors, Drs. Nicholas Coops and Sarah Gergel provided feedback on, and helped me to edit this manuscript, though I am the primary author of all sections.   Chapter 2 grew out of work that I pursued with my committee member Dr. Rob Kozak. The main line of inquiry and methodology is my own, but research questions and methods were refined with the help of Dr. Kozak and Dr. Gergel. I wrote the manuscript, with editing and feedback provided by both of my coauthors. A variant of this chapter is under review at the journal Ecosystem Services. Williams, David A.R., Kozak, Robert A., Gergel, Sarah E. (2018). Environmental justice and urban ecosystem services: A review of issues and methods of justice assessment. Ecosystem Serives (In Review).  Chapter 3 is based on a larger mapping project conceived jointly by Metro Vancouver and Dr. Coops. I worked closely with Dr. Coops to develop the mapping methodology, though many specific methodological decisions were mine, including the development of a GEOBIA-RandomForests workflow. My coauthor, Dr. Giona Matasci, was instrumental in writing scripts that could be adapted for implementing the RandomForests classifier. Dr. Matasci also counselled me in the interpretation of my results. I conducted all data processing and analysis, and wrote the manuscript with feedback and editing provided by Dr. Matasci, Dr. Coops and Dr. Gergel. Josephine Clark and Michael Coombes from Metro Vancouver also provided guidance and feedback on the mapping project throughout its development and implementation. A variant of this chapter is under review at the journal ISPRS Journal of Photogrammetry and Remote Sensing. Williams, David A.R., Matasci, Giona., Coops, Nicholas C. (2018). High-resolution urban landcover mapping using RapidEye and LiDAR. ISPRS Journal of Photogrammetry and Remote Sensing (In Review).  The conclusion alludes to work that further builds on and synthesizes Chapters 2 and 3. The methodology for this work has been developed with Dr. Gergel. The writing and data analysis has been undertaken by me.   v  Table of Contents  Abstract .......................................................................................................................................... ii Lay Summary ............................................................................................................................... iii Preface ........................................................................................................................................... iv Table of Contents ...........................................................................................................................v List of Tables ................................................................................................................................ ix List of Figures .................................................................................................................................x List of Abbreviations .................................................................................................................. xii Glossary ...................................................................................................................................... xiii Acknowledgements .................................................................................................................... xiv Dedication .....................................................................................................................................xv Chapter 1: Introduction ................................................................................................................1 1.1 Urban greenspace (UGS) provides ecosystem services to the majority of humanity ..... 1 1.2 Inequity in urban greenspace and the ecosystem services they provide is a global dilemma....................................................................................................................................... 3 1.3 The roots of environmental justice ................................................................................. 4 1.4 Mapping the distribution of UGS is a challenge ............................................................. 5 1.5 Thesis Overview ............................................................................................................. 6 Chapter 2: Cities and the danger of ecologists as technocrats: A quantitative review of issues and methods of justice assessment for urban ecosystem services. ..................................8 2.1 Introduction ..................................................................................................................... 8 2.2 Methods........................................................................................................................... 9 vi  2.2.1 Literature Search ......................................................................................................... 9 2.2.2 Article classification and ecosystem services tallying and encoding ......................... 9 2.2.3 Measuring the extent of environmental justice analysis: Creation of a justice index11 2.3 Results ........................................................................................................................... 13 2.3.1 ES mentioned and investigated ................................................................................. 13 2.3.2 Ecological justice index (EJI) ................................................................................... 18 2.3.3 Methods used for quantitative assessments .............................................................. 22 2.3.4 Qualitative Approaches and Conceptual Frameworks .............................................. 28 2.4 Discussion ..................................................................................................................... 30 2.4.1 Few elements of justice are usually addressed in urban ES research ....................... 30 2.4.2 Developing world urban ES justice literature is needed in the 21st century ............ 31 2.4.3 Quantitative justice assessment methods: Greenspace as ES proxy is low-hanging fruit …… ............................................................................................................................... 31 2.4.4 A message from geographers: The value of urban ES are contested, and can be used as a rationale for gentrification and dispossession ................................................................ 32 2.4.5 Urban green spaces provide ES that are often overlooked by ecologists ................. 33 2.4.6 Study Limitations ...................................................................................................... 34 2.5 Conclusion .................................................................................................................... 35 Chapter 3: High-resolution urban landcover mapping using RapidEye and LiDAR in a GEOBIA-RandomForests workflow ..........................................................................................36 3.1 Introduction ................................................................................................................... 36 3.2 Methods......................................................................................................................... 38 3.2.1 Study Location .......................................................................................................... 38 vii  3.2.2 Data ........................................................................................................................... 38 3.2.2.1 LiDAR processing ............................................................................................ 41 3.2.2.2 Class hierarchy development ............................................................................ 42 3.2.2.3 Spatial resolution and Minimum Mapping Unit (MMU) considerations ......... 42 3.2.2.4 Training Data Creation ..................................................................................... 43 3.2.3 Classification Workflow ........................................................................................... 45 3.2.3.1 GEOBIA expert ruleset creation and classification .......................................... 45 3.2.3.2 Random forests classification ........................................................................... 48 3.2.4 Accuracy Assessment ............................................................................................... 49 3.2.4.1 Validation Data ................................................................................................. 49 3.2.4.2 Accuracy Assessment ....................................................................................... 49 3.3 Results ........................................................................................................................... 49 3.3.1 LiDAR-RapidEye fusion in a GEOBIA workflow leads to high quantitative accuracies .............................................................................................................................. 49 3.3.2 Vegetation dominates landcover but tends to decrease from east to west ................ 52 3.4 Discussion ..................................................................................................................... 57 3.4.1 GEOBIA ruleset development can be tedious, but with significant payoff .............. 58 3.4.2 2m output resolution: an effective visual compromise to resolve buildings and trees ………................................................................................................................................... 60 3.4.3 The impact of secondary class assignments: flexibility and enhanced accuracy ...... 61 3.4.4 Temporal mismatch of data was a challenge, but not insurmountable ..................... 62 3.4.5 Analysis of landcover patterns: Vegetation dominates region but there are important differences among municipalities ......................................................................................... 63 viii  3.5 Conclusion .................................................................................................................... 65 Chapter 4: Conclusion .................................................................................................................66 4.1 Limitations of Research ................................................................................................ 68 4.2 Future Research ............................................................................................................ 68 4.3 Closing Thoughts .......................................................................................................... 70 References .....................................................................................................................................71 Appendices ....................................................................................................................................94 Appendix A ES Recoding Scheme Based On Wilkinson Et Al. (2013) ................................... 94 Appendix B Chapter 2 Annotated Bibliography ....................................................................... 96 Appendix C Object features exported for each type of object used for classification. ........... 153  ix  List of Tables Table 1 Attributes described for each journal article in the review during initial reading. .......... 10 Table 2 Elements of justice assessed in each paper. ..................................................................... 12 Table 3 Summary of papers reviewed that take a spatial and/or quantitative approach that employ or best inform urban ES justice assessment methodologies. ........................................................ 25 Table 4 Data used throughout the GEOBIA workflow. ............................................................... 40 Table 5 Final classification classes and classification criteria. ..................................................... 42 Table 6 Object Features used to create thresholds for the assign class algorithm in the GEOBIA rule set.. ......................................................................................................................................... 47 Table 7 Confusion matrix showing User's and Producer's Accuracies for each predicted class.. 50 Table 8 Top 20 RF classification variables ranked by importance as measured by the Mean Decrease in the Gini coefficient for ground-level and tree-canopy objects.. ................................ 50 Table 9 Area of landcover per class, and its percentage of total area (158 415 ha). .................... 54 x  List of Figures Figure 1 The number of study locations from each article ........................................................... 14 Figure 2 The number of study locations from each article classified as either Quantitative, Qualitative or Hybrid .................................................................................................................... 15 Figure 3 The number of articles published per year. .................................................................... 16 Figure 4 The number of ES and DS mentioned and investigated per year ................................... 17 Figure 5 ES investigated and not investigated as a proportion of the (recoded) total ES mentioned in each year ................................................................................................................................... 18 Figure 6 Frequency of EJI values of the articles reviewed  .......................................................... 19 Figure 7 Median EJI values per year ............................................................................................ 20 Figure 8 The frequency of EJI categories encoded for all articles reviewed ................................ 21 Figure 9 EJI values for Quantitative, Qualitative and Hybrid (both quantitative and qualitative) papers ............................................................................................................................................ 22 Figure 10 Study area boundary encompassing Metro Vancouver, Abbotsford and a 5km buffer........................................................................................................................................................ 39 Figure 11 LiDAR coverage of the study area ............................................................................... 40 Figure 12 Classification workflow showing input data, object segmentation, and classification stream ............................................................................................................................................ 45 Figure 13 Side-by-side comparison of classified building and tree objects in a single-family dwelling area of Burnaby, a municipality in Metro Vancouver ................................................... 46 Figure 14 Final landcover classification for LiDAR-coverage areas. .......................................... 53 Figure 15 From top to bottom, detail of a 1m nDSM, 5m RapidEye imagery and the 2m classification ................................................................................................................................. 55 xi  Figure 16 Proportion of landcover class by area for municipalities covered by the landcover classification ................................................................................................................................. 56   xii  List of Abbreviations ANT: Actor-Network Theory EDS: Ecosystem disservices EF: Ecological Footprint EJI: Environmental Justice Index ES: Ecosystem Services GEOBIA: Geographic Object-Based Image Analysis HDI: Human Development Index LiDAR: Light Detecting and Ranging MEA: Millennium Ecosystem Assessment MMU: Minimum Mapping Unit NDVI: Normalized Difference Vegetation Index SWIR: Short-Wave Infrared TEEB: The Economics of Ecosystems and Biodiversity UGS: Urban Greenspace UHI: Urban Heat Island xiii  Glossary LiDAR: Light Detection and Ranging is a technology used for mapping the three-dimensional structure of the Earth’s surface. It is also known as aerial laser scanning (ALS) when conducted from an aerial platform. Laser pulses are emitted and their return interval recorded by a sensor. Because the speed of light is a constant, the time of pulse return can be used to determine distance from the sensor. Accurate mapping of the Earth’s surface using LiDAR requires precise information about the position and orientation of the laser and sensor.   Multispectral sensor: Multispectral sensors are, in the context of remote sensing, passive optical sensors of the Earth’s surface that collect information in a number of electromagnetic (EM) bands. Common sensing bands are ‘atmospheric windows’ that can pass through the Earth’s atmosphere often ranging from the visible to the infrared portions of the EM spectrum.    xiv  Acknowledgements I would like to thank my supervisors, Dr. Nicholas Coops and Dr. Sarah Gergel, for their guidance, patience, and invaluable input. Thanks as well to Dr. Rob Kozak for agreeing to be on my committee and for his help preparing Chapter 2.   I would also like to thank Josephine Clark, Michael Coombes, others at Metro Vancouver and Nick Page at the City of Vancouver for their expert advice.   Thanks also to my folks, Miriam, members of the Integrated Remote Sensing Studio (IRSS) and Landscape Ecology Lab (LEL). Thanks to my friends that have hung around even though I haven’t. Special thanks to Trevor Luu and Lukas Jarron for photo-interpretation of ground truth data, as well as Piotr Tompalski and Giona Matasci for invaluable technical assistance. Thanks as well to Txomin Hermosilla for early research guidance and because his Pez collection might become sentient and hunt me down should I leave him out.   This research was supported by the University of British Columbia, Faculty of Forestry; Metro Vancouver; and the Natural Sciences and Engineering Research Council of Canada.  xv  Dedication  To my parents, Carol and Greg Williams, who have always supported me in every way. I love you both so much.   To Miriam, for sticking with me through it all.   And to all the citizens of every city everywhere – you have a right to your cities and a right to the greenspace in them.  You have the right to demand more.1  Chapter 1: Introduction Urban areas are home to over 50% of the global population (United Nations Department of Economic and Social Affairs Population Division 2014) and the primary cause of many global environmental problems, such as deforestation, pollution, and climate change (Weng 2014). Cities occupy only 3% of the global landscape (UN 2008), but are responsible for three quarters of global energy consumption and approximately 80% of anthropogenic greenhouse gas emissions (Ash et al. 2008). In Northern America, over 80% of the population live in cities (United Nations Department of Economic and Social Affairs Population Division 2014), and in Canada over 50% of the population live in just three metropolitan regions: Toronto, Montréal and Vancouver (Statistics Canada 2017). With cities expected to contain a further 2.5 billion people by 2050 (United Nations Department of Economic and Social Affairs Population Division 2014), a comprehensive understanding of the urban environment is fundamental to ensure sustainable and adaptive urban ecosystems.  1.1 Urban greenspace (UGS) provides ecosystem services to the majority of humanity Scholars of urban ecology argue that urban areas constitute novel social-ecological systems requiring an ecology of and for cities, which differs from merely conducting ecological research oft-suited to ‘natural’ ecosystems within cities (Childers et al. 2015; Steward T. A. Pickett et al. 2016). Cities are highly modified systems, with large material flows and external inputs compared to less-modified ecosystems (Agudelo-Vera et al. 2012; S.T.A. Pickett et al. 2013). A key characteristic of cities as social-ecological systems is extreme heterogeneity of land covers, habitat types and, often, of human inhabitants (Cadenasso, Pickett, and Schwarz 2007; S. T. A. Pickett et al. 2016). This social-ecological heterogeneity makes cities challenging to study and understand.  Urban greenspace (UGS) is a particularly heterogeneous, yet fundamental, feature of urban social-ecological systems. The definition of UGS has been shown to vary widely among disciplines and researchers (Taylor and Hochuli 2017). Here, however, UGS is urban and peri-urban vegetation including remnants of native vegetated communities, planted, introduced and invasive species of grasses, trees and shrubs. UGS can take the form of parks, golf courses, street 2  trees, lawns, private yards, gardens and agriculture, wetlands, green infrastructure such as bio-swales, green roofs and constructed wetlands, vacant lots and otherwise unmanaged vegetation (i.e. informal greenspace) (Colding et al. 2013; Davis et al. 2012; Faehnle et al. 2014; Kabisch et al. 2016; de la Barrera, Reyes-Paecke, and Banzhaf 2016; La Rosa 2014; Rupprecht et al. 2015; Wolch, Byrne, and Newell 2014a; Young 2010). UGS often exists as a patchwork of different vegetation cover types with a mix of fragmentation and connectivity amidst roads, pavement, infrastructure and buildings (Cen et al. 2015; Irwin and Bockstael 2007). Fragmentation of UGS can reduce the dispersal ability of plants and animals in the short term and the flow of genes and viability of metapopulations over longer time frames (Angelone and Holderegger 2009; Clergeau and Burel 1997; Dixo et al. 2009; Ferreras 2001; Hanski and Gilpin 1991). While ecologists have explored the impact of UGS fragmentation on a variety of animal species, the heterogeneous spatial patterns of UGS can also impact a wide variety of ecosystem services (ES) for people, including access to parks and greenspace (Alberti 2010; Gaston, Ávila-Jiménez, and Edmondson 2013a; Irwin and Bockstael 2007; Lepczyk et al. 2017; N. Schwarz et al. 2017).   At the same time, this spatial dynamism means that relatively ‘wild’ patches of UGS can be found throughout cities, sometimes in relatively small interstitial spaces between different landuses (Rupprecht et al., 2015; Steward T. A. Pickett et al., 2016; Threlfall & Kendal, 2017). Novel combinations of native and introduced species create unique biodiversity patterns in cities compared to their surrounding environments, producing new ecologies that can be commensal with human settlement (Morse et al. 2014). Along with the insects, birds and other species that use and live in these urban greenspace patches, humans benefit from ecosystem services (ES) that flow from UGS patches as a result of ecosystem functioning (Alberti 2010; Wu 2014).  Access to UGS improves health and well-being via reductions in air/noise pollution, heat stress, stress and anxiety (Suich, Howe, and Mace 2015; WHO Regional Office for Europe 2016; Wolch, Byrne, and Newell 2014b). Ecosystem disservices (EDS) causing inconvenience or even harm can include maintenance costs of street trees, storm and property damage, and allergens (Reviewed in von Döhren and Haase 2015).   3  Because globally most of humanity now lives in cities, understanding the delivery of ES by urban trees and greenspace is a research imperative (Gaston, Ávila-Jiménez, and Edmondson 2013a). However, the definition of what constitutes an ES is contested in a proliferation of ES frameworks (Díaz et al. 2015; Haines-Young and Potschin 2011; Luederitz et al. 2015; Villa, F., Bagstad, K. J., Voigt, B., Johnson, G. W., Athanasiadis, I. N., & Balbi 2014) and made even more challenging when applied to urban situations, thus complicating the exploration of urban ES. Perhaps the simplest definition of ES are the benefits that humans derive from ecosystems (Villa et al. 2014). The Millennium Ecosystem Services Assessment (MEA) divides ES into four broad categories of supporting, regulating, provisioning and cultural services (Fisher, Turner, and Morling 2009; ICSU, UNESCO, and UNU 2008). Building on the MEA, the TEEB project (The Economics of Ecosystems and Biodiversity) (TEEB 2011) specifically provides an “urban” ES framework directed at planners and policy-makers. The Intergovernmental Platform on Biodiversity and Ecosystem Services (IPBES) also includes equity and just governance (Díaz et al. 2015). Thus, issues of equity and inequity are particularly compelling to examine within the context of ecosystem services derived from UGS.   1.2 Inequity in urban greenspace and the ecosystem services they provide is a global dilemma   UGS tends to be distributed unevenly in cities – often along wealth, class and racial lines – with those better off socio-economically also living in greener environments (e.g. Dobbs, Kendal, and Nitschke 2014; Jesdale, Morello-Frosch, and Cushing 2013; Lakes, Brückner, and Krämer 2014; Wolch, Byrne, and Newell 2014a). At the same time, the poorest and most vulnerable often rely more heavily on ES than those that are better off (e.g. Jennings and Gaither 2015; Shackleton et al. 2015; Ward Thompson et al. 2012). For example, residents who cannot afford air conditioning might particularly benefit from the cooling services of nearby trees.   UGS also provides a suite of natural disaster-mitigating ES such as flood mitigation (Gómez-Baggethun and Barton 2013; McPhearson et al. 2015). Because anthropogenic climate change is leading to increased incidence of extreme weather events worldwide (IPCC 2014), bolstering 4  urban resilience to shocks from such events is perhaps more important than ever before – not only to safeguard cities for their inhabitants but in the interest of climate justice (Meerow, Newell, and Stults 2016) If the most vulnerable in a population tend to have the least access to UGS and associated ES, they might also be the most vulnerable to the effects of natural disasters. Indeed, urban UGS equity and, by extension, ES equity is a global environmental justice issue.   1.3 The roots of environmental justice Environmental justice is concerned with the distribution of environmental benefits and costs among different groups (Ernstson 2013a; Schlosberg 2007). It is, therefore, chiefly concerned with distributive justice, i.e. how best to distribute goods in a society that yields a socially just outcome. The problem of distributive justice, and a potential solution, is perhaps most famously articulated by the philosopher John Rawls (1971).  Rawls’ work has been critiqued by a number of philosophers (e.g. Bloom 1991; Nozick 1993; Wolff 1977), and it is important to note the efforts of Martha Nussbaum and Amartya Sen in further developing the philosophical underpinnings of distributive and environmental justice through the capabilities approach (Reviewed in Johnson 2012; Nussbaum and Sen 1993; Schlosberg 2007; Sen 2013).  However, my goal here is not to undertake a philosophical review, but rather provide adequate context about the origins of environmental justice.   Beginning in the 1960s and 70s the environmental and civil rights movements in the United States began to overlap, and by the 1980s the environmental justice movement was born (Reviewed in Ernstson 2013b; Reviewed in Tooke, Klinkenberg, and Coops 2010). It was initially primarily concerned with minority groups such as black and latinx people that were bearing the brunt of the emissions from polluting industries (i.e. environmental costs) (Mohai, Pellow, and Roberts 2009). Over time, however, the environmental justice movement has become concerned with the just distribution of environmental benefits among groups as well as costs. This has paved the way for scholars and activists to interrogate how UGS and ES are distributed in cities, which I attempt to add to here.  5  Vancouver, BC, which is the focus of Chapter 3, is often lauded for its high quality of life, however, it has the second-most unaffordable housing in the world (Cox and Pavletich 2014). Income and health outcomes have also previously been shown to be positively correlated with greenspace derived from remote sensing imagery (Crouse et al. 2017; Tooke, Klinkenberg, and Coops 2010). Given this evidence, issues of urban greenspace equity seem compelling to explore for the metropolitan Vancouver region.  1.4 Mapping the distribution of UGS is a challenge To understand the unequal distribution of UGS, mapping is essential. Mapping of urban areas is also a key priority for city managers and planners, yet is complicated by many factors. Metropolitan regions are often highly heterogeneous and complex (Cadenasso, Pickett, and Schwarz 2007; McGrath and Pickett 2011; S. T. A. Pickett et al. 2016; Wentz et al. 2014). This can lead to lower classification accuracies in urban areas, which require high resolution imagery to map effectively (Cadenasso, Pickett, and Schwarz 2007; Momeni, Aplin, and Boyd 2016; Pu and Landry 2012).   Furthermore, management of urban regions is decentralized at times, with municipalities not necessarily working in coordination, and with challenges that come with differing rates of sprawl in suburban and urban cores (Irwin and Bockstael 2007; UN Habitat 2013). Often, separate maps are maintained by a range of organisations and governments from different municipalities. As a result, region-wide modelling of urban growth and land-use change, in order to understand changes in greenspace losses, increases and/or fragmentation, presents significant challenges. Regional planners and economists, who must increasingly engage in regional planning processes over larger areas with diverse stakeholders, lack the geospatial information they need, or lack continuous coverage for large areas created with similar assumptions.   While there are significant challenges in creating accurate, comprehensive, and verified spatial layers of metropolitan regions, remote sensing offers a unique and critical approach to map urban landscapes. Indeed, extracting and classifying urban landcover types using satellite imagery at various spatial resolutions has been a major theme in urban landscape studies (i.e. Schneider 6  2012; Frolking et al. 2013; Castrence et al. 2014; Chen et al. 2015; Lin et al. 2014). Urban land cover is important for inferring landuse, mapping ecosystem services, or modelling complex processes like air quality, hydrology, or carbon stocks and flows (Reviewed in Wentz et al. 2014). Landcover and its change over time may also help with urban sustainability assessments (Kennedy, Cuddihy, and Engel-yan 2007; Moore, Kissinger, and Rees 2013).  1.5 Thesis Overview There are a number of barriers to understanding UGS equity, some conceptual and some technical. One barrier is the fine scale information needed to accurately map greenspace can be difficult and expensive to obtain (Alves and Ojima 2012; Lu, Coops, and Hermosilla 2017; Wolch, Byrne, and Newell 2014a). Another barrier is that few studies of environmental justice use similar quantitative methods, nor do they use concordant conceptual frameworks of UGS and ES equity. As such, this thesis seeks to contribute to UGS/ES equity studies in two fundamentally important ways. First, I explore the concepts of equity in ecosystem services as applied to urban settings. Second, I explore ways that new advances in remote sensing can better characterize UGS distributions via more accurate mapping, which is needed to address heterogeneous urban social-ecological systems.  In Chapter 2, I ask the question “How has environmental justice been considered and incorporated into urban ES research?” I conduct a systematic and qualitative literature review to catalogue conceptual frameworks for understanding urban ES justice and methods for justice inquiry/assessment. I create a novel justice index to assess the breadth of justice issues addressed in academic journal articles and summarize key quantitative methodologies for analysing UGS equity. These methods are synthesized with more qualitative approaches for a comprehensive overview of urban ES justice issues that I argue ecologists and planners must take into account.  In Chapter 3, I present a method for mapping urban land cover in Metro Vancouver, BC, using high spatial resolution approaches. High spatial resolution approaches include a fusion of Light Detection and Ranging (LiDAR) and multispectral satellite image data in an object-based workflow. The resulting landcover map achieved high individual class accuracies and can be 7  used as the basis for further work assessing UGS equity for this large, heterogeneous metropolitan region.   8  Chapter 2: Environmental justice and urban ecosystem services: A review of issues and methods of justice assessment.  2.1  Introduction Despite their importance, urban greenspace and associated ES are regularly distributed inequitably, often positively correlated with socio-economic status (e.g. Dobbs, Kendal, and Nitschke 2014; Lakes, Brückner, and Krämer 2014; K. Schwarz et al. 2015; Steenberg et al. 2015). For example, the greatest canopy cover occurs in the wealthiest Toronto neighbourhoods (Steenberg et al. 2015). In Baltimore, historical wealth corresponds better to contemporary canopy cover than contemporary wealth (2014). Despite such patterns, urban ES literature has inadequately addressed environmental justice, with many authors noting the importance of justice and equity concerns without investigating them thoroughly. Though there are literature reviews addressing issues of ES justice, such as Daw et al.’s investigation of the links between ES and poverty alleviation (2011) and Wolch, Byrne and Newell’s examination of justice impacts of urban greening initiatives (2014b), to my knowledge there has not been a review focused on methods for assessing urban ES justice.  The IPBES framework notwithstanding (Díaz et al. 2015), none of the primary ES frameworks explicitly incorporate methods for assessing the environmental justice and inequality of ES. Chan and Satterfield have used the MEA framework itself as a basis from which to explore justice and equity issues surrounding conservation initiatives (Chan and Satterfield 2013). In their review, Luederitz et al. (2015) mention stakeholder involvement and limited transferability of ES benefits, but do not directly tackle justice or equity. Though there are literature reviews addressing specific ES justice issues, such as Daw et al.’s links between ES and poverty (2011) and Wolch, Byrne and Newell’s justice impacts of urban greening initiatives (2014b), to my knowledge there has not been a review focused on methods for assessing urban ES justice.  Within this context, I ask “How has environmental justice been considered and incorporated into urban ES research?” To address this, I undertook a systematic review of trans-disciplinary literature on urban systems, using a broad and inclusive definition of ES as encompassing 9  benefits, values and disservices. I characterized types of urban ES and whether justice was examined quantitatively or qualitatively. I also highlighted the methods and results of key quantitative and qualitative papers that could be used to inform future urban ES justice frameworks. Furthermore, I discriminated among 10 different notions of environmental justice that were then reflected in a scoring system. My ancillary goal, therefore, was creating a systematic method for assessing the breadth of justice issues tackled in academic literature.  2.2  Methods 2.2.1 Literature Search A search was conducted using Thomsom Reuters’ Web of Science in November, 2015 using a search string modified from the syntax of Luederitz et al. (2015): TS=((ecosystem servic* OR ecosystem functio OR green infrastruc*) AND (urban OR city OR cities OR periurban OR town) AND (equity* OR equal* OR inequal* OR justice OR politic*)). The search returned 128 results with no results prior to the year 2000. The title and abstract of each result were read for relevance based on whether there was an explicit or implicit treatment of justice or equity issues related to urban ecosystem services. 60 journal articles were selected after reading titles and abstracts.   2.2.2 Article classification and ecosystem services tallying and encoding All 60 articles were assessed for 20 attributes (Table 1), both qualitative – such as location, and type of justice/equity issues raised – and quantitative, such as number of ES mentioned and/or explicitly investigated. Each article was then classified as to whether it employed primarily quantitative (using geospatial, statistical and other quantitative methods), qualitative (using literature review and analysis, conceptual frameworks and reasoned argument without substantial quantitative methods employed), or hybrid (a relatively balanced mixture of quantitative and qualitative methods) methodologies.   ES were described based on an article’s explicit or implicit description of the benefits of ecosystems provided to humans. This criterion was purposefully broad so ES in articles using relatively unique or implied conceptions of ES could still be categorized, enhancing the number 10  and types of ES codified. Ecosystem Disservices (EDS), ecosystem functions that yield negative results or harm for people, were tallied similarly to ES.  Further classification and encoding of ES were performed in order to ensure articles were directly comparable. First, ES types mentioned in articles were reclassified using a modified scheme from Wilkinson et al. (2013) (Appendix A: ES Recoding Scheme). Recoding enabled direct comparison of the numbers of ES mentioned, albeit via different names. ES investigated in each article were not recoded to maintain their identities; recoding would collapse some ES into a single category.  Table 1 Attributes described for each journal article in the literature review during initial reading.   † Justice-related attributes ‡ ES-related attributes Qualitative Quantitative Summary Spatial Scale Justice/Equity Issues Raised† Temporal Scale Justice Assessment Method/Test† Population Location Number of ES Mentioned‡ Spatial Scale Number of EDS Mentioned‡ Temporal Scale Number of ES Investigated‡ Socio-Economic Indicators Number of EDS Investigated‡ Data Types  Research Questions  Hypotheses/Assertions  ES Implied, Explicitly Mentioned, Or Measured/Investigated‡   EDS Implied, Mentioned, Measured/Investigated‡  ES Types‡ 11   2.2.3 Measuring the extent of environmental justice analysis: Creation of a justice index To quantitatively summarize justice issues in each article, a simple environmental justice index (EJI) was created using 10 elements of justice. Diverse elements of justice were mentioned across 60 articles, so the 10 elements are relatively broad categories. Three elements are reserved for explicit mentions of terms or concepts: (1) Environmental Justice, (2) Social Justice, and (3) Equality vs. Equity. Seven more elements required more interpretation of text and included: (4) Equality or Equity, (5) Distributive Justice, (6) Procedural/Governance Justice, (7) Interactional/Personal Justice, (8) Access, (9) Historical Inequality, and (10) Test/Measure. Table 2 provides a summary of the EJI categories.  A binary encoding of the elements mentioned or investigated in each article was performed (1=yes, 0=no). These values were then summed and normalized by the number of elements (i.e. divided by 10) to yield a number between 0 and 1, with higher values representing a more comprehensive investigation of justice issues.   The environmental and social justice categories (1 and 2) were tallied only if they were explicitly mentioned. The papers that mention environmental and social justice allude to broader justice movements at least enough to utilize the terms, if not also to provide definitions and frameworks for thinking about justice issues. Explicit references to environmental and social justice, therefore, were deemed to signify a broader understanding of justice thought and not just isolated incidences of injustice particular to the cases included in each paper.  Ethnicity Theory, the theory that different ethnic groups use parks in different ways (Wolch, Byrne, and Newell 2014b), is included under Access (7). Green spaces or other ES provisioning areas might not be used by groups due to cultural preferences, i.e. certain ethno-cultural groups may not access the benefits of urban ES even if they are proximal, because of preferred methods of, or settings for, ES access. Aspects of Ethnicity Theory fall under the umbrellas of distributive and procedural justice as well, especially concerning the need to involve ethno-cultural groups in, for example, green space planning. However, as a special case often mentioned in the 12  literature, it was deemed necessary to include mentions of Ethnicity Theory with the Access element of justice.  Historical injustices (9) can be distributional, procedural and interactional, so it could be argued that their encoding should fall under one of those elements. However, an historical perspective increases the depth of justice considerations, going beyond simple analyses of equality of distribution in space by addressing how the distribution of environmental costs and benefits has changed through history, and as a product of history – often due to power dynamics arising from colonialism and/or institutionalized racism.  Table 2 Elements of justice assessed in each paper. Each paper was rated using an Environmental Justice score from 0 - 1.0 based on the number of elements it mentioned and/or investigated.   Element # Justice Element Description/Rationale for Inclusion EJI Criteria Source 1 Environmental Justice Fair/equitable access to a healthy/clean environment and its associated benefits. Fair/equitable protection and freedom from environmental harms, such as pollution, among social groups. Mentioned explicitly Kabisch and Haase (2014)  2 Social Justice Social justice is commonly used by those addressing justice issues with respect to particular, often vulnerable, social groups (defined by, for example, race, gender, sexuality, class). It can be understood as the concept and study of how fairly benefits and burdens are distributed among groups in a society, often based on some allocation principle. A broad understanding of social justice includes distributive, procedural and interactional justice. What constitutes social justice and its application is still widely disputed. Mentioned explicitly Jost and Kay (2010) 3 Equality vs. Equity Included as a separate element because disadvantaged social groups might require more access to ES due to differential need, so that even where ES distribution is exactly equal it could still be inequitable. Mentioned explicitly Davis et al. (2012), Lakes et al. (2014) 4 Equality OR Equity General comments about equality or equity. These comments were mostly not specific enough to be placed in one of the other columns. Mentioned and/or implied. - 5 Distributive justice Fair allocation of resources among social groups. Generally understood to be articulated by John Rawls. Mentioned and/or implied. Jost and Kay (2010), Kabisch and Haase (2014), Low (2013), Rawls (1971) 6 Procedural justice Fair involvement of stakeholder/affected groups into planning and policy formulation processes. Mentioned and/or implied. Jost and Kay (2010), Kabisch and Haase (2014), Low (2013), Rawls (1971) 13  Element # Justice Element Description/Rationale for Inclusion EJI Criteria Source 7 Interactional justice Fairness of treatment when accessing resources, especially during interpersonal interactions. Mentioned and/or implied. Jost and Kay (2010), Kabisch and Haase (2014), Low (2013), Rawls (1971) 8 Access Issues of barriers to accessing ES in space and time due to poor public access, few easily accessible entrances, low public transit access, or ethno-cultural barriers to use. The issue of access is distinct from distributive justice, because regardless of where ES are located their flows must be accessible to people in order for services to be realized. Procedural and interactional justice are distinct from access as well, with procedural justice relating more to planning for and governance of ES, and interactional justice relating to how individuals are treated once they are already accessing ES. Ethnicity Theory is included here. Mentioned and/or implied. e.g. Booth et al. (2010), Jenerette et al. (2011), Wolch et al. (2014b) 9 Historical injustices Historical injustices can be distributional, procedural and interactional. Many authors highlight how historical injustices are often significant determinants of contemporary injustices, and therefore an important component of understanding the distribution of benefits and burdens seen among social, economic and ethno-cultural groups today. Mentioned and/or implied. e.g. Byrne and Wolch (2009), Safransky (2014) 10 Test/Measure The Test/Measure element was included to capture articles that provided a discursive or quantitative test or measure of justice in their analyses. Attempting to quantify the equitability of ES distribution goes beyond simple discussions of justice issues to provide the empirical basis for ES justice discussion and policy. Quantitative or qualitative method(s) to assess urban ES justice are put forward. Authors can explicitly state or allude to the methods as capable of being used to test the justice of distribution of ES. -  2.3 Results 2.3.1 ES mentioned and investigated Of 60 articles reviewed, the majority (57%) were from Europe, North America and Australia (Figure 1). Asia was primarily represented by case-studies in Chinese cities, with the exception of Baguio City in the Philippines (Estoque and Marayama (2014), and one pan-Asian approach (Brown, Dayal, and Rumbaitis Del Rio (2012). Africa, too, was poorly represented outside South Africa. For articles with Asian study locations, more than twice the number of quantitative articles were reviewed than qualitative articles, while there were no purely quantitative articles with African study locations (Figure 2). There was an even balance between quantitative and qualitative articles with European study locations, and for North American study locations there were slightly more qualitative than quantitative articles. 14   Figure 1 The number of study locations (n=71) from each article. Note that there are more locations than articles (n=60). The majority of studies were based in North America (31%) or Europe (20%), with the developing world most represented by Asian studies (17% - which were predominantly Chinese), and South African studies (10%). More studies (4%) had no explicit location (for example the modelling studies), than were set in the rest of Africa (3%), though with South African studies included African studies make up 13% of the total. 15   Figure 2 The number of study locations from each article classified as either Quantitative, Qualitative or Hybrid. All three article types have study locations in Asia, Europe and North America while in other areas there were fewer, and fewer types of articles. Articles set in Asia were primarily quantitative, European articles had a balance between quantitative and qualitative papers, while North American studies, the most numerous, had slightly more qualitative than quantitative articles. No primarily quantitative studies were conducted in Africa, including South Africa.  The number of articles published per year (Figure 3) steadily increased until 2014. Similarly, the total number of ES mentioned and investigated per paper increased over time (Figure 4). The number of ES mentioned and investigated peaked in 2013 and then declined despite more articles published the following year. The total number of ES mentioned (43), was the same before and after recoding, however the relative numbers of ES did change depending on the article. Fewer ES were investigated than mentioned in every year. Of the total ES mentioned, the proportion investigated (Figure 5) was always less than half, peaking at 0.4 in 2012 16  Far fewer EDS than ES were either mentioned or investigated (Figure 4) but EDS did increase over time. With the exception of Heynen’s article (2003), EDS weren’t mentioned until 2010, and though EDS mentions increased after that, 45 articles (75%) still did not discuss EDS at all.   Figure 3 The number of articles published per year (µ = 5.4, σ = 4.7). The most articles, 16, were published in 2014. The number of articles published on the justice of urban ES has tended to increase each year (2015 was the year of literature search). 17   Figure 4 The number of ES and DS mentioned and investigated per year. The range of ES mentioned per year is 4-164 with µ = 47 and σ = 52.3. The range of ES investigated is 0-45 with µ = 11.3, σ = 16. ES investigated are somewhat correlated with ES mentioned (rs = 0.38) indicating that articles that mention many ES are somewhat likely to also investigate or measure ES. The maximum number of EDS mentioned per year is 16, with σ = 6.0 across all years. EDS investigated vary between 0 and 8 (σ = 2.4). EDS investigated are quite correlated with ES investigated (rs = 0.68) indicating that articles that mention many EDS are more likely to also investigate or measure EDS. 18   Figure 5 ES investigated and not investigated as a proportion of the (recoded) total ES mentioned in each year. The proportion of ES investigated achieved a maximum of 0.4 in 2012 (µ = 0.15, σ = 0.15).    2.3.2 Ecological justice index (EJI) The majority (72%) of articles’ EJI scores were ≤0.4 (Figure 6) with the shape of the distribution slightly positively skewed. Median annual EJI (Figure 7) was quite similar to the median EJI value across all articles of 0.35 (σ = 0.2). Only 3 articles (5%) received an EJI score of 0.8, the maximum value recorded. The most common EJI variables were “Equity or Equality” and “Distributive Justice”, while the least frequently tallied EJI variable was “Equity vs. Equality” (Figure 8). No statistically significant difference was found among the EJI scores of qualitative, quantitative or hybrid papers, though quantitative papers did have the largest range of values and qualitative papers had the highest median EJI (Figure 9).  19   Figure 6 Frequency of EJI values of the articles reviewed (µ = 0.36, σ = 0.20). The majority of articles (72%) have values 0.4 or below indicating a narrow justice focus for most articles. 13 (22%) of the articles scored a value of 0.4, the modal value. The distribution was positively skewed with D’Agostino skewness = 0.29 and Anscombe-Glynn kurtosis = 2.74.    20   Figure 7 Median EJI values per year. No clear trend was found, with average values each year close to the mean EJI value for all articles of 0.35 (σ = 0.20).  21    Figure 8 The frequency of EJI categories encoded for all articles reviewed. The most common two categories are 4 and 5, corresponding to “Equity or Equality” and “Distributive Justice,” respectively. The other categories are (3) Equality vs. Equity, (6) Procedural/Governance, (2) Social Justice, (10) Test/Measure, (9) Historical Inequality, (1) Environmental Justice, (8) Access, and (7) Interactional/Personal.  22   Figure 9 EJI values for Quantitative (n=29), Qualitative (n=24) and Hybrid (n=7) (both quantitative and qualitative) papers. A Kruskall-Wallis rank sum test showed no significant difference among EJI values for the different paper types ( χ2 = 6.6728, p-value = 0.08309, α = 0.05).  2.3.3 Methods used for quantitative assessments Of the articles reviewed, 12 were deemed to have methods that can be used as, or to inform, tests of the justice of urban ES distribution (BenDor and Stewart 2011; Booth, Gaston, and 23  Armsworth 2010; Cohen et al. 2012; Davis et al. 2012; Ibes 2015; Jenerette et al. 2011; Kabisch and Haase 2014; Lakes, Brückner, and Krämer 2014; McDonald, Forman, and Kareiva 2010; Peng et al. 2015; Yao et al. 2014; Yin et al. 2007). Each of these papers was determined to be a quantitative paper, with no qualitative or “hybrid” papers fulfilling the “Test/Measure” criteria. A full annotated bibliography summarizing the 60 papers of the review, and detailing their treatment of justice issues, can be found in Appendix B  .   Many of the articles used greenspace as a proxy for urban ES. Greenspace was often derived from Geographic Information Systems (GIS) and/or remote sensing data. There was wide variation in how greenspace was used to define ES, but in general greenspace was used as a proxy for some combination of ES location, abundance and quality.   Use of spatially-explicit (mapped) greenspace as an ES proxy enabled integration with socioeconomic data. Censuses and surveys were used to identify ES beneficiaries and socio-economic gradients of ES benefits flows (see Table 3 for more details). Most common was statistical clustering and correlation analyses to identify relationships between groups of people and/or ES sources. Spatial overlays and proximity analyses were some of the simplest methods used to draw conclusions from relationships between green space and socio-demographic data (e.g. McDonald, Forman, and Kareiva 2010; Yin et al. 2007).  Using proximity analysis, the distance to ES and EDS can be compared among socio-economic groups or areas of a city. Examples include the BenDor and Stewart paper (2011), which compares the location of wetland impact and mitigation sites with the socioeconomics of the populations surrounding those sites, as well as Yin et al.’s (2007) accessibility analysis. Cohen et al. (2012), Davis et al.(2012), Ibes (2015), and Kabisch and Haase (2014) all used statistical clustering techniques, explained in more detail in Table 3 and Appendix B  .  Indices were used extensively to quantify the abundance and quality of ES source areas, as well as to measure the equity of their distribution. Jenerette et al., Kabisch and Haase and McDonald, Forman and Kareiva (2010) all employed the Gini index, while Lakes, Brückner, and Krämer 24  (2014) employed an environmental justice index of their own creation. Peng et al. (2015) used the Human Development Index (HDI) as a singular socio-demographic measurement, and combined that with Ecological Footprint (EF) measures. They then employed Gini to measure the equitability of the distribution of the HDI and EF measure. Yao et al. (2014), meanwhile, developed their own green space measure in the Effective Green Equivalent (EGE), creating an inequality coefficient for the EGE based on the Gini coefficient.    The majority of the authors in Table 3 found evidence that the distribution of greenspace or ES prompted equity concerns. A minority, however, labelled these distributions unjust with the exception of Davis et al. (2012) and Kabisch and Haase (2014) who claimed their results showed evidence of environmental injustice. In contrast, Peng et al.(Peng et al. 2015), Yao et al. (2014) and Yin et al. (2007) all asserted their results showed fairly equitable greenspace/ES distributions (though see Table 3 and Appendix B  for caveats).  25  Table 3 Summary of papers reviewed that take a spatial and/or quantitative approach that employ or best inform urban ES justice assessment methodologies. These 12 articles were deemed to have methods that can be used as, or to inform, tests of the justice of urban ES distribution  Authors Year Summary Analysis Method Results Justice assessment BenDor and Stewart 2011 Compared the location of wetland impact and mitigation sites with the socioeconomics of the populations surrounding those sites. Spatial overlay Populations near impact sites were generally more urban, whiter, and more educated, with higher incomes and lower poverty rates. Populations near mitigation sites were more often rural and people of colour, with higher poverty rates and lower incomes and/or wealth. Aquatic ecosystem losses and gains were therefore found to occur across a strong socio-economic gradient. Aquatic ecosystem losses and gains were found to occur across a strong socio-economic gradient. Displacement of aquatic ecosystems from urban areas, therefore, resulted in large numbers of urbanites losing aquatic ES and their counterparts gaining ES, while also shouldering the burden of concomitant disservices. Booth, Gaston and Armsworth 2010 Investigated the composition of user groups of recreation sites in England in terms of gender, age, socioeconomic level, and ethnicity. Visitor survey, geodemographic analysis Results showed significant bias among types of users, with fewer women, young people and ethnic minorities using recreation sites. Older, white males were more common site users. Justice was mostly assessed qualitatively, but also using statistical inference to compare characteristics of site users with local, regional and national populations to establish representativeness. Cohen et al. 2012 Compared the distribution of (semi) public greenspaces with socioeconomic data in Paris, France to assess the equity of greenspace distribution. Proximity/Overlay, Cluster analysis The distribution of greenspaces mostly benefit middle- and working-class – their neighbourhoods clustered at the edge of Paris, proximal to the majority of green space. Wealthier residents of Paris generally lived in denser areas of the city with low-medium levels of ES provision from semi-public green space, though they did have greater access to waterways. Justice assessed based on whether greenspace was distributed along a socio-economic gradient. A focus on semi-public green space was limiting because many wealthier people in Paris have access to private spaces within and out of the city. They also tend to live in Haussmannian districts that have high built density. Davis et al. 2012 Inquiry into the equitability of ecosystem services distribution in Chicago, IL, USA among socioeconomic groups. 4 variables used as ES proxies: proximity to Lake Michigan, proximity to open space, canopy cover and bird species abundance. Proximity analysis: distance to ES for each socio-economic group Mostly Hispanic census tracts had lower values for all 4 ES proxy variables (significant), while mostly-black census tracts had lower median values than the higher incomes tracts for all four variables (not-significant). Authors concluded that there is an unequal distribution of ES in Chicago, which they explicitly labelled an environmental justice issue. Noted that the equitability of ES distribution was not assessed. If disadvantaged social groups required more access to ES due to differential need, then even where ES distribution was exactly equal it could still be inequitable. Ibes 2015 Evaluated the equity of the distribution of park types in Phoenix, AZ, USA. Compared mean values of socio-economic variables among park cluster groups as well as the correlations of socio-economic variables with park types to establish significance. Cluster analysis (for park types), proximity analysis (socio-economic context) Large, more natural parks were associated with higher income areas, possibly because wealthier neighbourhoods were developed near them, or as a result of "white flight" from the city centre, or both. Lower access to these large parks could be an equity issue, but different ethno-cultural preferences mean it could not be. Ibes did not label the distribution of park types as just or unjust, noting equity analysis was not comprehensive. Study allows for a relative ranking of more or less environmentally just areas in Phoenix, however. Population density was highlighted as an important equity indicator – the higher the density, the more public outdoor spaces may be necessary. 26  Authors Year Summary Analysis Method Results Justice assessment Jenerette et al. 2011 Examined the spatio-temporal distribution of the cooling effect of vegetation in Phoenix, AZ, USA, and compare this with neighbourhood socioeconomic characteristics.   ‘Riskscapes’ of the distribution of 50°C surface temperature threshold.  Urban Heat Island (UHI) risk increases with minority status and decreasing wealth. Higher income areas of Phoenix had lower critical temperatures and more cooling vegetation. Cooling vegetation became more concentrated in wealthier areas over time.  Authors note that increasing vegetation to reduce UHI has consequences for equity. Increasing concentrations of cooling ES in wealthier areas through time, pointing to a growing disparity in wealth and equitable outcomes.  Kabisch and Haase 2014 Investigated the possible injustice of urban green space provision in Berlin, Germany. Per capita UGS is compared with population density, immigrant status and age. Two spatial scales of analysis: a city-wide scale, as well as a case study of the Tempelhof airport park.  Cluster analysis (for variance in UGS and socioeconomic variables). Gini index used to quantify evenness of distribution of UGS by district.  Three main clusters identified: (1) densely populated, low-vegetation cluster in the city centre with a high proportion of immigrants; (2) low density, high-vegetation cluster on the city periphery; (3) intermediate cluster. Gini results showed UGS distribution was highly unequal, especially for immigrants. Authors determined UGS distribution constituted a distributive injustice based on Gini results. The findings at Tempelhof were distinct from the pattern of distributive injustice found at the city scale, and may more closely relate to procedural justice concerns about effective consultation with nearby residents on appropriate park amenities. Lakes, Brückner, and Krämer 2014 Compare socio-economic inequality, ES and EDS in Berlin’s residential areas using an environmental justice index (EJI). Study is explicitly characterized as an environmental justice study assessing distributive justice Hotspot analysis of EJI.  Hotspots of inequality found to be in central Berlin (low greenness and high noise), with two clusters on city outskirts. 22 (5%) of 434 planning units had both low EI and SI, although 130 planning units had burdens attributed to one indicator or the other. 47 EJ ‘coldspots’ identified, mostly closer to the edge of the city, corresponding to areas of higher income and more abundant green space. EJI was not found to vary along a distinct gradient in Berlin, showing a more fragmented pattern. The authors hypothesize that wealthier inner-city areas that have been subject to recent gentrification may have contributed to this fragmented effect. Regardless, environmental quality was correlated with socio-economic status. McDonald, Forman and Kareiva 2010 Measured open space lost due to urban expansion for all US cities, and quantified the equality of per capita land consumption. Authors also evaluated the impact of zoning regulations on open space loss, as well as whether conservation funding tends to mitigate loss Landcover, zoning and demographic change analysis. Gini used to measure land consumption inequality. Open space loss was strongly correlated with population growth. Smaller land consumption per capita in larger cities and vice versa. Relatively few residents consumed most of the land at a low density. Cities that saw an increase in the proportion of high-density residential areas, and/or had reduced per capita land consumption saw an increase in the Gini coefficient, and inequality. The main equity concern of the paper was the measurement of land inequality using the Gini index. With relatively few people consuming both most of the land and driving urban expansion, results enable a justice assessment to be made. Authors, however, do not characterize their results as demonstrating injustice of land consumption/use. 27  Authors Year Summary Analysis Method Results Justice assessment Peng et al. 2015 Analyze the sustainability of Beijing, China, using an ecological footprint (EF) model. Distributive justice of resource use a primary concern addressed by the authors. Inter- and intragenerational equity also important temporal and spatial components of a full accounting of equity. Gini used to measure equity of natural capital utilization. Efficiency of natural capital use is measured using Human Development Index (HDI). Results showed obvious disparities in HDI and EF in Beijing, with intensity of natural capital use and HDI highest in the centre and decreasing towards peri-urban areas. The authors argue that when looking at Beijing as a whole that the EF is balanced. Curiously, inequality uncovered in the results are dismissed by aggregating the results to the city scale. Yao et al. 2014 Created an Effective Green Equivalent (EGE) metric to quantify the quality and accessibility of greenspace with two indicators: average EGE for all urban residents, and inequality coefficient of the EGE, measuring the equality of EGE distribution in an urban area. EGE applied to the case study city of Beijing, China. Distance to green space and NDVI used to assess green space quality. Inequality coefficient is based on the Gini coefficient EGE values per person are normally distributed, with an average value of 355.49ha per person, and an inequality coefficient of 0.24. 68.2% have access to moderate amount of greenspace. Authors argue that inequality coefficient values less than 0.4 are increasingly fair. It was noted that quality and accessibility affect the ES actually accessed by residents, with the EGE incorporating these elements to develop a more comprehensive picture of how green space is distributed in cities. Yin et al. 2007 Measured park accessibility using geospatial techniques in Qingdao, China. 1000m threshold between good and poorer access was used, calculated for each household in the study area. Percentages of very good (<500m), good (500-1000m), poor (1000-200m) and very poor (>2000m) access were then reported. Container and minimum distance methods were used to measure park accessibility. The vast majority of homes had very good access (64.04%) or good access (28.11%), with substantially fewer residents having Poor (7.69%) or very poor (0.16%) park access. The authors assert that their results show that there is a fairly equitable distribution of parks in Qingdao, with good access to green space at the household level. The authors note that accessibility studies had not been done for Chinese urban parks at the time of writing, nor indeed was it common practice to evaluate the accessibility of most other urban amenities, including schools and hospitals. This lack of accounting for distributional justice by Chinese planners, the authors argue, aggravates social inequities. 28  2.3.4 Qualitative Approaches and Conceptual Frameworks As shown in Figure 9, qualitative and hybrid articles didn’t necessarily discuss more justice issues than quantitative papers as encoded in the EJI. Instead, qualitative articles tended to engage in more focused discussions of fewer justice issues, sometimes in the style of argumentative essays. These articles tended to query the consequences of ways of valuing ES alongside the consequences of ignoring or embracing justice issues. Power dynamics as a result of current and historical processes were also a common theme.   Ernstson (2013b) provided one of the few theoretical frameworks published to date that specifically addresses environmental justice and ecosystem services. Fundamental is the idea that ES are not objective truths but rather are contested, socially produced artifacts. The concept of justice employed is made explicit, using a Rawlsian distributive justice definition. The framework is also rooted in ecological complexity theory and Actor Network Theory (ANT) (Latour 2005) which allows for the biophysical and social processes that underpin (often unequal) ES distribution patterns to be revealed.  Justice was also directly examined by Heynen (2003) who observed that urban ES distributions tend to be unequally distributed along socio-economic gradients. He also examined how scale affects production of injustice with regard to the provision of benefits from urban forests. The crux of Heynen’s argument is that what is just or not changes at different scales, therefore a multiscalar approach to justice assessment is required to assess justice trade-offs. Heynen also discusses how existing power structures tend to reinforce uneven urban environments. Those that have the most control tend also hold the lion’s share of capital, with the inverse also true. More generally, Heynen’s discussion of scale is an important addition to those concerned with temporal scales in the form of shifting baselines and historical injustices.   Power imbalances were the focus of other authors, too. Byrne and Wolch (2009) examined how ideology (class, race, nature) has shaped parks and park creation. They explored reasons for ethno-racial differences in park use, which is well-documented, as well as how power relations related to parks/nature spaces are investigated by geographers. The authors highlight many 29  justice and equality issues including how the environmental, cultural and political context of parks may result in non-use and avoidance by people of colour and other vulnerable groups. In general, gender, class and race combine to negatively impact access to urban ES. The authors also highlight areas where further research is needed, such as historical racism, gender, climate change and how they impact park use/access. The role that disadvantaged/people of colour play as agents in shaping new parks/access to nature also merits further inquiry.   Safransky (2014) seems to take up Byrne and Wolch’s call for a deeper historical perspective. Safransky investigated the role of US settler colonialism in the redevelopment of Detroit, MI, USA. They outline how, in the midst of a fiscal crisis, ‘green’ redevelopment was being touted as the way forward in Detroit by city managers. The underlying justice issue of green redevelopment, Safransky argues, is that green redevelopment leads to unequal costs and benefits as a result of gentrification and displacement. A key method by which land is opened for development is by categorizing it as empty. Treating land as empty can erase values held by those still attached to or using the land, such as community gardeners, or displace people themselves. Uncovering settler colonialism at work, Safransky argues, is an important part of decolonizing spaces, and of articulating alternative narratives of value.  Taking an approach much more rooted in ecology, and ecosystem services specifically, Unnikrishnan and Nagendra (2014) explored how restricting lake access in Bangalore, India affected ES access and provision, emphasizing cultural ES along with those more readily-quantified. Local authorities granted management and exclusion authority over some of Bangalore’s lakes to private companies based on the perception that nature was being overexploited with public access. By privatizing urban commons, Unnikrishnan and Nagendra argue that the actors and social networks accessing lake ES had been alienated. They also found that commoditization can lead to individual ES being valued over ES bundles and can also change the “cultural imaginary” of resources, both of which can negatively impact the provision of cultural ES. In the long run, the authors suggested that exclusion of those most dependent on lake ES, both for livelihoods and culture, might reduce social-ecological resiliency in Bangalore.  30  2.4 Discussion My results show a growing conversation around urban ES and ES justice. Far more ES and EDS are mentioned in articles than are investigated, yet investigations of ES/EDS are rising. At the same time, however, median EJI values were shown to have remained relatively static year-on-year. These results do seem to caution that the depth of ES justice inquiry is not expanding along with the depth of urban ES inquiry in general. The trend through time for EDS closely matches ES, with a peak in 2013 for ES absent due to the Wilkinson et al. paper’s lack of treatment of EDS.   Papers that deemed to include tests or measures of justice tended to find that inequalities in urban ES distribution or accessibility did exist, though results varied depending on the city and analysis method. Justice tests were sometimes not explicitly cast as such by their authors, and few papers decided whether or not an ES distribution was just or not. The strongest tests of justice tended to come from those papers that employed either cluster analysis or an inequality index (i.e. Gini) or both.   In general, papers with a qualitative focus argued for more nuanced justice assessments, noting that urban ES justice issues tend to arise from complex intersections of historical and current power imbalances, which themselves can be impacted by wealth, race, ethnicity, culture and ideology. Therefore, though no qualitative or “hybrid” paper posited conceptual tests of justice, their insights can be synthesized into meaningful qualitative criteria for justice assessments.   2.4.1 Few elements of justice are usually addressed in urban ES research At the same time, there was no significant difference the number of justice issues raised in qualitative articles as compared to quantitative or hybrid articles. Indeed, the majority (72%) of articles’ EJI scores were ≤ 0.4, indicating a fairly narrow range of justice issues considered by most articles. However, an EJI score does not necessarily indicate the depth of investigation of individual elements of justice. So though qualitative articles as a group did not tend to investigate 31  more elements of justice than quantitative papers, they did often explain particular elements in greater detail.  The two most common elements of justice tallied for all articles were also the most general – “Equity or Equality” and “Distributive Justice”. The first of these was tallied in the event of even a vague overture to equity or justice issues. The second, distributive justice, is arguably the most material of the justice issues in terms of consideration of the physical distribution of urban ES and their material benefits. It does not necessarily entail a conversation of power imbalances, governance, or historical inequalities. This finding lends further weight to the conclusion that the conversation of urban ES justice issues could be broader.  2.4.2 Developing world urban ES justice literature is needed in the 21st century The preponderance of developed world literature on urban ES and urban ES justice is of concern for a variety of reasons. Their results may not be transferable to the developing world cities where population growth is most expected to be concentrated this century (United Nations Department of Economic and Social Affairs Population Division 2014). Developed world literature might actually underestimate the extent to which ES are distributed inequitably, for instance, if their results were used to form global conclusions. The way that ES are valued may also be different – for developing world cities the benefits of street trees, for example, might not be seen to outweigh the costs of their maintenance (Cilliers et al. 2014). It should be noted, however, that the Western bias of the reviewed literature could in part be accounted for by the search being conducted in English.  2.4.3 Quantitative justice assessment methods: Greenspace as ES proxy is low-hanging fruit  Articles with more quantitative approaches generally had an emphasis on green space, mostly derived through geospatial data and methods. Using greenspace as a catch-all proxy for ES is not unproblematic (Andrew, Wulder, and Nelson 2014; Eigenbrod et al. 2010), however, it does have some obvious advantages: it is relatively simple to identify geospatially, it is spatially-32  explicit when mapped, and there is a substantial literature on the ES that greenspaces provide to people (Reviewed in Haaland and Konijnendijk van den Bosch 2015). Many articles used indices to measure the abundance of ES and EDS, as well as the equality of their distribution, most commonly using the Gini index. The spatially-explicit quality of greenspace allows it to be compared with socio-economic/demographic data from surveys and censuses, often in space and time, to establish whether ES and EDS vary along a gradient that can inform justice inquiries. These quantitative, spatially-explicit methods can better enable researchers and policy makers to compare the distribution of urban ES within cities and among cities for different socio-demographic groups. This is information that is of key importance for informed decision making that impacts urban environmental justice.   2.4.4 A message from geographers: The value of urban ES are contested, and can be used as a rationale for gentrification and dispossession  Perhaps as important for urban ES management that incorporates environmental justice considerations is the qualitative and conceptual work that was included in this review. Critical geographers, especially, have added much to an understanding of how ES are embedded in existing power structures and historical narratives, and how ES are not neutral artifacts in the political economy of a given social ecological system.   In one of the most synthetic papers in the review, Ernstson (Ernstson 2013b) argued that ES are socio-political constructs, meaning that ES are not objective ecological realities. Based on this literature review and other sources (for example, the excellent urban ecosystem services reviews of Hubacek and Kronenburg (2013), Luederitz et al. (2015) and Suich, Howe and Mace (2015)), it is a seminal work on the relationship between justice and ecosystem services. As Ernstson says, ES don’t exist “out there,” as an inherent part of ecosystem functioning, they require human interpretation to realize as benefits. If this is true, then it is also true that mechanistic assessments of trade-offs among ecosystem services must be approached with caution. The ES that are most salient for ascribing value to landscapes, and for informing management decisions might be in the eye of the beholder. Furthermore, because of power imbalances the types of ES consistently valued by one group may be consistently undervalued by those of another. Gaining a 33  full understanding of how ES vary among urban socio-demographic groups, therefore, is not merely a process of mapping green space, but also understanding how the groups in question ascribe value to the ES supply sources around them.  More generally, critical geographers are worried about the use of ES due to their complicity in processes of neoliberalization, the commodification of nature, as well as the consequences of that commodification resulting from the inherent injustices of markets (e.g. elite capture, unequal distribution of costs and benefits, externalities, etc.). Unnikrishnan and Nagendra (2014) showed that privatization of ES-provisioning areas can lead to fewer ES being provided, with cultural ES provision particularly curtailed. Indeed, multiple authors noted how the valuing of ES in urban areas can be used as a rationale for gentrification and dispossession.   Historical injustices, too, can drive current patterns and policy surrounding the distribution of ES. For example, Safransky’s (2014) examination of colonial methods of land dispossession and the use of revisionist history in Detroit, MI showed how historical injustices can be perpetuated by unjust methods, themselves a product of a history of injustice. Also important to understand is that not all green space is created equal from the perspective of different ethnocultural groups. As Byrne and Wolch show, different ethno-cultural groups use green space differently, with important implications for procedural and interactional justice.   2.4.5 Urban green spaces provide ES that are often overlooked by ecologists As spaces for community organizing and place making, green spaces provide ES that are often overlooked by ecologists, though perhaps less often by landscape and urban planners. Though ES flowing from green spaces can be a rationale for dispossession, green spaces can also be sites of community organization – where people come together to construct their own narratives of place, and their own narratives of the value of nature, perhaps in opposition to narratives imposed on them from outside. Though a thorough investigation of place-making, and of the relationship between space and place is beyond the scope of this review (See Harvey 1994; and Massey 1991), it figures prominently in Heynen and Safransky’s papers, and is also touched on in other papers that can be found Appendix B: see, for example, Anguelovski and Martínez 34  Alier’s ‘Environmentalism of the Poor’(2014); Bendt, Barthel and Colding’s analysis of environmental learning in community gardens in Berlin (2013); Colding et al.’s look at urban green commons in Sweden, Germany and South Africa (2013); as well as Shackleton et al.’s look at perceived ES of residents in two South African towns (2015).  2.4.6 Study Limitations Sources of error for this review include the fact that there was only one reviewer, whose subjectivity in choosing and interpreting articles could not be mitigated by others. Many other articles, or sections of articles could have been relevant as well, with other search terms or methods likely yielding different results. The Luederitz et al. urban ecosystem services review is a good example of this – it was not included in the Web of Knowledge search despite being an important summary of urban ES in 2015 that included social and justice-concerned papers itself.  The justice index employed here is relatively coarse, with categories chosen subjectively. It focuses on mentions of the elements of justice without quantifying the depth of their investigation by using numerical encoding rather than binary variables. Each element of justice was treated equally, as well, while it could be argued that justice tests or assessments should deserve more weight in the determination of the index.  A peak in the ES mentioned in 2013 is in part a function of the Wilkinson et al. (2013) paper being published in 2013. Despite some bias inherent in tallying the ES in the Wilkinson et al. paper using a modified version of their own ES scheme, that paper still provided the most comprehensive list of specific ES in the literature review.    Another issue is the necessarily interdisciplinary nature of urban environmental justice and urban ES literature, spanning ecological, urban planning, critical geography and other disciplines. ES provide a common language for articulating natural values to an extent, but there are still concepts and jargon that have yet to cross disciplinary boundaries. Ernstson (2013b) likely provides the most synthetic approach, bridging ecology and social science by integrating social-ecological systems, ecosystem services, social construction of value and Actor Network Theory 35  all in one framework. It was relatively rare for the more ecology focused papers to, for example, incorporate historical injustice and contemporary colonialism as employed by Safransky (2014), or for papers that had a strong qualitative treatment of justice issues to incorporate quantitative justice assessment metrics such as the environmental justice index of Lakes, Brückner, & Krämer (2014). Individual journal articles cannot do it all, however, and papers like that of Lakes, Brückner, & Krämer (2014) or Cohen et al. (2012) show that it is possible to have a nuanced disucssion of justice while also conducting quantitative analyses that provide a strong empirical basis for further inquiry.   2.5 Conclusion For those advocating the use of ecosystems services in urban areas it is critical that they understand that ES are rarely distributed evenly in cities. This unequal distribution has important consequences. With poorer people the world over more reliant on ES and more vulnerable to EDS, equity of access to ES should be a chief concern of those that are seeking to create social-ecological systems that are both resilient and equitable. Another important lesson is that ES are socio-political constructs, valued differently by different people, and created through the use of competing narratives and (often scientific) artifacts (Ernstson in 2013). As Ernstson says, ES don’t exist “out there,” as an inherent part of ecosystem functioning, they require human interpretation to realize as benefits. If this is true, then it is also true that mechanistic assessments of trade-offs among ecosystem services must be approached with caution. The ES that are most salient for ascribing value to landscapes, and for informing management decisions, might be in the eye of the beholder. Furthermore, because of power imbalances, and the host of justice issues described above, the types of ES consistently valued by one group may be consistently undervalued by those of another. Gaining a full understanding of how ES vary among urban socio-demographic groups, therefore, is not merely a process of mapping green space, but also understanding how the groups in question ascribe value to the ES supply sources around them.    36  Chapter 3: High-resolution urban landcover mapping using RapidEye and LiDAR  3.1 Introduction To date, landcover mapping for metropolitan regions has generally fallen into two broad types from the point of spatial resolution: a very-high spatial resolution approach and a moderate spatial resolution approach. The first utilizes very-high spatial resolution  (< 2m) imagery, including aerial photography and, increasingly, hyperspectral and LiDAR-derived 3D structure information (Chance et al. 2016; Y. Chen et al. 2009; Z. Chen and Gao 2014; Hermosilla et al. 2011; Kim and Kim 2013; J. P. M. O’Neil-Dunne et al. 2013; Plowright, Coops, and Aven 2015; Sasaki et al. 2011). The very-high spatial resolution approach can lead to high accuracies at fine spatial scales, however, it can also be expensive, difficult to scale to large areas and can require considerable time and computing power for data processing.  The very-high spatial resolution approach is also increasingly characterized by data fusion, particularly of spectral, thematic and 3D structure data (Rashed and Jurgens 2010, chap. 11). LiDAR, especially, is regularly employed in urban settings alongside spectral data to improve classification accuracies (Hartfield, Landau, and Leeuwen 2011; Tompalski and Wężyk 2012; Zhan, Molenaar, and Tempfli 2002; Zhou et al. 2009). A key benefit of LiDAR is the capacity to reliably obtain high-precision, three-dimensional measurements of buildings and trees over broad spatial scales which, as a result, has attracted significant interest among urban and natural resource managers (Hudak, Evans, and Smith 2009). LiDAR pulses can also penetrate vegetation, which allows analysts to derive useful information about the vertical structure of tree canopies (Coops, Duffe, and Koot 2010; Plowright, Coops, and Aven 2015).  The moderate spatial resolution approach utilizes imagery with pixel sizes from 30m - 500m such as Landsat Enhanced Thematic Mapper Plus (ETM+) or the Moderate Resolution Imaging Spectroradiometer (MODIS) (Jing et al. 2015; Lu, Coops, and Hermosilla 2016, 2017; Nichol and Lee 2005; Shao and Liu 2014). Moderate spatial resolution imagery is often inexpensive to obtain, and has been used for time-series analysis and global comparisons. However, moderate-37  spatial-resolution approaches necessarily lack the fine spatial detail that is a key determinant of urban classification accuracy (Momeni, Aplin, and Boyd 2016) and, in urban settings, must overcome the challenge of mixed pixels. Mixed pixels are not well suited to traditional hard classification methods, which classify land cover on the basis of spectral reflectance properties while assuming homogeneity at the pixel scale within any given class land cover class, and so techniques such as spectral mixture analysis must be employed (Lu, Coops, and Hermosilla 2017; Small 2003; Welch 1982).  Neither moderate nor very-high resolution as defined here, RapidEye is a satellite constellation that makes use of five identical sensors to provide 5 x 5 m spatial resolution imagery with a return time of 5.5 days at nadir (and daily if considering off-nadir imagery). The RapidEye sensors consist of five bands situated in the visible and NIR portions of the electromagnetic spectrum with a range of 440-850 nm. RapidEye also employs a Red Edge band at 690-730nm that has been shown to improve the spectral separability of landcover classes, especially those that are vegetated (BlackBridge 2014).  In this chapter RapidEye imagery is integrated with LiDAR data in a Geographic Object-Based Image Analysis (GEOBIA) workflow (Figure 12). GEOBIA has been shown to be more accurate than pixel-based approaches for urban landcover classification (Y. Chen et al. 2009; Jebur et al. 2013; Momeni, Aplin, and Boyd 2016; Myint et al. 2011; Reviewed in J. O’Neil-Dunne, MacFaden, and Royar 2014), in part because it can accommodate dissimilar data types as well as textural, shape and context metrics (Thomas Blaschke et al. 2014). The goal was to create a high-spatial-resolution landcover product, with relevant classes and accuracies for landcover management, which could precisely resolve both buildings and tree canopies for a large metropolitan area.   38  3.2 Methods 3.2.1 Study Location Located in southwestern British Columbia, Canada, Metro Vancouver is a partnership of 21 municipalities, one Electoral Area and one Treaty First Nation that collaboratively plans for and delivers regional-scale services (Metro Vancouver 2016a). The study area includes Metro  Vancouver and a neighbouring municipality, Abbotsford, with a 5km buffer (Figure 10). Key geographical features of Metro Vancouver are the Coast Mountains, which make up the northern, more remote parts of the study area, and the Fraser River, which runs through a broad valley in the eastern sections of the study area before terminating in a delta at the Pacific Ocean in the West. Landcover and landuse types vary widely across the study area and range from densely built-up areas, to extensive areas of lower-density single-family homes, agricultural lands, as well as large undeveloped tracts of forests, and fresh water.   Much of the region was once, or still is, temperate rain forest dominated by western hemlock (Tsuga heterophylla (Raf.) Sarg.), western red cedar (Thuja plicata Donn ex D.Don), and Douglas fir (Pseudotsuga menziesii (Mirb.) Franco). At higher elevations, the growing season is considerably shorter, with forests dominated by mountain hemlock (Tsuga mertensiana (Bong.) Carr.) and yellow cedar (Cupressus nootkatensis D.Don), before giving way to alpine meadows and bare rock, snow and ice.   3.2.2 Data 2014 RapidEye images were delivered radiometrically corrected and georeferenced (Level 3A product specification, RapidEye, 2015). RapidEye is a constellation of 5 multispectral satellites delivering 16-bit multispectral imagery at a 6.5m ground sampling distance, and a 5m orthorectified pixel (RapidEye 2015). RapidEye uses a 5-band sensor with Blue, Green, Red, Red Edge and NIR bands (bands 1-5, respectively). Cross calibration studies have shown individual RapidEye sensors to be interchangeable, with recorded radiometric differences within 2% amongst sensors (Thiele, Anderson, and Brunn 2014). RapidEye’s spectral band configuration is ideal for vegetation monitoring (Bindel et al. 2011), while the temporal coverage 39  may be useful for short term change, as well as longer term land cover transitions (Metternicht, Hurni, and Gogu 2005).   Figure 10 Study area boundary encompassing Metro Vancouver, Abbotsford and a 5km buffer. The total study area is 5715.97 km2 and is composed of diverse urban, peri-urban and undeveloped areas. Topographic map from ArcMap 10.x (Sources: Esri, HERE, DeLorme, Intermap, increment P Corp., GEBCO, USGS, FAO, NPS, NRCAN, GeoBase, IGN, Kadaster NL, Ordnance Survey, Esri Japan, METI, Esri China (Hong Kong), swisstopo, MapmyIndia, © OpenStreetMap contributors, and the GIS User Community).  Three images were acquired from August 2nd, 24th and 26th to cover the central, east and west portions of the study area, respectively (Table 4, Figure 11). The August 2nd image covers the majority of the study area and was used preferentially over the other images, especially where scattered clouds existed in the August 24th image. An exception was made for a small eastern section of the August 2nd image where haze from wildfire smoke was present. Here, the August 24th image was used.   40  Table 4 Data used throughout the GEOBIA workflow. Three RapidEye images provided multispectral information at a 5m pixel resolution. Air photos with a 20cm pixel covering a northwestern portion of the study area were used for ground truth point classification. LiDAR data came from multiple sensors and years with a considerable range in point density. Vector features, provided by the metropolitan regional planning authority, were used to establish study area boundaries, to identify port areas in the GEOBIA ruleset and as masks during manual reclassification.  Imagery LiDAR Vector Features  RapidEye (NIR-RE-RGB, August 2014) Air photos (RGB, Summer 2014)  Multiple Sensors (2008-2015, 3.2-36.5 pts/m²)  Land Use (2011) Land Cover (2010) Municipal Boundaries Project Boundary Waterbody Transport Line   Figure 11 Coral polygons totalling 1584.15 km2 showing LiDAR coverage of the study area, corresponding to the majority of the built-up areas in the region (61% of incorporated municipalities). The hashed area shows air photo coverage used for ground truth assessment in conjunction with Google Earth imagery used for the rest of the study area. Background images are, from left to right, the August 26th, August 2nd, and August 24th, 2014 RapidEye images used for the classification (RGB = 532). 41   LiDAR data was collected over multiple years for individual municipalities or watersheds and varied with respect to sensor, date (2008-2015) and point density (3.2-36.5 pts/m²). This semi-contiguous patchwork of data covers about 158 415 ha or 61% of the built-up, developed and/or municipally incorporated sections of the study area (see Figure 11).  3.2.2.1 LiDAR processing All LiDAR was processed into normalized digital surface model (nDSM), standard deviation of height (zDev), and nDSM slope rasters. Showing the relative height of features above the ground, nDSMs are derived by subtracting a digital elevation/terrain model (DEM/DTM) from a digital surface model (DSM) (Haala and Brenner 1999; Hermosilla et al. 2014). They have been used widely in urban landcover classification studies to identify above-ground features such as buildings and trees (Blaschke, 2010). Helpful for separating above-ground features, zDev layers can be separated into rough-textured features often corresponding to vegetation, and smooth-textured features corresponding to buildings (MacFaden et al. 2012). Slope values are helpful for delineating features that begin or end at relatively steep height boundaries such as the tops of many buildings, roads, and tree crowns (Trimble eCognition 2010).   For nDSM creation, I followed the methodology of Tompalski and Wężyk (2012). The nDSM and nDSM slope layers were created with a 1m spatial resolution, and the zDev layers were created with a 2m spatial resolution to yield more informative grid metrics. Point cloud processing for nDSM and zDev rasters was done using lastools (Isenburg 2014, 2017) and FUSION (McGaughey 2015), while the slope raster was produced as part of the GEOBIA rule set using individual nDSM rasters (see below). For all LiDAR data powerlines were not removed. For pre-classified LIDAR data, all classes except water and noise were included in nDSM and zDev creation.   42  3.2.2.2 Class hierarchy development A classification hierarchy was informed by a literature review, existing Metro Vancouver landcover data from 2010, and a consideration of the classification methodology. The hierarchy follows many of the criteria of Anderson et al. (1976): i.e. it is applicable over extensive areas, repeatable, comparable through time, and follows a hierarchical structure from the general to the specific. Most of the final classes used, however, still correspond to level 1 of Anderson’s class hierarchy, with some exceptions (see Table 5). The HERCULES model of urban landscape structure proposed by Cadenasso et al.  (2007), which breaks urban landscapes into three main elements (buildings, surfaces and vegetation) and their sub-features, was also influential. In discussion with Metro Vancouver, 12 final classes were decided upon for the classification hierarchy (Table 5).  3.2.2.3 Spatial resolution and Minimum Mapping Unit (MMU) considerations LiDAR-derived rasters provided vertical structure information at a 1m spatial resolution allowing very fine spatial detail of building outlines as well as tree canopy shape and position. Conversely, the 5m spatial resolution of RapidEye has a poorer spatial resolution but allows objects to be spectrally distinguished in 5 spectral bands. As a result, a map spatial resolution of 2 m was confirmed as optimum – a hybrid of the fine spatial resolution of the LiDAR and the spectral detail of the RapidEye imagery.   Table 5 Final classification classes and classification criteria. Level 1 Level 2 Level 3  Criteria Built-up Buildings --  Houses, garages, warehouses, towers, industrial structures, etc.   Paved --  Asphalt and concrete surfaces including sidewalks, parking lots, alleys and highways.  Other Built --  Not concrete/asphalt built surfaces or building roofs. Sports surfaces (artificial turf and running tacks), transit or rail areas, other impervious surfaces, etc. 43  Level 1 Level 2 Level 3  Criteria Bare Barren --  Beaches, alpine rock, shoreline rock, etc. Lack of vegetation. Likely not soil (colour/context suggests no organic matter and/or imperviousness). Also quarries, gravel pits, gravel roads.  Soil --  Agricultural soils (light or dark), cleared/open areas where darker colours indicate organic matter present (as compared to, e.g. sand) Vegetation Tree canopy Coniferous  Predominantly coniferous (>75%)   Broadleaf  Predominantly Broadleaf (>75%)   Shrub  Woody, leafy, and generally rough-textured vegetation shorter than trees (approx. <3-4m), taller than grass  Grass-herb --  Most crops, golf course greens, city park grass, lawns, etc. Alpine meadows, near-shore grass areas, fine-textured bog/wetland areas.  Non-photosynthetic vegetation --  Dead grass, cutblock slash, log booms  Water -- --  Lakes, rivers, inlets, irrigation channels, retention ponds, pools, etc. Shadow -- --  Dark pixels with very low reflectance values. Image features not easily visible. RapidEye images used for shadow locations.  3.2.2.4 Training Data Creation Training points were randomly generated, with the majority stratified using an existing 2010 landcover layer. Stratification was used to obtain a representative sample of the rarer land cover classes, with the number of points generated within each stratum (a 2010 landcover class layer) based on the proportion of that stratum’s area of the 2010 landcover class layer. Approximately 200 additional points were added non-randomly to improve classifier training for a number of classes, specifically Shadow, Other Built (especially artificial sports turf and running tracks), Soil, and Non-Photosynthetic Vegetation.   44  A total of 2993 training points were established and buffered at 5m, corresponding to the pixel size of the RapidEye imagery. For each point, primary and secondary class assignments were attributed at each level of the class hierarchy of Table 5. This was done by interpreting underlying image features for the 5m buffer around each point. If a buffer straddled two land cover types, a primary and secondary class could be assigned. So, for each ground truth point, up to 6 class assignments could be made considering every level of the classification hierarchy (i.e. level 1, first and second choice; level 2, first and second choice; level 3, first and second choice).  Training points were mostly those that did not have multiple target class assignments. An exception was for those points whose secondary classes were either Shadow or Other Built; these points’ secondary class assignments were often promoted to primary assignments to increase the number of points of those classes. Cases where primary and secondary class assignments were Grass-Herb and Non-Photosynthetic Vegetation were also included, as these classes were often interchangeable.   Points were classified using manual interpretation of Google Earth Imagery (Google Earth Pro v. 7.1.5.1557) – which consisted principally of World View and air photo data with pixel sizes less than 50cm – or aerial photography provided by Metro Vancouver. In Google Earth, contiguous, very-high-resolution imagery was available for the majority of the developed part of the study area from July 14th and August 2nd, 2014, and this imagery was used wherever possible. Seasonal imagery was used to differentiate broadleaved from coniferous vegetation. Where the landcover underlying point buffers changed between dates, the most common landcover type was chosen, followed by imagery closest in time to August 2nd, 2014.   45  3.2.3 Classification Workflow  Figure 12 Classification workflow showing input data, object segmentation, and classification stream. LiDAR data was processed into rasters and used for object segmentation. LIDAR and RapidEye data were combined to delineate and classify building and tree objects, and to delineate unclassified ground-level objects. After the first segmentation of the LiDAR-derived rasters into initial objects, objects were classified into temporary classes. Objects with high mean nDSM slope and zDev values were identified as tree objects. Objects with lower slope and zDev values, but nDSM values > 2m were classified into a temporary class that was a precursor to building objects. From this initial separation, objects were reclassified into other temporary object classes, merged together, and re-segmented where appropriate, with the goal of exporting semantically meaningful objects accurately classified into Buildings and Trees, or left unclassified. Unclassified objects were re-segmented using RapidEye pixel values before export. Trees and unclassified objects were further classified using Random Forests in R. The draft classification underwent some manual reclassification using vector masks before accuracy assessment was conducted using the caret package in R.   3.2.3.1 GEOBIA expert ruleset creation and classification A standard GEOBIA workflow was followed beginning with segmentation, followed by feature extraction and selection, and then classification (Hermosilla et al. 2011; Schöpfer, Lang, and Strobl 2010; Trimble 2012b) (Figure 12). Initial segmentation was conducted on the nDSM, smoothed nDSM, nDSM slope and zDev layers using a multiresolution (MR) segmentation algorithm (Baatz and Schäpe 1999; Trimble 2012a), which delineates pixels into homogenous objects using multiple input data layers, with the aim of segmenting buildings, tree canopy and 46  other, mostly ground-level objects. Relatively small objects were created using a scale parameter of 15, a shape parameter of 0.2, indicating that pixel values have a strong influence over object shape, and a compactness parameter of 0.8 to prevent overly complicated objects. Image layer weights of 2, 1, 1 and 3 were given to the nDSM slope, nDSM, smoothed nDSM and zDev layers, respectively. LiDAR-derived layers were solely used for initial segmentation because they represented high-resolution data: the pixels of the LiDAR rasters were smaller than features of interest (buildings and trees) (Figure 13).  Figure 13 Side-by-side comparison of classified building and tree objects in a single-family dwelling area of Burnaby, a municipality in Metro Vancouver. The top image is a 1m nDSM, the bottom image is 5m RapidEye imagery (RGB = 532). Individual trees and buildings are easily identifiable in the nDSM, while in the RapidEye image there are many mixed pixels obscuring objects of interest, though the general pattern of built-up and vegetated land cover is apparent.    A total of 23 pixel-based, geometric and contextual features from the RapidEye and LiDAR data were used to develop the rule set (Table 6). Pixel-based features included average the Soil 47  Adjusted Vegetation Index (SAVI), and, in the case of LIDAR, average values for all three rasters, along with measures of variation were included. Geometric features were used to classify objects based on their shape, while contextual features were useful for classifying objects based on neighbouring object classifications. Using these features, a classification ruleset was constructed to mimic as closely as possible decision trees employing on simple thresholds.   After ruleset classification, ground-level objects were re-segmented using the MR algorithm and RapidEye data. Building, Tree Canopy, and unclassified ground-level objects were then exported with a number of object features that varied by class (see Appendix C  for full list of exported features). Tree Canopy and unclassified objects were further classified using Random Forests (RF) in R (Breiman and Cutler 2015; R Development Core Team 2011; RStudio Team 2016).   Table 6 Object Features used to create thresholds for the assign class algorithm in the GEOBIA rule set. For each assign class algorithm, up to two threshold conditions can be applied to any number of existing classes to assign a new classification. Thresholds are taken from object features, which are pixel, geometric and context-based metrics describing objects. Further features descriptions can be found in the eCognition Reference Book (Trimble 2012a), and on the eCognition Community Wiki (eCognition Community Members 2009).  Object Feature Formula/Description Pixel-based   SAVI ((1+0.5))*(([Mean NIR]-[Mean Red])/([Mean NIR]+[Mean Red]+0.5)) Max height - Min height [Max. pixel value nDSM]-[Min. pixel value nDSM] Coef_Var_nDSM [Standard deviation nDSM]/[Mean nDSM] Mean_nDSM/SD_nDSM [Mean nDSM]/[Standard deviation nDSM] Mean nDSM Slope  Mean nDSM  Mean zDev  Max. pixel value  Min. pixel value  Standard deviation zDev    48  Object Feature Formula/Description Geometric   Area  Border Index  Compactness  Elliptic Fit  Length: Width  Rectangular Fit  Roundness  Shape Index  Width (only main line)  Width (only main line)  Contextual   Relative Border Relative border of objects of one class to objects of another class Distance to class In 1m pixels Land Use 2011 = Port Metro Vancouver Used to identify/reclassify objects in Port Areas.  3.2.3.2 Random forests classification Tree canopy and unclassified objects were further classified whereas building objects remained as a single class. Each RF was grown to 1000 trees, with an mtry parameter (number of predictors per node) set to the square root of the total number of predictor variables. Any tree or building training points were, except for the shrub class, not used for ground-level objects on the assumption that the building and tree classes were already captured through GEOBIA ruleset classification. After classification, vector objects were converted to rasters using the Polygon to Raster tool in ArcMap at a 2m spatial resolution ((ESRI (Environmental Systems Research Institute 2014)). A 3x3 majority vote smoothing filter was then applied to each raster to fill small data holes and smooth object boundaries.  49  3.2.4 Accuracy Assessment 3.2.4.1 Validation Data Validation points were generated randomly and stratified by the final classified raster. An average of 64 validation points per class (min = 20, max = 88), and a total of 768 validation points were used for accuracy assessment. For validation, only locations which had a single class designation and those which were well located on both the LiDAR coverage and the RapidEye were used. Classification of validation points followed a similar methodology to that of the training points detailed above.  Additional validation points were added to capture rarer but high-profile classes, such as artificial sports turf (classified as Other Built). Validation points also underwent correction after their initial classification by a photo-interpreter. This was done to (1) correct photo-interpretation errors; (2) remove points whose buffers straddled multiple landcovers (and did not have multiple class assignments); (3) account for spatial shifts in GE imagery and the classified raster; (4) choose between multiple potentially accurate class assignments.   3.2.4.2 Accuracy Assessment Accuracy assessment was conducted using the caret package in R (Kuhn 2017). A confusion matrix was produced using the validation points (5m buffers) described in section 3.2.2.4. A predicted class was assigned to each 5m buffer by choosing the largest proportion of the predicted class in each buffer. The output of the accuracy assessment was a confusion matrix, overall, producer’s and user’s accuracies, as well as Cohen’s kappa coefficient.  3.3 Results 3.3.1 LiDAR-RapidEye fusion in a GEOBIA workflow leads to high quantitative accuracies  Quantitative accuracy overall was high, with an overall accuracy of 88% and a kappa of 0.87 (Table 7). Buildings, the only class solely classified using the object ruleset, had user’s and 50  producer’s accuracies of class at 95% for both (Table 7). Tree Canopy – Coniferous, Broadleaf and Shrub – also had high individual class accuracies, ranging from 83-94%. Grass-Herb had a higher producer’s than user’s accuracy, in contrast to Non-Photosynthetic Vegetation which had the opposite pattern. This was due both to Grass-Herb’s confusion with Shrub for predicted classes and Non-Photosynthetic Vegetation’s confusion with Barren and Soil compared to the reference data. Non-vegetated ground-level classes (i.e. Paved, Soil, Barren, etc.) did see some confusion, however, despite these issues, individual class accuracies were still above 80% with the exception of the user’s Accuracy of Soil.   Table 7 Confusion matrix showing user's and producer's Accuracies for each predicted class. Overall accuracy is 88% with a kappa of 0.87 and upper and lower 95% confidence intervals of 89.9% and 85.0%, respectively.  Reference   Prediction Barren Buildings Conifer-ous Broad-leaf Grass-Herb Non-Photo Veg Other Built Paved Shadow Shrub Soil Water User's Accuracy Barren 48 0 0 0 0 4 0 5 0 1 1 0 0.81 Buildings 1 70 1 0 0 0 0 1 0 0 1 0 0.95 Coniferous 1 0 49 3 0 0 0 0 1 0 0 0 0.91 Broadleaf 1 0 5 78 1 0 0 2 0 0 0 1 0.89 Grass-herb 0 0 0 2 69 0 0 1 0 9 2 0 0.83 Non-photo Veg 0 0 0 0 0 55 0 0 0 0 0 0 1.00 Other Built 0 2 0 0 0 0 60 3 0 0 0 1 0.91 Paved 5 2 0 0 0 1 2 66 0 0 1 0 0.86 Shadow 0 0 0 1 0 0 0 1 16 2 0 0 0.80 Shrub 0 0 0 3 1 0 0 0 0 58 0 0 0.94 Soil 1 0 0 0 1 4 12 2 0 0 70 1 0.77 Water 0 0 0 0 0 0 0 0 3 0 0 36 0.92 Producer's Accuracy 0.84 0.95 0.89 0.90 0.96 0.86 0.81 0.81 0.80 0.83 0.93 0.92   RapidEye’s Red Edge band was important for class separation in each of the RF classifications of ground-level and tree canopy objects (Table 8), however, NIR band variables had a greater 51  importance for each object type. LiDAR-derived variables were also among the top 10 most important classification variables for both, ground-level and tree canopy objects, but spectral variables outnumbered LiDAR-derived variables for each object type.  Geometric and contextual variables were much less important than spectral or LiDAR-derived variables for RF, though were commonly combined with spectral thresholds for the GEOBIA ruleset (Table 6).  Table 8 Top 20 RF classification variables ranked by importance as measured by the Mean Decrease in the Gini coefficient for ground-level and tree-canopy objects. Spectral variables are the most numerous, and generally the most important, though mean nDSM values were the second most important variable for tree-canopy object classification. Individual band values and derived variables such as vegetation indices or band ratios were important, as was object brightness. Few contextual or geometric variables were important for object classification, with the exception of the Relative Border to Unclassified Objects (RelBord unclass) variable ranked 18th for ground-level objects. Other abbreviations are as follows: sd = standard deviation; CoefVar = coefficient of variation; Vis = visible bands (i.e. RGB); RE = Red Edge band; NDVI = Normalized Difference Vegetation Index; NDRE = Normalized Difference Red Edge; SAVI = Soil-Adjusted Vegetation Index; Ht = Height.  Ground-level objects  Tree-canopy objects  RF Variable Mean Decrease Gini Rank RF Variable Mean Decrease Gini Rank Mean NIR 132.91588 1 Image Brightness 48.588335 1 NDVI 126.01202 2 Mean nDSM 46.657251 2 NDVIRE 122.67819 3 Brightness Vis 46.431732 3 SAVI 112.31402 4 Mean NIR 40.234183 4 NIR:RE 94.32547 5 Mean Red 37.607138 5 NDRE 88.4334 6 CoefVar nDSM 32.849803 6 sd zDev 86.79146 7 Mean Green 32.19683 7 Mean Blue 85.78478 8 Mean Blue 29.184645 8 Brightness Vis 71.25298 9 NIR:RE 26.626301 9 sd slope 65.76563 10 NDRE 22.999547 10 Mean Green 65.40238 11 sd NIR 21.110174 11 Mean RE 64.68071 12 NDVI 20.658662 12 Mean Red 64.21073 13 Mean Slope 19.617318 13 sd Blue 63.92032 14 SAVI 19.192081 14 Mean zDev 61.43657 15 sd nDSM 18.415951 15 sd Green 61.32469 16 NDVIRE 18.379996 16 Image Brightness 60.92248 17 sd Red 18.284298 17 RelBord unclass 57.68813 18 Mean RE 17.958517 18 sd NIR 54.45 19 MaxHt - MinHt 17.91343 19 MaxHt - MinHt 53.07446 20 Mean zDev 17.724243 20 52   3.3.2 Vegetation dominates landcover but tends to decrease from east to west The 12-class classification at 2m spatial resolution is shown in Figure 14. The same rule set effectively segmented and classified building and tree canopy objects throughout the study area. Object mean and standard deviation were commonly used, as were their relative border to another class. Key customized features were the Coefficient of Variation of the nDSM (and its inverse), and the difference between maximum and minimum nDSM values. The classification covers the majority of the built-up sections of the study area, but also significant portions of the agricultural land. At coarser scales, the classification accurately captures the patterns of built-up and vegetated landcovers present on the landscape. At finer scales, the map achieved the goal of precisely resolving buildings and tree canopy at a high spatial resolution (Figure 15). Despite a number of dense urban areas, for example Vancouver’s downtown peninsula, built-up land covers only accounted for 22% of the study area (Table 9). Vegetated landcovers, by contrast, made up almost three quarters of the study area at 72.5% of the total. Tree canopy classes accounted for 40% of the total, while grass-herb made up 33.4%. Barren and soil, though often visually striking, only made up a combined 3.8% of the study area, with the Shadow, Water, and Non-Photosynthetic Vegetation classes constituting the balance.  53    Figure 14 Final landcover classification for LiDAR-coverage areas. Patterns of built-up, agricultural and forested landcovers are readily discernible. Landcover is overlaid on two RapidEye images used in the study (from left to right August 2nd and August 24th, 2014). 54   Table 9 Area of landcover per class, and its percentage of total area (158 415 ha). Vegetation covers the majority of the study area at 72.5%. One third of the study area is Grass-herb, mostly in the form of agricultural fields. Another third is made up by Tree Canopy classes, approaching 40% including Shrub. Built-up landcovers total 22% of the area, dominated by paved surfaces. Soil and Barren are similar in size, making up less than 4% of the total. With few large water features included in the study region, Water makes up only 0.6% of the total landcover by area.  Class Area (ha) Percent Area Grass-Herb 52863 33.4 Broadleaf 37358 23.6 Coniferous 20652 13.0 Paved 19837 12.5 Buildings 14850 9.4 Shrub 3407 2.2 Barren 3070 1.9 Soil 2934 1.9 Shadow 1794 1.1 Water 1008 0.6 Non-Photosynthetic Vegetation 534 0.3 Other Built 107 0.1  The patterns seen for the entire classification change from a municipal perspective (Figure 16). The more densely populated municipalities in the west such as Vancouver, North Vancouver (City), and Burnaby have higher percentages of built-up classes (Buildings, Paved and Other Built) than the other municipalities, with Vancouver and North Vancouver reaching 58% and 59%, respectively. In contrast, the University of British Columbia at the western tip of Vancouver has about 26% Built-up landcover and 52% coverage by Coniferous and Broadleaf classes. Municipalities in the southern part of the study area become increasingly dominated by the Grass-herb class the further to the east they are situated. Surrey has almost equal parts Built-up and Grass-herb landcovers at close to 31% for each class, respectively. For Langley (City and District) Grass-herb’s share of landcover grows to 43%, with Built-up classes covering only 15% of the area, while for Abbotsford, the furthest east municipality, those values grow further apart at 50% and 13%, respectively. Maple Ridge, a municipality in the northeast of the study area, showed a separate trend dominated by forest classes, with 59% of landcover being either Coniferous or Broadleaf, followed by Grass-herb at 19%, and built-up classes at 13%. 55    Figure 15 Detail of a 1m nDSM (A), 5m RapidEye imagery (B) and the 2m classification (C). Each image also shows vector objects as they looked before rasterization. Tree and building objects (coloured green and red in the top two images, respectively) match well with the features shown in the nDSM. In contrast, ground-level objects tend to be less precise, with coarser edges due to the use of RapidEye to segment those objects. In the final classification, rasterization led to simplified object borders which tended to make the classification appear more visually accurate. 56    Figure 16 Proportion of landcover class by area for municipalities covered by the landcover classification. With the exception of the University of British Columbia (UBC) at the extreme west, western municipalities have higher percentages of Built-up landcovers than other municipalities. Burnaby, to the east of Vancouver, still has greater Broadleaf than Paved coverage, however.  Grass-herb becomes increasingly dominant from west to east in the south of the study area, while Maple Ridge in the northeast and UBC in the west have greater coverage by Coniferous and Broadleaf classes than any other.57   3.4 Discussion These results indicate that it is possible to accurately resolve buildings and tree canopy (as well as more traditional urban land cover classes) by incorporating RapidEye imagery and LiDAR-derived rasters into a GEOBIA workflow. The use of the high-density LiDAR-derived layers allowed for accurate building detection and aided in the classification of coarse-textured vegetation such as trees and shrubs, consequently helping to bypass the need for very-high-resolution spectral data. At the same time, however, RapidEye’s Red Edge and NIR bands were important for class separation (Table 8), outnumbering LiDAR-derived variables for each object type. Both LiDAR-derived and spectral variables far outweighed geometric and contextual variables, which were important nonetheless for the GEOBIA ruleset (Table 6).  Although higher urban landcover classification accuracies (>90%) have been achieved using GEOBIA (e.g. Myint et al. 2011; J. P. M. O’Neil-Dunne et al. 2013; Voltersen et al. 2014), these studies have usually used true high-resolution spectral data – where pixels are generally smaller than objects of interest (T. Blaschke 2010) – as well as fewer classes (i.e. < 10). In this case, commensurate high spatial-resolution spectral data would also need to have < 2m pixel resolution, which can be costly to obtain. Using existing LiDAR datasets, some of which were initially obtained solely for DEM creation and hydrological modelling (Metro Vancouver 2016b), enabled the use of coarser RapidEye spectral data to achieve satisfactory results.   Other landcover classification studies using RapidEye have achieved similar or higher accuracies, but under quite different conditions. Adam et al. achieved 93% overall accuracy (kappa 0.92) for 11 classes using RapidEye and an RF classifier. Their study area was relatively small, however, at 25km2 (this study’s classification covers 1584.15 km2), and urban areas were a relatively minor proportion of the study area at only 1.56% (2014). Dupuy et al. achieved 90% accuracy mapping 2 classes, ‘artificial’ and non-artificial, with a GEOBIA approach that integrated RapidEye and thematic data into a classification ruleset. Tigges et al. achieved a maximum kappa of 0.83 using LiDAR and multi-temporal RapidEye to classify 8 tree genera in Berlin, Germany. For single date RapidEye, however, the maximum kappa fell to 0.52 (2013). 58   3.4.1 GEOBIA ruleset development can be tedious, but with significant payoff Developing a GEOBIA ruleset can be a time consuming, iterative process (Thomas Blaschke et al. 2014; Duro, Franklin, and Dubé 2012; J. O’Neil-Dunne, MacFaden, and Royar 2014), but it led to high class accuracies for buildings and trees (89-95% user’s and producer’s accuracies). As others have noted, it also enabled me to further classify objects using Random Forests with a much wider range of metrics than a per-pixel classification (Thomas Blaschke et al. 2014; Duro, Franklin, and Dubé 2012). Shape and context were important, particularly in the ruleset itself, with over half of the object features used for classification either geometric or contextual (Table 6). Metrics derived from these characteristics enabled me to classify objects based on, for example, their relative border to another class or whether they were very thin, thick, regularly shaped or more complex – difficult or impossible tasks from a pixel-based approach. At the same time, however, spectral and LiDAR-derived variables were the most important for RF classification, and many of these would have been useful in a pixel-based approach, as demonstrated by Duro et al. (2012).   By using spectral indices and conservative height thresholds the ruleset was transferable to portions of the study area covered by different RapidEye images and with different mixtures of landuse and landcover. This is not an unimportant consideration when needing to subset a heterogeneous study area due to large area or high data dimensionality. Ruleset transferability is also important for classification reproducibility as might be required for periodic updating.   High accuracies for Coniferous and Broadleaf classes support my assumption that they were successfully captured by the ruleset under the Tree Canopy class (i.e. did not need to be separated from other classes by Random Forests). The shrub class did see some confusion with Grass-herb, though conceptually this class exists on a gradient from grass-herb to trees and, therefore, some confusion may be inevitable. There was, however, little confusion between Shrub and either Broadleaf or Coniferous, indicating that the height threshold of 3m used in the ruleset for distinguishing Shrub from the Coniferous and Broadleaf classes was appropriate.   59  RapidEye’s lack of SWIR band meant that built-up indices such as the NDBI were not usable during the classification, nor were band ratios that could easily exploit the differences between soil, rock and pavement/asphalt in the longer wavelengths of the infrared. This led to unavoidable spectral overlap among many non-vegetated ground level classes. Though Paved and Barren did see some confusion, at only 1.9% of the study area, Barren’s impact on regional estimates of landcover, or derived measurements such as impervious surface extent, is likely to be small.   Other Built and Soil were also likely confused in part due to training points placed along rail infrastructure that were classified as Other Built. I assumed that a combination of spectral, geometric and contextual metrics exported from GEOBIA software would enable Random Forests to separate rail lines as Other Built from the Paved, Building, and Barren classes. RandomForests, however, classified rail infrastructure as a mixture of all of these classes. This also may have ‘contaminated’ the Soil class, with Other Built objects being predicted where Soil should have been, and vice versa, though that could also be due to RapidEye’s lack of SWIR band. Though RF is robust to training data noise and error (Adam et al. 2014) it could not accommodate training data that had too large a degree of spectral overlap. In hindsight, then, it might have been better to reserve the Other Built class for sports surfaces and try to use classes truly representative of the landcover of rail yards – such as paved, barren and buildings – to train the classifier.   Another limitation of the study is the number of validation points used for the Shadow and Water classes. With only 20 validation points for Shadow and 39 for Water, the robustness of the results for those classes is arguably lower than for other classes. However, internal tests using n-fold cross validation and the accuracy results from the entire study area show that the results for Shadow and Water are relatively accurate. Furthermore, with a total of 768 validation points used, I feel that an appropriate balance between efficiency of effort and explanatory power of an accuracy assessment was achieved (Foody 2009).  60  3.4.2 2m output resolution: an effective visual compromise to resolve buildings and trees One of the unique characteristics of the study is the use of LiDAR data at finer resolutions than the spectral data used. Many studies combining LiDAR and spectral data use spectral data of a finer resolution than LiDAR data (i.e. sub-metre resolution air photos) (e.g. Tigges, Lakes, and Hostert 2013; Zhou 2013), or coarsen higher-resolution LiDAR data to match spectral data (e.g. Nijland et al. 2015). In this study, however, I found that the heterogeneity of landcover present required the use of very-high-resolution LiDAR-derived rasters to accurately segment and classify objects. Coarser spectral information could then be applied to these objects. A GEOBIA workflow made the conservation of the detail obtained from LiDAR data possible.  Polygon object boundaries, the result of multiple segmentation and growing algorithms, sometimes showed complex artifacts not representative of features of interest. Rasterization of vector objects actually tended to reduce the appearance of these artifacts, with polygon-to-raster conversion and majority-smoothing filters leading to more realistic object shapes and little information loss (Figure 15). The segmentation of ground-level objects using RapidEye imagery necessarily resulted in coarser object boundaries than those objects delineated from LiDAR-derived rasters. The resultant saw-tooth edges of some Paved and Grass-Herb objects can also be seen in Figure 15.  The intention of using a GEOBIA workflow was to improve classification accuracy, not deliver polygons that could reliably be interpreted as the objects they represented. Furthermore, assessing the accuracy of objects themselves is time-intensive – requiring extensive digitization of validation objects. It was not possible to undertake such a difficult process, especially given the size of the study area. Therefore, no positional or structural (Castilla et al. 2012) accuracy assessment was conducted using vector objects, nor was count-based accuracy (Radoux and Bogaert 2017) assessed. Optimal segmentation is important for accurate classification results (Stefanski, Mack, and Waske 2013), and conducting only a pixel/area-based accuracy assessment limits my ability to say with confidence that the shape of, for example, a building or agricultural field are representative of image features. However, with a raster output as the final desired data product, I determined that a pixel-based accuracy assessment was appropriate.   61   The choice of a 2m output resolution for the classification enabled me to conserve object edges and objects of a small size. The 2m resolution does mean that ground-level objects are oversampled to about 1/6 the size of the pixels on which they are based – however, ground-level objects were in part defined in relation to objects segmented using 1 and 2m resolution rasters. Further, as can be seen in Figure 15, the oversampling of ground level objects does little to change their boundaries, while the coarsening of the edges of LiDAR-derived objects upon rasterization can actually improve their shape by reducing complexity and artifacts.  3.4.3 The impact of secondary class assignments: flexibility and enhanced accuracy Fuzzy classification has been integrated with GEOBIA for urban landuse and landcover classifications for enhanced classification accuracy (Hamedianfar and Shafri 2014; Krellenberg 2007; M. Li, Bijker, and Stein 2015; Lizarazo and Barros 2010; Wentz et al. 2014; Zhan, Molenaar, and Gorte 2000). Multiple class assignments, though not fuzzy classification using membership functions or quantitative criteria, are a way of capturing the uncertainty of class membership in heterogeneous environments like cities, or when multiple reference data types are used – especially if those types differ in MMU or spatial resolution (e.g. Thompson and Gergel 2008).   Secondary class assignments were useful in three main ways. First, points with secondary classes that were being under- or misclassified could have those classes “promoted” to the primary class position in order to improve prediction accuracies (done for shadow and other built points). Second, points with secondary classifications could be discarded as ‘impure’ points, reducing classifier and accuracy assessment confusion. Thirdly, having each level of the class hierarchy detailed for each point made data cleaning and data troubleshooting easier. Points that had been mislabelled due to interpreter error could sometimes be corrected by comparing the classifications along the levels of the class hierarchy.   The very high user’s accuracy for Non-Photosynthetic Vegetation is in large part a function of the flexibility afforded by multiple class assignments. For example, a point may have been 62  assigned both a grass-herb and a non-photosynthetic vegetation label if it represented an area of chlorotic grass. In a case like this either classification would be accepted as correct if it occurred in the classified raster. Soil was occasionally also accepted as correct, as it was found that Soil, Non-Photosynthetic Vegetation and Grass-herb existed on the landscape in a continuum depending on the extent of desiccation and density of grass coverage.   3.4.4 Temporal mismatch of data was a challenge, but not insurmountable The temporal mismatch among RapidEye and Google Earth imagery also posed some challenges. This was especially the case in agricultural areas where fields that were covered in crops in one type of imagery may have been harvested or plowed in another type. In highly dynamic areas like these, differences of even a few weeks could make a significant spectral difference. Points that were incorrectly classified were either deleted, reclassified, or, in the case of ambiguous points less than 5 pixels from a more unambiguous location, moved. Using these criteria, points on or near the edge of more than one class could be moved to a more representative location.   In contrast, the difference in time between RapidEye imagery and LiDAR tended to be less pronounced. There were cases where trees existed in the LiDAR but had been cleared for development in the RapidEye imagery. At the same time, the magnitude of discrepancies between LiDAR and RapidEye were surprisingly small, and, where discovered, were fixed manually, either by editing objects before export from GEOBIA software, or by raster reclassification. Though there was evidence of urban development and associated land clearing – particularly in suburban and rural areas – because of the durable nature of built infrastructure and the longevity of trees, already-developed areas of the classification tended to have good agreement between LiDAR and RapidEye data.   63  3.4.5 Analysis of landcover patterns: Vegetation dominates region but there are important differences among municipalities  Though the study area covered the majority of urban areas and incorporated municipalities in the region, it is still very green (Table 9), with Grass-Herb, Broadleaf and Coniferous the largest classes by area, covering 33.4%, 23.6% and 13.0% respectively. In contrast, only 22% of the study area is built-up landcover. To my knowledge, this study represents the first accurate regional assessment of landcover for Metro Vancouver, and so comparisons to other landcover estimates for the region are not easily made. However, the City of Vancouver had previously estimated its canopy cover from LiDAR data to be 18%, comparable to the nearby cities of Victoria, BC (18%) and Seattle, WA (23%) (City of Vancouver 2014). The results show that Broadleaf and Coniferous trees cover about 19% and 6% of Vancouver’s area, respectively, an increase of 7% on the previous estimate for a total canopy cover of about 25%. Methods developed by Li et al. (2015) and Seiferling et al. (2017) using Google Street view imagery to estimate street-level canopy cover have recently been applied to Vancouver by the Treepedia project of the Senseable City Lab at MIT (Ratti et al. 2017). Treepedia has estimated a median 25.9% street-level canopy cover for the City of Vancouver. Though the Treepedia estimate is only for canopy visible from the street, it is still within 1% of this study’s estimate of total canopy cover in Vancouver. As the most densely populated municipality in the Metro Vancouver region, the closeness in estimates is likely due to Vancouver’s relatively dense road network, and estimates may deviate for municipalities with larger patches of forest that are less road accessible.  In general, other studies have found North American cities to have canopy cover percentages of similar magnitudes, though other landcovers differed substantially. Canada’s other major cities have fairly similar canopy cover estimates, with Toronto estimating 26.6-28% tree canopy (City of Toronto Parks Forestry and Recreation Urban Forestry 2013) and Montreal estimating 20% c. 2007 (Beauchesne et al. 2016). In the United States Zhou, Troy and Grove (2008) used aerial imagery, LiDAR and vector data in Baltimore, MD, finding coarse vegetation (trees and shrubs) covering 34.5% of the study area, and Building and Pavement classes covering 12 and 25.9%, respectively. MacFaden et al. (2012) combined LiDAR, very high resolution orthoimagery and 64  vector data for LULC mapping of New York City with results of 20% tree canopy, 18% grass/shrubs and 58% built-up landcovers, including buildings. Sing et al. (2012) found a range of 26.6-28.9% forest and 15-15.6% impervious surface using a LiDAR-Landsat TM approach in the Charlotte, NC metropolitan region. In a tree canopy equity analysis, Schwarz et al. (2015) summarized canopy cover derived from 0.6m Quickbird, 1m National Agriculture Imagery Program, aerial photography and LiDAR data for seven U.S. cities. They found a wide range for mean tree canopy (by census block group) in each city: Philadelphia, PA, New York, NY, and Los Angeles, CA each had less than 20%, with 12.65, 16.35, and 17.61%, respectively; Baltimore, MD, Sacramento, CA and Washington, D.C. had 22.34, 23.66 and 27.52%, respectively; while at the high end, Raleigh, NC had mean tree canopy cover of 54.64%.   Though each municipality in Metro Vancouver has its own areas of greater development, possibly indicating a polycentric city, there is still a general pattern of a greater proportion of built-up landcover in the west than east (Figure 14, Figure 16). Vancouver and neighbouring municipalities Burnaby and North Vancouver (city) show much higher proportions of Buildings and Paved classes – with combined percentages of 58.4, 42.3, and 59.1%, respectively – than municipalities further east. That said, there are still considerable variations within municipalities, with Burnaby, to the immediate east of Vancouver, seeing higher levels of Broadleaf tree canopy (26.3%) than any other landcover, in part due to large parks. Vancouver, too, has a greener west than east side which I plan to explore as part of a greenspace equity analysis in future work.   These landcover patterns are a result of a number of factors, which go beyond the scope of this analysis, but landuse and the history of development play a large role. For example, the fertile soil of the Fraser valley is prime agricultural land, and despite recent development much of the south of the study area is still relatively rural (Metro Vancouver 2013). Agriculture, and especially protected agricultural land, therefore, is a key driver of the landcover found in Surrey, Langley and Abbotsford, with their high proportions of the Grass-Herb class (31.5, 43.2, and 49.6%, respectively). These more suburban and rural municipalities are also more likely to have residents commuting to another part of Metro Vancouver for work. Vancouver and neighbouring UBC together have the highest percentage of residents working in the same area in which they 65  live, at 66%, while other subregions in Metro Vancouver range from 24-50% (Metro Vancouver 2014).  Vancouver’s influence in the region as the centre of commerce (22% of the region’s employment is in and around downtown Vancouver) and the location of highest population densities (Metro Vancouver 2013) can be seen in the pattern of built-up landcover, which generally decreases in proportion with increasing distance from Vancouver, and lends support to those that have conceptualized a concentric urban greenness gradient from a single densely built-up area (S. S. Chen et al. 2015; Guérois and Pumain 2008). Indeed, Lu et al. (2016) specifically found that vegetative cover increases as distance from downtown Vancouver increased. Cities in developed economies are rarely monocentric, however (Kloosterman and Musterd 2001; Reviewed in McKinney 2006), and the City of Vancouver’s dominance in the region is expected to diminish as other municipalities grow and develop (Metro Vancouver 2013), indicating the need for regular landcover updates to effectively monitor how the region is changing.  3.5 Conclusion This work shows that even using what is often medium-resolution spectral data in an urban context, and with more than 10 landcover classes, overall accuracies approaching 90% can be achieved. The addition of LiDAR data improved class accuracies, especially for buildings and coarse-textured vegetation. Used in concert with GEOBIA, LiDAR also improved the visual accuracy of the final classified map, with well-defined building edges and tree crowns. There are often many disparate geospatial datasets available in metropolitan regions, which can be challenging to integrate. However, both GEOBIA and Random Forests can accommodate multiple data types, and Random Forests is well-positioned to utilize the many metrics that can be exported from GEOBIA software. These characteristics enabled me to use spectral, LIDAR-derived and geometric metrics from multiple datasets to accurately classify a large portion of a metropolitan region at a high spatial resolution. 66  Chapter 4: Conclusion This research makes contributions to urban ecology, urban remote sensing and UGS/ES equity studies via a literature review of urban ES justice issues (Chapter 2:) and an accurate, high spatial resolution method for mapping urban land cover (Chapter 3:). The aim of Chapter 2 was to provide a transdisciplinary investigation of urban ES justice issues for both ecologists and social scientists. Chapter 3 presents a novel method for accurately mapping urban landcover using multispectral imagery and 3D structure information from LiDAR. Together, the two chapters provide a basis for further work to assess UGS/ES equity in Metro Vancouver, and elsewhere, in a way that accurately captures public and private greenspace along with a suite of equity assessment methods.   Chapter 2 answers the question “How has environmental justice been considered and incorporated into urban ES research?” The chapter highlights how quantitative, spatially-explicit methods can be used to uncover inequity in urban ES (and UGS) as well as qualitative arguments that ES are not objective realities or neutral constructs. Using both quantitative and qualitative methodologies, I catalogued methods for urban ES justice assessment and related conceptual frameworks from papers obtained through a Web of Science search. A novel environmental justice index for quantifying the breadth of justice inquiry in a document was developed to better understand the types of justice issues considered in qualitative, quantitative and hybrid articles. I found that studies were mainly based in the developed world and China, and incorporated an average of 4 justice issues out of a possible 10. Though I found a number of articles with quantitative methods that can be used to understand UGS/ES equity issues, few studies were willing to explicitly label a given distribution as unjust. Similarly, no qualitative papers provided a definitive test of justice, but many still offered valuable critiques of, and insights into, ES. To my knowledge, this review is the first of its kind focused on urban ES justice issues from a transdisciplinary body of work.   Chapter 3 integrates multispectral imagery with multiple LiDAR datasets of varying point densities to map landcover in a large metropolitan region, achieving an overall accuracy approaching 90% with 12 landcover classes.  A GEOBIA workflow and a RandomForests 67  classifier enabled the use of a variety of data types and derived metrics, improving classification accuracies. Using very-high spatial resolution rasters derived from LiDAR I created objects for classification at a finer scale than using only multispectral data. LiDAR’s 3D structure information also allowed me to accurately map both buildings and trees.   While other studies have combined LiDAR and RapidEye, to my knowledge none have done it over such a large area, for so many classes, or with multiple LiDAR data sets. Also unique was the use of a GEOBIA ruleset to classify buildings and trees first, before using RandomForests to further refine and expand the classification. The resultant landcover map is a strong basis for analyzing greenspace in Metro Vancouver – especially tree canopy. Indeed, as the first accurate landcover map of the region, this work will provide a new baseline for city managers and researchers to assess landcover change, water quality, carbon storage and other features and processes at a landscape level.  Together, these two chapters provide methodological guidance on how to accurately map greenspace and assess the equity of its distribution. Mapping greenspace is an essential baseline for quantitative inquiry (Gaston, Ávila-Jiménez, and Edmondson 2013b; Holt et al. 2015), however, as shown in Chapter 2, simply mapping greenspace is not enough. Because ES are valued differently by different groups (Ernstson 2013a; Wolch, Byrne, and Newell 2014b), making mechanistic assessments of trade-offs among ES that flow from UGS is difficult. With a lack of stakeholder involvement and governance perspectives in urban ES literature (Luederitz et al. 2015), a lack of quantitative case studies in political ecology and critical geography (Chapter 2), and a rapidly urbanizing world (United Nations 2014), an understanding of UGS and urban ES equity is of critical importance.   For Metro Vancouver, this work can help support regional planning goals that include protection and enhancement of natural features and their connectivity, as well as the development of healthy communities with a variety of services and amenities (Metro Vancouver 2013). Metro Vancouver even has a specific goal of supplementing ecosystem services by improving and adding to sources of ES flows (Metro Vancouver Regional District 2011). Chapter 3 will help 68  regional planners to better account for the range of sites around the region that are providing ecosystem services, while Chapter 2 will help to push regional and city planners to include local stakeholders in ES assessment and valuation.  4.1 Limitations of Research  Each chapter has some important limitations. Chapter 2 results were compiled by a single reviewer, meaning that the subjectivity inherent choosing and interpreting articles could not be mitigated by others. No review can be totally exhaustive, and Chapter 2 certainly misses papers germane to the research question. Further, the interdisciplinary nature of the review likely means that articles were missed due to differing disciplinary terminology, especially regarding the benefits people derive from ecosystems. Finally, the justice index developed in Chapter 2 was developed to effectively examine the breadth of justice issues in a journal articles, but does not address the depth of inquiry of any one issue.   The accuracy of results in Chapter 3 could have been improved with a short-wave infrared (SWIR) band to better differentiate non-vegetated landcovers. Confusion between the Soil and Other Built classes was partly due to methodological shortcomings as well. The robustness of accuracy results for the shadow and water classes may also be criticized due to the small number of validation points for each of those classes, though internal tests using many more points were consistent with the presented results.  Finally, temporal mismatches between the LiDAR and RapidEye data, as well as the Google Earth imagery used for validation likely reduced classification accuracies. The extent of this impact was not quantified, but based on visual assessments and the extent of manual corrections to object polygons and validation points it was not a serious enough issue to undermine the overall classification methodology.   4.2 Future Research Existing research has highlighted UGS inequalities in a number of North American cities. Boone et al. (2009) found that African Americans in Baltimore, MD had better access to parks than whites, but that predominantly black areas had higher park congestion due to a number of small urban parks. Dai (2011) also found that racialized people had poorer access to public 69  greenspaces in Atlanta, GA, while Su et al. (2009) showed that cumulative environmental hazard increased with the percentage of non-whites in Los Angeles, CA. Meanwhile, Pham et al. found evidence that lower income correlated with less greenspace in Montréal, QC, and Nesbitt and Meitner found that greenness varied with income, education and race in Portland, OR, showing evidence of inequity. Similar patterns have been found the world over (e.g. Dobbs, Kendal, and Nitschke 2014; Lakes, Brückner, and Krämer 2014; K. Schwarz et al. 2015; Steenberg et al. 2015).  In light of this, future work should build on the landcover classification developed in Chapter 3 to explore greenspace connectivity and equity in the Metro Vancouver region. By combining greenspace equity and connectivity analyses I hope to put ecocentric conservation goals in the context of environmental justice. I hope to determine the extent to which greenspace falls along socio-economic gradients in the region, and whether those trends can be related to the role individual patches play in greenspace connectivity. I hope to determine whether there is evidence for greenspace inequity in Metro Vancouver, and how the region compares with other urban areas in the developed world. Growing out of a connectivity analysis, I will also investigate whether larger, well-connected patches tend to have, for example, wealthier neighbourhoods, and the role that patches with poorer neighbourhoods play in Metro Vancouver’s greenspace network. If smaller, more isolated patches tend to have less-wealthy surroundings, are they nonetheless important stepping stone patches?   This work will require an explicitly spatial methodology, combining census data and greenspace data. A connectivity analysis will be conducted using a graph theory approach (Pascual-Hortal and Saura 2006; Saura and Fuente 2011) for a suite of species that use habitat represented by the land cover classes mapped in Chapter 3. Resulting connectivity metrics can then be summarized for a census area along with socio-economic variables of interest representing income, wealth, and race, among others. Metrics describing the amount and accessibility of greenspace will also be computed for each census area. Connectivity and greenspace metrics can then be compared with socio-economic variables, and spatially explicit models developed. Possible methods 70  include overlay and cluster analysis methods noted in Chapter 2, as well as spatial autoregression (Reviewed in Long 1998; Pace and Barry 1997).  4.3 Closing Thoughts Thomas Piketty has shown that wealth inequality is at an historic high (Piketty 2014), resulting in a range of negative outcomes including lower quality of life for the less-wealthy, the undermining of democratic institutions, economic stagnation, and civil unrest. If wealth and income vary with UGS, then there is reason to be concerned that UGS inequity may also be on the rise. Comprehensively addressing the problem of inequality, therefore, requires addressing inequalities in UGS distribution and the ecosystem services that flow from them. To not do so is to allow the benefits of UGS to disproportionately flow to those that are already better off. If we accept that the pursuit of justice is right, and that the pursuit of a more equal distribution of benefits is just, then we must work to end environmental injustice by creating a more equitable distribution of UGS and the ES that flow from them.  71  References Adam, Elhadi, Onisimo Mutanga, John Odindi, and Elfatih M. Abdel-Rahman. 2014. “Land-Use/cover Classification in a Heterogeneous Coastal Landscape Using RapidEye Imagery: Evaluating the Performance of Random Forest and Support Vector Machines Classifiers.” International Journal of Remote Sensing 35(10): 3440–58. http://www-tandfonline-com.ezproxy.library.ubc.ca/doi/abs/10.1080/01431161.2014.903435?src=recsys (October 1, 2015). Agudelo-Vera, Claudia Marcela, Adriaan Mels, Karel Keesman, and Huub Rijnaarts. 2012. “The Urban Harvest Approach as an Aid for Sustainable Urban Resource Planning.” Journal of Industrial Ecology 16(6): 839–50. Alberti, Marina. 2010. “Maintaining Ecological Integrity and Sustaining Ecosystem Function in Urban Areas.” Current Opinion in Environmental Sustainability 2(3): 178–84. http://www.sciencedirect.com/science/article/pii/S1877343510000576 (January 13, 2016). Alves, Humberto P. F., and Ricardo Ojima. 2012. “Environmental Inequality in São Paulo City: An Analysis of Differential Exposure of Social Groups to Situations of Environmental Risk.” In Urbanization and Sustainability: Linking Urban Ecology, Environmental Justice and Global Environmental Change, eds. Christopher G. Boone and Michail Fragkias. Springer, 117–35. Anderson, By James R, Ernest E Hardy, John T Roach, and Richard E Witmer. 1976. Geological Survey Professional Paper 964 A Land Use And Land Cover Classification System For Use With Remote Sensor Data. http://igett.delmar.edu/08/Cohort2/LearningUnitExamplesnodata/LUDanScollon/Support_Docs/refDoc.LULC_class_system_1976.pdf. Andrew, M. E., M. A. Wulder, and T. A. Nelson. 2014. “Potential Contributions of Remote Sensing to Ecosystem Service Assessments.” Progress in Physical Geography 38(3): 328–53. http://ppg.sagepub.com/content/38/3/328.abstract (May 1, 2015). Angelone, Sonia, and Rolf Holderegger. 2009. “Population Genetics Suggests Effectiveness of Habitat Connectivity Measures for the European Tree Frog in Switzerland.” Journal of Applied Ecology 46(4): 879–87. Anguelovski, Isabelle, and Joan Martínez Alier. 2014. “The ‘Environmentalism of the Poor’ 72  Revisited: Territory and Place in Disconnected Glocal Struggles.” Ecological Economics 102: 167–76. https://apps.webofknowledge.com/full_record.do?product=UA&search_mode=AdvancedSearch&qid=15&SID=2Bw5iVlCLEa8hfJJc65&page=12&doc=116 (August 16, 2015). Ash, Caroline et al. 2008. “Reimagining Cities.” Science 319(February): 739. Baatz, Martin, and Arno Schäpe. 1999. “Object-Oriented and Multi-Scale Image Analysis in Semantic Networks.” In 2nd International Symposium: Operationalization of Remote Sensing, , 7–13. Beauchesne, Natacha et al. 2016. Sustainable Montréal 2016-2020. Montreal, QC. http://ville.montreal.qc.ca/pls/portal/docs/page/d_durable_en/media/documents/plan_de_dd_en_lr.pdf. BenDor, Todd, and Audrey Stewart. 2011. “Land Use Planning and Social Equity in North Carolina’s Compensatory Wetland and Stream Mitigation Programs.” Environmental management 47(2): 239–53. https://apps.webofknowledge.com/full_record.do?product=UA&search_mode=AdvancedSearch&qid=15&SID=2Bw5iVlCLEa8hfJJc65&page=9&doc=89 (November 2, 2015). Bendt, Pim, Stephan Barthel, and Johan Colding. 2013. “Civic Greening and Environmental Learning in Public-Access Community Gardens in Berlin.” Landscape and Urban Planning 109(1): 18–30. http://www.sciencedirect.com/science/article/pii/S0169204612002745 (October 14, 2015). Bigsby, Kevin M, Melissa R McHale, and George R Hess. 2014. “Urban Morphology Drives the Homogenization of Tree Cover in Baltimore, MD, and Raleigh, NC.” Ecosystems 17(2): 212–27. Bindel, Marcus, Sören Hese, Christian Berger, and Christiane Schmullius. 2011. “Evaluation of Red-Edge Spectral Information for Biotope Mapping Using RapidEye.” SPIE Remote Sensing 8174: 81740X–81740X–9. http://proceedings.spiedigitallibrary.org/proceeding.aspx?articleid=1269599. BlackBridge. 2014. The RapidEye Red Edge Band. https://resa.blackbridge.com/files/2014-06/Red_Edge_White_Paper.pdf. Blaschke, T. 2010. “Object Based Image Analysis for Remote Sensing.” ISPRS Journal of 73  Photogrammetry and Remote Sensing 65(1): 2–16. http://www.sciencedirect.com/science/article/pii/S0924271609000884 (July 10, 2014). Blaschke, Thomas et al. 2014. “Geographic Object-Based Image Analysis - Towards a New Paradigm.” ISPRS journal of photogrammetry and remote sensing : official publication of the International Society for Photogrammetry and Remote Sensing (ISPRS) 87(100): 180–91. http://www.sciencedirect.com/science/article/pii/S0924271613002220 (April 30, 2016). Bloom, Allan. 1991. Giants and Dwarfs: Essays 1960-1990. New York: Simon & Schuster. Boone, Christopher G., Geoffrey L. Buckley, J. Morgan Grove, and Chona Sister. 2009. “Parks and People: An Environmental Justice Inquiry in Baltimore, Maryland.” Annals of the Association of American Geographers 99(4): 767–87. http://www.tandfonline.com/doi/abs/10.1080/00045600903102949 (August 23, 2017). Booth, Josephine E, Kevin J Gaston, and Paul R Armsworth. 2010. “Who Benefits from Recreational Use of Protected Areas ?” Ecology And Society 15(3): 19. http://www.ecologyandsociety.org/vol15/iss3/art19/. Breiman, Leo, and Adele Cutler. 2015. “randomForest: Breiman and Cutler’s Random Forests for Classification and Regression.” https://cran.r-project.org/package=randomForest. Brown, A., A. Dayal, and C. Rumbaitis Del Rio. 2012. “From Practice to Theory: Emerging Lessons from Asia for Building Urban Climate Change Resilience.” Environment and Urbanization 24(2): 531–56. http://eau.sagepub.com/content/24/2/531 (March 10, 2015). Byrne, J., and J. Wolch. 2009. “Nature, Race, and Parks: Past Research and Future Directions for Geographic Research.” Progress in Human Geography 33(6): 743–65. http://phg.sagepub.com/cgi/doi/10.1177/0309132509103156. Cadenasso, Mary L., Steward T. A. Pickett, and Kirsten Schwarz. 2007. “Spatial Heterogeneity in Urban Ecosystems: Reconceptualizing Land Cover and a Framework for Classification.” Frontiers in Ecology and the Environment 5(2): 80–88. http://doi.wiley.com/10.1890/1540-9295(2007)5[80:SHIUER]2.0.CO;2 (June 21, 2016). Castilla, G et al. 2012. “An Integrated Framework for Assessing the Accuracy of Geobia Landcover Products.” In Proceedings of the 4th GEOBIA, , 572–75. http://oa.upm.es/21039/1/INVE_MEM_2012_129483.pdf. Castrence, Miguel et al. 2014. “Mapping Urban Transitions Using Multi-Temporal Landsat and 74  DMSP-OLS Night-Time Lights Imagery of the Red River Delta in Vietnam.” Land 3(1): 148–66. Cen, Xiaoteng et al. 2015. “Coupling Intensive Land Use and Landscape Ecological Security for Urban Sustainability: An Integrated Socioeconomic Data and Spatial Metrics Analysis in Hangzhou City.” Sustainability 7(2): 1459–82. http://www.mdpi.com/2071-1050/7/2/1459/. Chan, Kai M A, and Terre Satterfield. 2013. “Justice, Equity, and Biodiversity.” In The Encyclopedia of Biodiversity., ed. S. A. Levin. Oxford: Elsevier Ltd, 434–41. Chance, Curtis M., Nicholas C. Coops, Ken Crosby, and Neal Aven. 2016. “Spectral Wavelength Selection and Detection of Two Invasive Plant Species in an Urban Area.” Canadian Journal of Remote Sensing 42(1): 27–40. http://www.tandfonline.com/doi/full/10.1080/07038992.2016.1143330 (July 31, 2017). Chen, S S, Yue Yan, Qun Gao, and Dene Liu. 2015. “Quantifying Circular Urban Expansion Patterns of Compact Chinese Cities: The Case of Yangtze River Delta, China.” Environment and Planning B: Planning and Design 42: 279–99. Chen, Yunhao, Wei Su, Jing Li, and Zhongping Sun. 2009. “Hierarchical Object Oriented Classification Using Very High Resolution Imagery and LIDAR Data over Urban Areas.” Advances in Space Research 43(7): 1101–10. http://www.sciencedirect.com/science/article/pii/S0273117708005735 (November 4, 2015). Chen, Ziyue, and Bingbo Gao. 2014. “An Object-Based Method for Urban Land Cover Classification Using Airborne Lidar Data.” IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing 7(10): 4243–54. http://ieeexplore.ieee.org/lpdocs/epic03/wrapper.htm?arnumber=6856202 (November 9, 2015). Childers, Daniel et al. 2015. “An Ecology for Cities: A Transformational Nexus of Design and Ecology to Advance Climate Change Resilience and Urban Sustainability.” Sustainability 7(4): 3774–91. http://www.mdpi.com/2071-1050/7/4/3774/ (October 13, 2016). Cilliers, Sarel et al. 2014. “Sustainable Urban Landscapes: South African Perspectives on Transdisciplinary Possibilities.” Landscape and Urban Planning 125: 260–70. https://apps.webofknowledge.com/full_record.do?product=UA&search_mode=AdvancedSearch&qid=15&SID=2Bw5iVlCLEa8hfJJc65&page=10&doc=98 (August 27, 2015). 75  City of Toronto Parks Forestry and Recreation Urban Forestry. 2013. Every Tree Counts: A Portrait of Toronto’s Urban Forest. Toronto, Ontario. https://www1.toronto.ca/City Of Toronto/Parks Forestry & Recreation/Urban Forestry/Files/pdf/E/every_tree_counts.pdf. City of Vancouver. 2014. City of Vancouver Urban Forest Strategy. Vancouver. http://vancouver.ca/home-property-development/urban-forest-strategy.aspx. Clergeau, Philippe, and Françoise Burel. 1997. “The Role of Spatio-Temporal Patch Connectivity at the Landscape Level: An Example in a Bird Distribution.” Landscape and Urban Planning 38(1–2): 37–43. Cohen, Marianne et al. 2012. “Urban Biodiversity and Social Inequalities in Built-up Cities: New Evidences, next Questions. The Example of Paris, France.” Landscape and Urban Planning 106(3): 277–87. http://www.sciencedirect.com/science/article/pii/S0169204612000941 (September 27, 2015). Colding, Johan et al. 2013. “Urban Green Commons: Insights on Urban Common Property Systems.” Global Environmental Change 23(5): 1039–51. http://www.sciencedirect.com/science/article/pii/S0959378013000800 (November 1, 2015). Coops, Nicholas C., Jason Duffe, and Cathy Koot. 2010. “Assessing the Utility of Lidar Remote Sensing Technology to Identify Mule Deer Winter Habitat.” Canadian Journal of Remote Sensing 36(2): 81–88. http://www.tandfonline.com/doi/abs/10.5589/m10-029 (December 30, 2016). Cox, Wendell, and Hugh Pavletich. 2014. 10th Annual Demographia International Housing Affordability Survey : 2014. Belleville, Illinois. http://www.demographia.com/dhi.pdf. Crouse, Dan L et al. 2017. “Urban Greenness and Mortality in Canada’s Largest Cities: A National Cohort Study.” The Lancet Planetary Health 1(7): e289–97. http://linkinghub.elsevier.com/retrieve/pii/S2542519617301183. Dai, Dajun. 2011. “Racial/ethnic and Socioeconomic Disparities in Urban Green Space Accessibility: Where to Intervene?” Landscape and Urban Planning 102(4): 234–44. http://dx.doi.org/10.1016/j.landurbplan.2011.05.002. Davis, Amélie Y. et al. 2012. “Green Infrastructure and Bird Diversity across an Urban Socioeconomic Gradient.” http://www.esajournals.org/doi/abs/10.1890/ES12-00126.1 (November 2, 2015). 76  Daw, Tim, Katrina Brown, Sergio Rosendo, and Robert Pomeroy. 2011. “Applying the Ecosystem Services Concept to Poverty Alleviation: The Need to Disaggregate Human Well-Being.” Environmental Conservation 38(4): 370–79. http://journals.cambridge.org/abstract_S0376892911000506 (June 22, 2015). Díaz, Sandra et al. 2015. “The IPBES Conceptual Framework — Connecting Nature and People.” Current Opinion in Environmental Sustainability 14: 1–16. http://www.sciencedirect.com/science/article/pii/S187734351400116X (May 15, 2017). Dixo, Marianna, Jean Paul Metzger, Jo??o S. Morgante, and Kelly R. Zamudio. 2009. “Habitat Fragmentation Reduces Genetic Diversity and Connectivity among Toad Populations in the Brazilian Atlantic Coastal Forest.” Biological Conservation 142(8): 1560–69. Dobbs, Cynnamon, Dave Kendal, and Craig R. Nitschke. 2014. “Multiple Ecosystem Services and Disservices of the Urban Forest Establishing Their Connections with Landscape Structure and Sociodemographics.” Ecological Indicators 43: 44–55. http://www.sciencedirect.com/science/article/pii/S1470160X14000594 (January 16, 2015). von Döhren, Peer, and Dagmar Haase. 2015. “Ecosystem Disservices Research: A Review of the State of the Art with a Focus on Cities.” Ecological Indicators 52: 490–97. Duro, Dennis C., Steven E. Franklin, and Monique G. Dubé. 2012. “A Comparison of Pixel-Based and Object-Based Image Analysis with Selected Machine Learning Algorithms for the Classification of Agricultural Landscapes Using SPOT-5 HRG Imagery.” Remote Sensing of Environment 118: 259–72. eCognition Community Members. 2009. “Features Reference.” eCognition Community Wiki. http://community.ecognition.com/home/features-reference (February 12, 2017). Eigenbrod, Felix et al. 2010. “The Impact of Proxy-Based Methods on Mapping the Distribution of Ecosystem Services.” Journal of Applied Ecology 47(2): 377–85. Ernstson, Henrik. 2013a. “The Social Production of Ecosystem Services: A Framework for Studying Environmental Justice and Ecological Complexity in Urbanized Landscapes.” Landscape and Urban Planning 109(1): 7–17. http://www.sciencedirect.com/science/article/pii/S0169204612002861 (June 12, 2015). ———. 2013b. “The Social Production of Ecosystem Services: A Framework for Studying Environmental Justice and Ecological Complexity in Urbanized Landscapes.” Landscape 77  and Urban Planning 109(1): 7–17. http://linkinghub.elsevier.com/retrieve/pii/S0169204612002861 (August 25, 2017). ESRI (Environmental Systems Research Institute, Inc.). 2014. “ArcGIS 10.3.0.4322.” www.esri.com. Estoque, Ronald C., and Yuji Murayama. 2014. “Measuring Sustainability Based Upon Various Perspectives: A Case Study of a Hill Station in Southeast Asia.” AMBIO 43(7): 943–56. https://apps.webofknowledge.com/full_record.do?product=UA&search_mode=AdvancedSearch&qid=15&SID=2Bw5iVlCLEa8hfJJc65&page=10&doc=96 (October 9, 2015). Faehnle, Maija et al. 2014. “How Can Residents’ Experiences Inform Planning of Urban Green Infrastructure? Case Finland.” Landscape and Urban Planning 130: 171–83. http://www.sciencedirect.com/science/article/pii/S0169204614001753 (September 24, 2015). Ferreras, P. 2001. “Landscape Structure and Asymmetrical Inter-Patch Connectivity in a Metapopulation of the Endangered Iberian Lynx.” Biological Conservation 100(1): 125–36. Fisher, Brendan, R. Kerry Turner, and Paul Morling. 2009. “Defining and Classifying Ecosystem Services for Decision Making.” Ecological Economics 68(3): 643–53. http://www.sciencedirect.com/science/article/pii/S0921800908004424 (July 9, 2014). Foody, Giles M. 2009. “Sample Size Determination for Image Classification Accuracy Assessment and Comparison.” International Journal of Remote Sensing 30(20): 5273–91. http://www.tandfonline.com/doi/abs/10.1080/01431160903130937 (June 23, 2017). Frolking, Steve, Tom Milliman, Karen C Seto, and Mark a Friedl. 2013. “A Global Fingerprint of Macro-Scale Changes in Urban Structure from 1999 to 2009.” Environmental Research Letters 8(2): 24004. Gaston, Kevin J., María L. Ávila-Jiménez, and Jill L. Edmondson. 2013a. “Managing Urban Ecosystems for Goods and Services” ed. Julia Jones. Journal of Applied Ecology 50(4): 830–40. http://www.readcube.com/articles/10.1111/1365-2664.12087? (July 14, 2014). ———. 2013b. “Managing Urban Ecosystems for Goods and Services.” Journal of Applied Ecology 50(Gaston 2003): 830–40. Gómez-Baggethun, Erik, and David N. Barton. 2013. “Classifying and Valuing Ecosystem Services for Urban Planning.” Ecological Economics 86: 235–45. 78  http://www.sciencedirect.com/science/article/pii/S092180091200362X (July 11, 2014). Guérois, Marianne, and Denise Pumain. 2008. “Built-up Encroachment and the Urban Field: A Comparison of Forty European Cities.” Environment and Planning A 40(9): 2186–2203. Haala, Norbert, and Claus Brenner. 1999. “Extraction of Buildings and Trees in Urban Environments.” ISPRS Journal of Photogrammetry and Remote Sensing 54(2): 130–37. Haaland, Christine, and Cecil Konijnendijk van den Bosch. 2015. “Challenges and Strategies for Urban Green-Space Planning in Cities Undergoing Densification: A Review.” Urban Forestry and Urban Greening 14: 760–71. Haines-Young, R, and Marion Potschin. 2011. “Common International Classification of Ecosystem Services (CICES): 2011 Update.” Expert Meeting on Ecosystem Accounts …: 1–17. http://www.cices.eu/wp-content/uploads/2009/11/CICES_Update_Nov2011.pdf. Hamedianfar, Alireza, and Helmi Z.M. Shafri. 2014. “Development of Fuzzy Rule-Based Parameters for Urban Object-Oriented Classification Using Very High Resolution Imagery.” Geocarto International 29(3): 268–92. http://www.tandfonline.com/doi/abs/10.1080/10106049.2012.760006 (November 4, 2015). Hanski, Ilkka, and Michael Gilpin. 1991. “Metapopulation Dynamics : Brief History and Conceptual Domain.” Biological Journal of the Linnean Society 42: 3–16. Hartfield, Kyle A., Katheryn I. Landau, and Willem J. D. van Leeuwen. 2011. “Fusion of High Resolution Aerial Multispectral and LiDAR Data: Land Cover in the Context of Urban Mosquito Habitat.” Remote Sensing 3(12): 2364–83. http://www.mdpi.com/2072-4292/3/11/2364 (November 3, 2015). Harvey, David. 1994. “The Social Construction of Space and Time: A Relational Theory.” Geographical Review of Japan 67(2): 126–35. Hermosilla, Txomin et al. 2014. “Using Street Based Metrics to Characterize Urban Typologies.” Computers, Environment and Urban Systems 44: 68–79. http://www.sciencedirect.com/science/article/pii/S0198971513001166 (September 6, 2015). Hermosilla, Txomin, Luis A. Ruiz, Jorge A. Recio, and Javier Estornell. 2011. “Evaluation of Automatic Building Detection Approaches Combining High Resolution Images and LiDAR Data.” Remote Sensing 3(12): 1188–1210. http://www.mdpi.com/2072-4292/3/6/1188 (August 15, 2015). 79  Heynen, Nikolas C. 2003. “The Scalar Production of Injustice within the Urban Forest.” Antipode 35(5): 980–98. http://doi.wiley.com/10.1111/j.1467-8330.2003.00367.x (November 2, 2015). Holt, Alison R., Meghann Mears, Lorraine Maltby, and Philip Warren. 2015. “Understanding Spatial Patterns in the Production of Multiple Urban Ecosystem Services.” Ecosystem Services 16: 33–46. http://www.sciencedirect.com/science/article/pii/S2212041615300243 (October 9, 2015). Hubacek, Klaus, and Jakub Kronenberg. 2013. “Synthesizing Different Perspectives on the Value of Urban Ecosystem Services.” Landscape and Urban Planning 109(1): 1–6. http://www.sciencedirect.com/science/article/pii/S0169204612002915 (October 20, 2015). Hudak, Andrew Thomas, Jeffrey Scott Evans, and Alistair Matthew Stuart Smith. 2009. “LiDAR Utility for Natural Resource Managers.” Remote Sensing 1(4): 934–51. Ibes, Dorothy C. 2015. “A Multi-Dimensional Classification and Equity Analysis of an Urban Park System: A Novel Methodology and Case Study Application.” Landscape and Urban Planning 137: 122–37. https://apps.webofknowledge.com/full_record.do?product=UA&search_mode=AdvancedSearch&qid=15&SID=2Bw5iVlCLEa8hfJJc65&page=10&doc=91 (September 27, 2015). ICSU, UNESCO, and UNU. 2008. Ecosystem Change and Human Well-Being: Research and Monitoring Priorities Based on the Millennium Ecosystem Assessment. Paris. IPCC. 2014. Climate Change 2014: Impacts, Adaptation and Vulnerability - Contributions of the Working Group II to the Fifth Assessment Report Summary for Policymakers. Irwin, Elena G, and Nancy E Bockstael. 2007. “The Evolution of Urban Sprawl: Evidence of Spatial Heterogeneity and Increasing Land Fragmentation.” Proceedings of the National Academy of Sciences of the United States of America 104(52): 20672–77. http://www.pnas.org/content/104/52/20672.abstract (July 28, 2015). Isenburg, Martin. 2014. “Lastools.” https://rapidlasso.com/lastools/. ———. 2017. “Lastools.” https://rapidlasso.com/lastools/. Jebur, Mustafa Neamah, Helmi Zulhaidi Mohd Shafri, Biswajeet Pradhan, and Mahyat Shafapour Tehrany. 2013. “Per-Pixel and Object-Oriented Classification Methods for Mapping Urban Land Cover Extraction Using SPOT 5 Imagery.” Geocarto International 80  29(7): 792–806. http://www-tandfonline-com.ezproxy.library.ubc.ca/doi/abs/10.1080/10106049.2013.848944 (November 9, 2015). Jenerette, G. Darrel, Sharon L. Harlan, William L. Stefanov, and Chris A. Martin. 2011. “Ecosystem Services and Urban Heat Riskscape Moderation: Water, Green Spaces, and Social Inequality in Phoenix, USA.” Ecological Applications 21(7): 2637–51. http://www.esajournals.org/doi/abs/10.1890/10-1493.1 (October 9, 2015). Jennings, Viniece, and Cassandra Johnson Gaither. 2015. “Approaching Environmental Health Disparities and Green Spaces: An Ecosystem Services Perspective.” International journal of environmental research and public health 12(2): 1952–68. https://apps.webofknowledge.com/full_record.do?product=UA&search_mode=AdvancedSearch&qid=15&SID=2Bw5iVlCLEa8hfJJc65&page=10&doc=93 (September 3, 2015). Jesdale, Bill M., Rachel Morello-Frosch, and Lara Cushing. 2013. “The Racial/ Ethnic Distribution of Heat Risk-Related Land Cover in Relation to Residential Segregation.” Environmental Health Perspectives 121(7): 811–17. Jing, Wenlong, Yaping Yang, Xiafang Yue, and Xiaodan Zhao. 2015. “Mapping Urban Areas with Integration of DMSP/OLS Nighttime Light and MODIS Data Using Machine Learning Techniques.” Remote Sensing 7(9): 12419–39. http://www.mdpi.com/2072-4292/7/9/12419/. Johnson, Craig A. 2012. “Governing Climate Displacement: The Ethics and Politics of Human Resettlement.” Environmental Politics 21(2): 308–28. http://www.tandfonline.com/doi/abs/10.1080/09644016.2012.651905#.VInfcaaZOTU (December 11, 2014). Jost, John T., and Aaron C. Kay. 2010. “Social Justice: History, Theory, and Research.” In Handbook of Social Psychology, Vol. 2, eds. Susan T. Fiske, Daniel T. Gilbert, and Gardner Lindzey. Hoboken, N.J.: John Wiley & Sons Inc, 1122–65. http://resolve.library.ubc.ca/cgi-bin/catsearch?bid=4424754. Kabisch, Nadja, and Dagmar Haase. 2014. “Green Justice or Just Green? Provision of Urban Green Spaces in Berlin, Germany.” Landscape and Urban Planning 122: 129–39. https://apps.webofknowledge.com/full_record.do?product=UA&search_mode=AdvancedSearch&qid=15&SID=2Bw5iVlCLEa8hfJJc65&page=5&doc=44 (December 2, 2014). 81  Kabisch, Nadja, Michael Strohbach, Dagmar Haase, and Jakub Kronenberg. 2016. “Urban Green Space Availability in European Cities.” Ecological Indicators 70: 586–96. http://linkinghub.elsevier.com/retrieve/pii/S1470160X16300504 (March 20, 2017). Kennedy, Christopher, John Cuddihy, and Joshua Engel-yan. 2007. “The Changing Metabolism of Cities.” Journal of Industrial Ecology 11(2): 43–59. Kim, Yongmin, and Yongil Kim. 2013. “Improved Classification Accuracy Based on the Output-Level Fusion of High-Resolution Satellite Images and Airborne LiDAR Data in Urban Area.” IEEE Geoscience and Remote Sensing Letters 11(3): 636–40. Kloosterman, Robert C., and Sako Musterd. 2001. “The Polycentric Urban Region: Towards a Research Agenda.” Urban Studies 38(4): 623–33. http://journals.sagepub.com/doi/10.1080/00420980120035259 (October 5, 2017). Krellenberg, Kerstin. 2007. “Urban Environmental Monitoring in Buenos Aires – Determining Green Areas.” In Applied Remote Sensing for Urban Planning, Governance and Sustainability, eds. Maik Netzband, William L. Stefanov, and Charles Redman. Berlin, Heidelberg: Springer, 77–95. Kuhn, Max. 2017. “Classification and Regression Training (Caret).” https://github.com/topepo/caret/. de la Barrera, Francisco, Sonia Reyes-Paecke, and Ellen Banzhaf. 2016. “Indicators for Green Spaces in Contrasting Urban Settings.” Ecological Indicators 62: 212–19. Lakes, Tobia, Maria Brückner, and Alexander Krämer. 2014. “Development of an Environmental Justice Index to Determine Socio-Economic Disparities of Noise Pollution and Green Space in Residential Areas in Berlin.” Journal of Environmental Planning and Management 57(4): 538–56. http://www.tandfonline.com/doi/abs/10.1080/09640568.2012.755461. Latour, Bruno. 2005. Reassembling the Social: An Introduction to Actor-Network-Theory. Oxford university press. Lepczyk, Christopher A. et al. 2017. “Biodiversity in the City: Fundamental Questions for Understanding the Ecology of Urban Green Spaces for Biodiversity Conservation.” BioScience 27: 41–50. http://academic.oup.com/bioscience/article/doi/10.1093/biosci/bix079/4056044/Biodiversity82  -in-the-City-Fundamental-Questions-for (August 10, 2017). Li, Mengmeng, Wietske Bijker, and Alfred Stein. 2015. “Use of Binary Partition Tree and Energy Minimization for Object-Based Classification of Urban Land Cover.” ISPRS Journal of Photogrammetry and Remote Sensing 102: 48–61. http://www.sciencedirect.com/science/article/pii/S0924271615000167 (November 9, 2015). Li, Xiaojiang et al. 2015. “Assessing Street-Level Urban Greenery Using Google Street View and a Modified Green View Index.” Urban Forestry and Urban Greening 14(3): 675–85. http://dx.doi.org/10.1016/j.ufug.2015.06.006. Lin, Jinyao, Xiaoping Liu, Kai Li, and Xia Li. 2014. “A Maximum Entropy Method to Extract Urban Land by Combining MODIS Reflectance, MODIS NDVI, and DMSP-OLS Data.” International Journal of Remote Sensing 35(18): 6708–27. Lizarazo, Ivan, and Joana Barros. 2010. “Fuzzy Image Segmentation for Urban Land-Cover Classification.” Photogrammetric Engineering & Remote Sensing 76(2): 151–62. http://essential.metapress.com/openurl.asp?genre=article&id=doi:10.14358/PERS.76.2.151. Long, Dan S. 1998. “Spatial Autoregression Modeling of Site-Specific Wheat Yield.” Geoderma 85(2–3): 181–97. Low, S. 2013. “Public Space and Diversity: Distributive, Procedural and Interactional Justice for Parks.” In The Ashgate Research Companion to Planning and Culture, eds. G. Young and D. Stevenson. Surrey: Ashgate Publishing, 295–310. Lu, Yuhao, Nicholas C. Coops, and Txomin Hermosilla. 2016. “Regional Assessment of Pan-Pacific Urban Environments over 25 Years Using Annual Gap Free Landsat Data.” International Journal of Applied Earth Observation and Geoinformation 50: 198–210. http://linkinghub.elsevier.com/retrieve/pii/S0303243416300447 (July 26, 2017). ———. 2017. “Estimating Urban Vegetation Fraction across 25 Cities in Pan-Pacific Using Landsat Time Series Data.” ISPRS Journal of Photogrammetry and Remote Sensing 126: 11–23. http://linkinghub.elsevier.com/retrieve/pii/S092427161630449X (July 26, 2017). Luederitz, Christopher et al. 2015. “A Review of Urban Ecosystem Services: Six Key Challenges for Future Research.” Ecosystem Services 14: 98–112. http://www.sciencedirect.com/science/article/pii/S2212041615300024 (May 27, 2015). MacFaden, Sean W. et al. 2012. “High-Resolution Tree Canopy Mapping for New York City 83  Using LIDAR and Object-Based Image Analysis.” Journal of Applied Remote Sensing 6(1): 63567-1-063567–23. http://remotesensing.spiedigitallibrary.org/article.aspx?doi=10.1117/1.JRS.6.063567 (July 18, 2016). Massey, Doreen. 1991. Space, Place and Gender A Global Sense of Place. Minneapolis, Minnesota: University of Minnesota Press. http://www.aughty.org/pdf/global_sense_place.pdf (October 29, 2014). McDonald, Robert I, Richard T T Forman, and Peter Kareiva. 2010. “Open Space Loss and Land Inequality in United States’ Cities, 1990-2000.” PloS one 5(3): e9509. http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0009509 (November 2, 2015). McGaughey, Robert J. 2015. “FUSION.” http://forsys.cfr.washington.edu/fusion/fusionlatest.html. McGrath, Brian, and S. T. A. Pickett. 2011. “The Metacity: A Conceptual Framework for Integrating Ecology and Urban Design.” Challenges 2(4): 55–72. http://www.mdpi.com/2078-1547/2/4/55/htm (October 16, 2015). McKinney, Michael L. 2006. “Urbanization as a Major Cause of Biotic Homogenization.” Biological Conservation 127(3): 247–60. McPhearson, Timon, Erik Andersson, Thomas Elmqvist, and Niki Frantzeskaki. 2015. “Resilience of and through Urban Ecosystem Services.” Ecosystem Services 12: 152–56. Meerow, Sara, Joshua P. Newell, and Melissa Stults. 2016. “Defining Urban Resilience: A Review.” Landscape and Urban Planning 147. Metro Vancouver. 2013. Metro Vancouver 2040: Shaping Our Future. Vancouver. http://www.metrovancouver.org/services/regional-planning/PlanningPublications/RGSAdoptedbyGVRDBoardJuly292011.pdf. ———. 2014. Progress Toward Shaping Our Future: 2014 Annual Report. Vancouver. http://www.metrovancouver.org/services/regional-planning/PlanningPublications/MV_2040_Progress_toward_Shaping_our_Future_2014_Annual_Report.pdf. ———. 2016a. “About US.” http://www.metrovancouver.org/about/Pages/default.aspx. 84  ———. 2016b. “Personal Communication.” Metro Vancouver Regional District. 2011. Ecological Health Action Plan. Metternicht, Graciela, Lorenz Hurni, and Radu Gogu. 2005. “Remote Sensing of Landslides: An Analysis of the Potential Contribution to Geo-Spatial Systems for Hazard Assessment in Mountainous Environments.” Remote Sensing of Environment 98(2–3): 284–303. Mohai, Paul, David N Pellow, and J. Timmons Roberts. 2009. “Environmental Justice.” Annu. Rev. Environ. Resour (34): 405–30. 10.1146/annurev-environ-082508-094348. Momeni, Rahman, Paul Aplin, and Doreen Boyd. 2016. “Mapping Complex Urban Land Cover from Spaceborne Imagery: The Influence of Spatial Resolution, Spectral Band Set and Classification Approach.” Remote Sensing 8(2): 88. http://www.mdpi.com/2072-4292/8/2/88. Moore, Jennie, Meidad Kissinger, and William E. Rees. 2013. “An Urban Metabolism and Ecological Footprint Assessment of Metro Vancouver.” Journal of Environmental Management 124: 51–61. http://linkinghub.elsevier.com/retrieve/pii/S0301479713001618 (August 23, 2017). Morse, Nathaniel B. et al. 2014. “Novel Ecosystems in the Anthropocene: A Revision of the Novel Ecosystem Concept for Pragmatic Applications.” Ecology and Society 19(2). Myint, Soe W. et al. 2011. “Per-Pixel vs. Object-Based Classification of Urban Land Cover Extraction Using High Spatial Resolution Imagery.” Remote Sensing of Environment 115(5): 1145–61. http://www.sciencedirect.com/science/article/pii/S0034425711000034 (May 19, 2015). Nichol, J., and C. M. Lee. 2005. “Urban Vegetation Monitoring in Hong Kong Using High Resolution Multispectral Images.” International Journal of Remote Sensing 26(0): 903–18. Nijland, W., N.C. Coops, S.E. Nielsen, and G. Stenhouse. 2015. “Integrating Optical Satellite Data and Airborne Laser Scanning in Habitat Classification for Wildlife Management.” International Journal of Applied Earth Observation and Geoinformation 38: 242–50. http://linkinghub.elsevier.com/retrieve/pii/S0303243414002670. Nozick, Robert. 1993. Anarchy, State, and Utopia. Oxford: Blackwell Publishers. Nussbaum, Martha, and Amartya Sen, eds. 1993. The Quality of Life. Oxford, UK: Oxford University Press. 85  O’Neil-Dunne, Jarlath, Sean MacFaden, and Anna Royar. 2014. “A Versatile, Production-Oriented Approach to High-Resolution Tree-Canopy Mapping in Urban and Suburban Landscapes Using GEOBIA and Data Fusion.” Remote Sensing 6(12): 12837–65. http://www.mdpi.com/2072-4292/6/12/12837/ (July 15, 2016). O’Neil-Dunne, Jarlath P.M., Sean W. MacFaden, Anna R. Royar, and Keith C. Pelletier. 2013. “An Object-Based System for LiDAR Data Fusion and Feature Extraction.” Geocarto International 28(3): 227–42. Pace, R.Kelley, and Ronald Barry. 1997. “Sparse Spatial Autoregressions.” Statistics and Probability Letters 33: 291–97. Pascual-Hortal, Lucía, and Santiago Saura. 2006. “Comparison and Development of New Graph-Based Landscape Connectivity Indices: Towards the Priorization of Habitat Patches and Corridors for Conservation.” Landscape Ecology 21(7): 959–67. http://link.springer.com/10.1007/s10980-006-0013-z (April 7, 2017). Peng, Jian et al. 2015. “Sustainability Evaluation of Natural Capital Utilization Based on 3DEF Model: A Case Study in Beijing City, China.” Ecological Indicators 58: 254–66. https://apps.webofknowledge.com/full_record.do?product=UA&search_mode=AdvancedSearch&qid=15&SID=2Bw5iVlCLEa8hfJJc65&page=11&doc=107 (July 11, 2015). Pickett, S. T. A. et al. 2016. “Dynamic Heterogeneity: A Framework to Promote Ecological Integration and Hypothesis Generation in Urban Systems.” Urban Ecosystems. http://link.springer.com/10.1007/s11252-016-0574-9. Pickett, S.T.A. et al. 2013. “Ecological Science and Transformation to the Sustainable City.” Cities 32: S10–20. http://www.sciencedirect.com/science/article/pii/S0264275113000267 (August 19, 2015). Pickett, Steward T. A. et al. 2016. “Evolution and Future of Urban Ecological Science: Ecology in , of , and for the City.” Ecosystem Health and Sustainability 2(7): 1–16. http://doi.wiley.com/10.1002/ehs2.1229 (July 26, 2016). Piketty, T. 2014. Capital in the Twenty-First Century. Cambridge, Massechusettes: The Belknap Press of Harvard University Press. http://books.google.com/books?hl=en&lr=&id=T8zuAgAAQBAJ&oi=fnd&pg=PR5&dq=Capital+in+the+Twenty-86  First+Century&ots=ZiLGJeYDGi&sig=_qKfJpvQn1oCF3xIfpwcJg8DKzA (December 14, 2014). Plowright, Andrew A., Nicholas C. Coops, and Neal W. Aven. 2015. “Evaluating the Health of Urban Forests Using Airborne LiDAR.” In 2015 Joint Urban Remote Sensing Event (JURSE), IEEE, 1–4. http://ieeexplore.ieee.org/document/7120461/ (July 31, 2017). Pu, Ruiliang, and Shawn Landry. 2012. “A Comparative Analysis of High Spatial Resolution IKONOS and WorldView-2 Imagery for Mapping Urban Tree Species.” Remote Sensing of Environment 124: 516–33. http://dx.doi.org/10.1016/j.rse.2012.06.011. R Development Core Team, R. 2011. “R: A Language and Environment for Statistical Computing” ed. R Development Core Team. R Foundation for Statistical Computing 1(2.11.1): 409. http://www.r-project.org. Radoux, Julien, and Patrick Bogaert. 2017. “Good Practices for Object-Based Accuracy Assessment.” Remote Sensing 9(7): 646. http://www.mdpi.com/2072-4292/9/7/646 (June 22, 2017). RapidEye. 2015. Satellite Imagery Product Specifications V6.1. http://blackbridge.com/rapideye/upload/RE_Product_Specifications_ENG.pdf. Rashed, Tarek, and Carsten Jurgens, eds. 2010. 10 Remote Sensing of Urban and Suburban Areas. http://link.springer.com/10.1007/978-1-4020-4385-7. Ratti, Carlo et al. 2017. “Treepedia.” MIT Senseable City Lab. http://senseable.mit.edu/treepedia (October 5, 2017). Rawls, John. 1971. A Theory of Justice. Cambridge, Massachusetts: The Belknap Press of Harvard University Press. La Rosa, Daniele. 2014. “Accessibility to Greenspaces: GIS Based Indicators for Sustainable Planning in a Dense Urban Context.” Ecological Indicators 42: 122–34. http://dx.doi.org/10.1016/j.ecolind.2013.11.011. RStudio Team. 2016. “RStudio: Integrated Development for R. RStudio, Inc.” Rupprecht, Christoph D D, Jason A. Byrne, Jenni G. Garden, and Jean Marc Hero. 2015. “Informal Urban Green Space: A Trilingual Systematic Review of Its Role for Biodiversity and Trends in the Literature.” Urban Forestry and Urban Greening 14(4). Safransky, Sara. 2014. “Greening the Urban Frontier: Race, Property, and Resettlement in 87  Detroit.” Geoforum 56: 237–48. http://www.sciencedirect.com/science/article/pii/S0016718514001377 (September 17, 2015). Sasaki, Takeshi et al. 2011. “Object-Based Classification of Land Cover and Tree Species by Integrating Airborne LiDAR and High Spatial Resolution Imagery Data.” Landscape and Ecological Engineering 8(2): 157–71. http://link.springer.com/10.1007/s11355-011-0158-z (November 12, 2015). Saura, Santiago, and Begoña de la Fuente. 2011. “Connectivity as the Amount of Reachable Habitat in the Landscape: Conservation Priorities and the Roles of Habitat Patches in Landscape Networks.” In , 1–27. Schlosberg, David. 2007. Defining Environmental Justice: Theories, Movements, and Nature. Oxford University Press. Schneider, Annemarie. 2012. “Monitoring Land Cover Change in Urban and Peri-Urban Areas Using Dense Time Stacks of Landsat Satellite Data and a Data Mining Approach.” Remote Sensing of Environment 124: 689–704. Schöpfer, Elisabeth, Stefan Lang, and Josef Strobl. 2010. “Segmentation and Object-Based Image Analysis.” In Remote Sensing of Urban and Suburban Areas, eds. Tarek Rashed and Carsten Jurgens. Springer Science & Business Media. Schwarz, Kirsten et al. 2015. “Trees Grow on Money: Urban Tree Canopy Cover and Environmental Justice.” PLoS ONE 10(4): 1–17. Schwarz, Nina et al. 2017. “Understanding Biodiversity-Ecosystem Service Relationships in Urban Areas: A Comprehensive Literature Review.” Ecosystem Services 27: 161–71. http://linkinghub.elsevier.com/retrieve/pii/S2212041616305502 (September 23, 2017). Seiferling, Ian, Nikhil Naik, Carlo Ratti, and Raphäel Proulx. 2017. “Green Streets − Quantifying and Mapping Urban Trees with Street-Level Imagery and Computer Vision.” Landscape and Urban Planning 165: 93–101. http://www.sciencedirect.com/science/article/pii/S0169204617301147?via%3Dihub. Sen, Amartya. 2013. “The Ends and Means of Sustainability.” Journal of Human Development and Capabilities 14(1): 6–20. http://www.tandfonline.com/doi/abs/10.1080/19452829.2012.747492#.VInfTaaZOTU 88  (November 18, 2014). Shackleton, Sheona et al. 2015. “Multiple Benefits and Values of Trees in Urban Landscapes in Two Towns in Northern South Africa.” Landscape and Urban Planning 136: 76–86. https://apps.webofknowledge.com/full_record.do?product=UA&search_mode=AdvancedSearch&qid=15&SID=2Bw5iVlCLEa8hfJJc65&page=12&doc=112 (January 5, 2015). Shao, Zhenfeng, and Chong Liu. 2014. “The Integrated Use of DMSP-OLS Nighttime Light and MODIS Data for Monitoring Large-Scale Impervious Surface Dynamics: A Case Study in the Yangtze River Delta.” Remote Sensing 6(10): 9359–78. http://www.mdpi.com/2072-4292/6/10/9359/ (December 30, 2016). Singh, Kunwar K., John B. Vogler, Douglas A. Shoemaker, and Ross K. Meentemeyer. 2012. “LiDAR-Landsat Data Fusion for Large-Area Assessment of Urban Land Cover: Balancing Spatial Resolution, Data Volume and Mapping Accuracy.” ISPRS Journal of Photogrammetry and Remote Sensing 74: 110–21. http://www.sciencedirect.com/science/article/pii/S0924271612001785 (November 12, 2015). Small, Christopher. 2003. “High Spatial Resolution Spectral Mixture Analysis of Urban Reflectance.” Remote Sensing of Environment 88(1–2): 170–86. http://www.sciencedirect.com/science/article/pii/S0034425703002396 (October 15, 2015). Statistics Canada. 2017. “Population of Census Metropolitan Areas.” http://www.statcan.gc.ca/tables-tableaux/sum-som/l01/cst01/demo05a-eng.htm. Steenberg, James W N et al. 2015. “Neighbourhood-Scale Urban Forest Ecosystem Classification.” Journal of environmental management 163: 134–45. http://www.sciencedirect.com/science/article/pii/S0301479715302127 (September 1, 2015). Stefanski, Jan, Benjamin Mack, and Orn Waske. 2013. “Optimization of Object-Based Image Analysis With Random Forests for Land Cover Mapping.” IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing 6(6): 2492–2504. http://ieeexplore.ieee.org/lpdocs/epic03/wrapper.htm?arnumber=6494344 (June 14, 2016). Su, Jason G. et al. 2009. “An Index for Assessing Demographic Inequalities in Cumulative Environmental Hazards with Application to Los Angeles, California.” Environmental Science and Technology 43(20): 7626–34. 89  Suich, Helen, Caroline Howe, and Georgina Mace. 2015. “Ecosystem Services and Poverty Alleviation: A Review of the Empirical Links.” Ecosystem Services 12: 137–47. http://www.sciencedirect.com/science/article/pii/S2212041615000236 (May 7, 2015). Taylor, Lucy, and Dieter F. Hochuli. 2017. “Defining Greenspace: Multiple Uses across Multiple Disciplines.” Landscape and Urban Planning 158. TEEB. 2011. “TEEB Manual for Cities: Ecosystem Services in Urban Management.” The Economics of ecosystems and biodiversity (TEEB): 48. Thiele, Michael, Cody Anderson, and Andreas Brunn. 2014. “Cross-Calibration of the RapidEye Multispectral Imager Payloads Using near Simultaneous Acquisitions of Pseudo-Invariant Test Sites.” In eds. Roland Meynart, Steven P. Neeck, and Haruhisa Shimoda. International Society for Optics and Photonics, 924114. http://proceedings.spiedigitallibrary.org/proceeding.aspx?doi=10.1117/12.2066481 (December 30, 2016). Thompson, Shanley D., and Sarah E. Gergel. 2008. “Conservation Implications of Mapping Rare Ecosystems Using High Spatial Resolution Imagery: Recommendations for Heterogeneous and Fragmented Landscapes.” Landscape Ecology 23(9): 1023–37. Threlfall, Caragh G., and Dave Kendal. 2017. “The Distinct Ecological and Social Roles That Wild Spaces Play in Urban Ecosystems.” Urban Forestry & Urban Greening. http://www.sciencedirect.com/science/article/pii/S1618866716304721 (July 10, 2017). Tigges, Jan, Tobia Lakes, and Patrick Hostert. 2013. “Urban Vegetation Classification: Benefits of Multitemporal RapidEye Satellite Data.” Remote Sensing of Environment 136: 66–75. http://www.sciencedirect.com/science/article/pii/S0034425713001429. Tompalski, P., and P. Wężyk. 2012. “Lidar and VHRS Data for Assessing Living Quality in Cities - An Approach Based on 3d Spatial Indices.” ISPRS - International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences XXXIX-B6(September): 173–76. Tooke, Thoreau R, Brian Klinkenberg, and Nicholas C Coops. 2010. “A Geographical Approach to Identifying Vegetation-Related Environmental Equity in Canadian Cities.” Environment and Planning B: Planning and Design 37(6): 1040–56. http://epb.sagepub.com/content/37/6/1040.abstract (September 18, 2015). 90  Trimble. 2012a. “eCognition® Developer 8.8 Reference Book.” http://www.ecognition.com/. ———. 2012b. “eCognition® Developer 8.8 User Guide.” Trimble eCognition. 2010. Guide Tour L3 - Advanced Example of Building Extraction. UN. 2008. “Economic Social World Urbanization Prospects The 2007 Revision Highlights.” United Nations (February). UN Habitat. 2013. State of the World’s Cities 2012/2013: Prosperity of Cities. New York. www.unhabitat.org. United Nations. 2014. “World’s Population Increasingly Urban with More than Half Living in Urban Areas.” http://www.un.org/en/development/desa/news/population/world-urbanization-prospects-2014.html (April 28, 2015). United Nations Department of Economic and Social Affairs Population Division. 2014. World Urbanization Prospects: The 2014 Revision: Highlights. http://esa.un.org/unpd/wup/Highlights/WUP2014-Highlights.pdf. Unnikrishnan, Hita, and Harini Nagendra. 2014. “Privatizing the Commons: Impact on Ecosystem Services in Bangalore’s Lakes.” Urban Ecosystems 18(2): 613–32. https://apps.webofknowledge.com/full_record.do?product=UA&search_mode=AdvancedSearch&qid=15&SID=2Bw5iVlCLEa8hfJJc65&page=12&doc=111 (November 2, 2015). Villa, F., Bagstad, K. J., Voigt, B., Johnson, G. W., Athanasiadis, I. N., & Balbi, S. 2014. “The Misconception of Ecosystem Disservices: How a Catchy Term May Yield the Wrong Messages for Science and Society.” Ecosystem Services 10: 52–53. Villa, Ferdinando et al. 2014. “A Methodology for Adaptable and Robust Ecosystem Services Assessment.” PloS one 9(3): e91001. http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0091001 (July 31, 2015). Voltersen, Michael, Christian Berger, Sören Hese, and Christiane Schmullius. 2014. “Object-Based Land Cover Mapping and Comprehensive Feature Calculation for an Automated Derivation of Urban Structure Types at Block Level.” Remote Sensing of Environment 154: 192–201. http://www.sciencedirect.com/science/article/pii/S0034425714003216 (September 29, 2015). Ward Thompson, Catharine et al. 2012. “More Green Space Is Linked to Less Stress in Deprived Communities: Evidence from Salivary Cortisol Patterns.” Landscape and Urban Planning 91  105(3): 221–29. http://www.sciencedirect.com/science/article/pii/S0169204611003665 (April 7, 2017). Welch, R. 1982. “Spatial Resolution Requirements for Urban Studies.” International Journal of Remote Sensing 3(2): 139–46. http://www.tandfonline.com/doi/abs/10.1080/01431168208948387 (December 30, 2016). Weng, Qihao. 2014. Global Urban Monitoring and Assessment through Earth Observation. CRC Press. Wentz, Elizabeth et al. 2014. “Supporting Global Environmental Change Research: A Review of Trends and Knowledge Gaps in Urban Remote Sensing.” Remote Sensing 6(5): 3879–3905. http://www.mdpi.com/2072-4292/6/5/3879/htm (August 27, 2015). WHO Regional Office for Europe. 2016. Urban Green Spaces and Health: A Review of the Evidence. Copenhagen. http://www.euro.who.int/__data/assets/pdf_file/0005/321971/Urban-green-spaces-and-health-review-evidence.pdf?ua=1. Wilkinson, Cathy, Toomas Saarne, Garry D Peterson, and Johan Colding. 2013. “Strategic Spatial Planning and the Ecosystem Services Concept-an Historical Exploration.” Ecology and Society 18(1). http://search.ebscohost.com/login.aspx?direct=true&profile=ehost&scope=site&authtype=crawler&jrnl=17083087&AN=86991050&h=oVPbbMPLmDSCj1tZplarlDGabZcgVhVwN+c1hg+ivk0qI+Zh+0oyIQEzd02fKN6WaLlWHFl/K/LetdB3JN2jlQ==&crl=c. Wolch, Jennifer R., Jason Byrne, and Joshua P. Newell. 2014a. “Urban Green Space, Public Health, and Environmental Justice: The Challenge of Making Cities ‘just Green Enough.’” Landscape and Urban Planning 125: 234–44. http://linkinghub.elsevier.com/retrieve/pii/S0169204614000310 (August 25, 2017). ———. 2014b. “Urban Green Space, Public Health, and Environmental Justice: The Challenge of Making Cities ‘just Green Enough.’” Landscape and Urban Planning 125: 234–44. http://www.sciencedirect.com/science/article/pii/S0169204614000310 (July 10, 2014). Wolff, Robert Paul. 1977. Understanding Rawls: A Reconstruction and Critique of A Theory of Justice. Princeton, New Jersey: Princeton University Press. Wu, Jianguo. 2014. “Urban Ecology and Sustainability: The State-of-the-Science and Future 92  Directions.” Landscape and Urban Planning 125: 209–21. http://www.sciencedirect.com/science/article/pii/S0169204614000322 (July 14, 2014). Yao, Liang et al. 2014. “Effective Green equivalent—A Measure of Public Green Spaces for Cities.” Ecological Indicators 47: 123–27. http://www.sciencedirect.com/science/article/pii/S1470160X14003161 (August 19, 2015). Yin, Haiwei, Yongjun Song, Fanhua Kong, and Yi Qi. 2007. “Measuring Spatial Accessibility of Urban Parks: A Case Study of Qingdao City, China.” In Geoinformatics 2007, International Society for Optics and Photonics, 67531L–67531L. https://apps.webofknowledge.com/full_record.do?product=UA&search_mode=AdvancedSearch&qid=15&SID=2Bw5iVlCLEa8hfJJc65&page=13&doc=127 (November 2, 2015). Young, Robert F. 2010. “Managing Municipal Green Space for Ecosystem Services.” Urban Forestry & Urban Greening 9(4): 313–21. http://www.sciencedirect.com/science/article/pii/S1618866710000543 (August 6, 2015). Zhan, Qingming, Martien Molenaar, and Ben Gorte. 2000. “Urban Land Use Classes With Fuzzy Membership and Classification Based on Integration of Remote Sensing and Gis.” International Archives of Photogrammetry and Remote Sensing XXXIII(art B7): 1751–60. http://www.isprs.org/proceedings/XXXIII/congress/part7/1751_XXXIII-part7.pdf. Zhan, Qingming, Martien Molenaar, and Klaus Tempfli. 2002. “Hierarchical Image Object-Based Structural Analysis toward Urban Land Use Classification Using High-Resolution Imagery and Airborne LiDAR Data.” Remote Sensing and Data Fusion over Urban Areas, IEEE/ISPRS Joint Workshop 2001 (November 2015): 251–58. itc.eu. Zhou, Weiqi. 2013. “An Object-Based Approach for Urban Land Cover Classification: Integrating LiDAR Height and Intensity Data.” IEEE Geoscience and Remote Sensing Letters 10(4): 928–31. Zhou, Weiqi, Ganlin Huang, Austin Troy, and M.L. Cadenasso. 2009. “Object-Based Land Cover Classification of Shaded Areas in High Spatial Resolution Imagery of Urban Areas: A Comparison Study.” Remote Sensing of Environment 113(8): 1769–77. http://www.sciencedirect.com/science/article/pii/S0034425709001199 (November 4, 2015). Zhou, Weiqi, Austin Troy, and Morgan Grove. 2008. “Object-Based Land Cover Classification and Change Analysis in the Baltimore Metropolitan Area Using Multitemporal High 93  Resolution Remote Sensing Data.” Sensors 8(3): 1613–36. http://www.mdpi.com/1424-8220/8/3/1613/htm (January 12, 2016). 94  Appendices Appendix A   ES Recoding Scheme Based On Wilkinson Et Al. (2013)1 ES Type ES Code Supporting A Water cycling (groundwater recharge) A1 Soil formation A2 Nutrient cycling C cycle A3a Nutrient cycling-nitrogen A3b Nutrient cycling-sulphur A3c Nutrient cycling-phosphorous A3d Primary production(green space) A4 Photosynthesis A5 Biodiversity A6 Habitat A7 Supporting subtotal  Provisioning B Food-Agriculture B1 Food-Fishing B2 Food-Wild B3 Fresh water B4 Water-energy B4b Water-transportation B4c Biochemicals/genetic resource B5 Fiber/raw materials B6 Biological energy sources/Fuel B7 Medicine B8 Provisioning subtotal  Regulating C Climate regulation – local (UHI/temperature) C1a Climate regulation – global (carbon sequestration) C1b Air quality regulation C2 Water purification/waste treatment C3 Water regulation (infiltration, stormwater regulation) C4 Wave attenuation/Coastal protection C4a                                                  1 Bold entries have been added to the original Wilkinson et al. scheme, found in Wilkinson, C., Saarne, T., Peterson, G. D., & Colding, J. (2013). Strategic Spatial Planning and the Ecosystem Services Concept-an Historical Exploration. Ecology and Society, 18(1). http://doi.org/10.5751/ES-05368-180137  95  Disease regulation C5 Pest regulation C6 Natural hazard regulation (floods, storm surge, etc.) C7 Erosion regulation - soil retention C8 Pollination C9 Seed dispersal C10 Noise regulation/light pollution reduction C11 Street cleaning C12 Wind regulation C13 Gas regulation C14 Fire regulation C15 Energy conservation C16 Regulating subtotal  Cultural D Social relations D1 Cultural landscape, heritage values D2 Sense of place/Connection to nature D3 Aesthetic D4 Inspirational D5 Recreation and eco-tourism D6 Educational & knowledge D7 Physical health D8 Psychological health D9 Spiritual and religious values D10 Green commons D11 Community organization/empowerment D12 Economic multipliers(retail sales, property values) D13 Safety D14 Cultural subtotal     96  Appendix B  Chapter 2 Annotated Bibliography 1. Anguelovski, I., & Martínez Alier, J. (2014). The “Environmentalism of the Poor” revisited: Territory and place in disconnected glocal struggles. Ecological Economics, 102, 167–176. http://doi.org/10.1016/j.ecolecon.2014.04.005  Anguelovski and Martínez Alier present four currents of environmentalism: Cult of Wilderness, Gospel of Eco-Efficiency, Mantra of Environmental Justice, and “Environmental Justice and the Environmentalism of the Poor.” They focus on the latter, first reviewing urban environmental struggles before arguing for an environmentalism of the poor grounded in social justice and place-based struggles. The authors note that nature itself is a political construct, and that cities are sites of strong power differentials and acute environmental struggles, especially with regard to gentrification and displacement. Highlighting different environmentalist approaches, the authors compare the commonplace marriage of Eco-efficiency and Conservation initiatives with the relative rarity of collaboration between conservation and environmental justice groups. The authors imply that conservation groups' acquiescence to global capitalism is allowing for its expansion to, perhaps, the detriment of nature and environmental justice. A general argument in favour of environmental justice (EJ) is presented, including environmental justice outcomes that consider race, gender, and poverty. The article is global in its spatial scope, though focuses on urban environmental justice issues in the Global South especially, such as the vulnerability of informal settlements (slums, favelas, etc.) to natural disasters and disease. Also highlighted is the often unequal distribution of environmental services in urban areas such as access to clean transit, infrastructure upgrades, and access to good quality (healthy/fresh/local/affordable) food. The authors also note how Indigenous worldviews differ from non-indigenous worldviews, leading to conflicting conceptions of land ownership, use, and participatory frames. They also criticise traditional GDP accounting, comparing it with alternatives like Buen Vivir and "GDP of the Poor".   2. Apostolopoulou, E., Bormpoudakis, D., Paloniemi, R., Cent, J., Grodzińska-Jurczak, M., Pietrzyk-Kaszyńska, A., & Pantis, J. D. (2014). Governance rescaling and the neoliberalization of nature: the case of biodiversity conservation in four EU 97  countries. International Journal of Sustainable Development & World Ecology, 21(6), 481–494. http://doi.org/10.1080/13504509.2014.979904  Apostolopoulou et al. examine how the rescaling of biodiversity governance and rescaling of political economy of conservation in the European Union is linked with a neoliberalizing trend in nature conservation. They also look at who might benefit and who might be harmed by this governance rescaling and neoliberalization. Their results show that EU states have aligned conservation and capitalist interests under a neoliberal framework. Despite this homogenization, the political past and the socio-political idiosyncrasies of the countries studied influenced the types and extent of governance rescaling and the changing relationship between conservation and capitalism. The authors’ work shows that private, non-state actors (businesses and NGOs) have benefitted from the neoliberalization of nature conservation, sometimes to the detriment of conservation areas themselves – especially if they have profit-oriented budgets and management – and sometimes to the detriment of effective, accountable governance or oversight. This can result in, for example, a lack of enforcement of environmental regulations on firms operating in or near protected areas. Despite potentially deleterious outcomes, some government policies have made participation by the private sector in conservation governance mandatory. The authors also highlight how the neoliberalization of biodiversity governance has led to its commodification (through e.g. natural capital and ES frames) with the implication that it is then susceptible to injustices inherent in markets.  Rescaling of biodiversity governance has also involved deregulation, sometimes jeopardizing conservation areas because of encroaching development or lack of state-provided resources.  Devolution is billed as a democratic reform, but financing for conservation (or other) projects can be depoliticized through market competition which can then enable elite capture. Neoliberal policies often employ austerity, which can lead to the government pulling out of its central role of conservation governance and letting market mechanisms fill its former role. This can lead to elite capture (because without state involvement elites have the resources to engage in governance), negative externalities (pollution, degradation), and less democratic/egalitarian, 98  more authoritarian conservation governance (like exclusion from governance of or access to the land).  3. Baptista, S. R. (2008). Metropolitanization and Forest Recovery in Southern Brazil: A Multiscale Analysis of the Florianópolis City-Region, Santa Catarina State, 1970 to 2005. Ecology and Society, 13(2), 5. http://doi.org/5  Baptista asks questions about how land change can be studied in urban regions with respect to socio-economic, demographic and ecological transitions, especially with regard to forests. The purpose of the paper was to encourage interdisciplinary dialogue, and did not present new research or frameworks for justice. Social and environmental justice is mentioned in the abstract just slightly less than in the rest of the paper, with a focus on how urban growth might impact people and ecosystems now and in the future.    4. Baptista, S. R. (2010). Metropolitan land-change science: A framework for research on tropical and subtropical forest recovery in city-regions. Land Use Policy, 27(2), 139–147. http://doi.org/10.1016/j.landusepol.2008.12.009  In this paper Baptista looked at land cover change in a southern Brazilian metropolitan region based on population and agricultural census data. Results showed that the area of agricultural lands declined, as did their populations, while forested areas grew over the time frame investigated. Justice issues were mentioned, though not thoroughly addressed. These include questions of how urban expansion and status quo environmental management might lead to uneven development and environmental injustice: Who has or does not have access to environmental amenities? How are environmental disamenities and their impacts distributed? How are ecosystem benefits distributed? The author also notes how "... surprisingly little has been published relating forest transition dynamics to spatial configurations of socioeconomic inequality and poverty.”   99  5. Baustian, M. M., Mavrommati, G., Dreelin, E. A., Esselman, P., Schultze, S. R., Qian, L., … Rose, J. B. (2014). A one hundred year review of the socioeconomic and ecological systems of Lake St. Clair, North America. Journal of Great Lakes Research, 40(1), 15–26. http://doi.org/10.1016/j.jglr.2013.11.006  This paper examined the watershed of Lake St. Claire between Michigan and Ontario. Specifically, indicators of the watershed’s health as an ecological system were related to human influence from c. 1900 to the early 2000s. Justice issues were not investigated, though the study is an interesting quantitative investigation into how social-ecological systems change over time.  6. BenDor, T., & Stewart, A. (2011). Land use planning and social equity in North Carolina’s compensatory wetland and stream mitigation programs. Environmental Management, 47(2), 239–53. http://doi.org/10.1007/s00267-010-9594-z  BenDor and Stewart use the North Carolina Ecosystem Enhancement Program (EEP) as a case to evaluate the benefits and burdens generated from a program designed to mitigate the environmental impacts of urban growth. The EEP specifically focuses on mitigating impacts to aquatic ecosystems by allocating mitigation credits to developers and governments. The authors compared the location of wetland impact and mitigation sites with the socioeconomics of the populations surrounding those sites. Populations near impact sites were generally more urban, whiter, and more educated with higher incomes and lower poverty rates. Populations near mitigation sites were more often rural and people of colour, with higher poverty rates and lower incomes and/or wealth. Aquatic ecosystem losses and gains were therefore found to occur across a strong socio-economic gradient. The authors noted that the EEP might allow significant losses of streams and wetlands in urban areas, impairing “aquatic integrity” in those areas. Further, wetlands might be a disservice to rural populations, reducing property values because of lower use values of land. Justice issues of the scale of benefits and burdens were also highlighted, with aquatic ecosystems redistributed within jurisdictional boundaries that might not allow for representative relocation across a landscape. Furthermore, many ecosystem services (ES) flows from aquatic ecosystems are local, 100  “…including services with both direct and indirect ‘‘use value,’’ such as water purification, flood control, and shoreline stabilization." Displacement of aquatic ecosystems from urban areas, therefore, results in large numbers of urbanites losing aquatic ES and their counterparts gaining ES, while also shouldering the burden of concomitant disservices.  A definitive test or metric of justice was not put forward, however the comparison of the location of wetland impact and mitigation sites with the socioeconomics of the populations surrounding those sites does allow for a quantitative and spatially explicit approach.   7. Bendt, P., Barthel, S., & Colding, J. (2013). Civic greening and environmental learning in public-access community gardens in Berlin. Landscape and Urban Planning, 109(1), 18–30. http://doi.org/10.1016/j.landurbplan.2012.10.003  Bendt, Barthel and Colding investigate the role that community gardens in Berlin that are open to the public and have low barriers have in facilitating ecological education and in helping people value nature. The authors “...analyse social learning and boundary activity—negotiation between external competences of individuals and socially defined competences—in public-access community gardens." Their results show that boundary activity in community gardens is impacted by three main features: “…socio-material resources, agency and activities, as well as [the gardens’] geographical locations." Justice issues were mostly related to how the institutions of community gardens affect participation within the garden, and how the garden is situated in a broader community. Participation enables access to green space and food ES (at a minimum) while also building community.  8. Bengston, D. N., & Youn, Y. (2006). Urban Containment Policies and the Protection of Natural Areas: The Case of Seoul’s Greenbelt. Ecology and Society, 11(1), 3. http://doi.org/10.1016/j.amepre.2008.01.018  Bengston and Youn look at the costs and benefits of the urban green belt, a form of urban containment policy, of Seoul, South Korea. Costs include increased land prices in the city surrounded by the green belt, decreased land prices and loss of property rights within the green 101  belt itself, and increased congestion and commuting times, especially for people commuting through the greenbelt into Seoul. Benefits include access to nature, green belt ES (especially recreation and air quality) and containing urban sprawl. Two equity concerns were highlighted as being the lack of compensation for loss of development rights of landholders in the greenbelt areas, as the negative impact on Seoul’s housing affordability.  9. Bigsby, K. M., McHale, M. R., & Hess, G. R. (2014). Urban Morphology Drives the Homogenization of Tree Cover in Baltimore, MD, and Raleigh, NC. Ecosystems, 17(2), 212–227. http://doi.org/10.1007/s10021-013-9718-4  Bigsby et al. study the drivers of tree cover in Baltimore, MD and Raleigh, NC in the United States. The cities are ecologically similar but have different development histories and urban morphologies. To understand how tree cover distribution varied between the two cities, the authors investigated the impacts of urban morphology, socioeconomics and lifestyle through time. The authors found that morphology (e.g. dense old towns versus sprawling suburbs) was the main driver of canopy cover. They also hypothesize that tree coverage distributions may converge in US cities if the trend of single-family residence parcels becoming smaller and more similar in size continues.  Results also showed that tree cover is positively correlated with affluence, an environmental justice concern. Not only can wealthier residents afford more extensive private green space, but there is also more public green space near their homes. The authors also noted how past socioeconomic distributions can affect contemporary tree cover, with some currently poorer neighbourhoods with relatively high canopy cover being inhabited by wealthier groups when the trees were planted.   10. Bodurow, C. C., Creech, C., Hoback, A., & Martin, J. (2009). Multivariable Value Densification Modeling Using GIS. Transactions in GIS, 13, 147–175. http://doi.org/10.1111/j.1467-9671.2009.01163.x  102  Bodurow et al. created a "community value densification tool" that allows, through participatory GIS techniques, for members of a community to identify resources, physical features, and their density. The authors apply their tool in a case study in SW Detroit, MI, USA to showcase its utility. The goal of the “Value Densification Community Mapping Project (VDCmp)” is to articulate a broad understanding of value in communities with which to advocate for equality and sustainability to decision-makers.  The VDCmp enables communities to visually assess the distribution of ES, environmental health impacts, social inequality and other patterns in a spatially explicit way “… with an end-goal of developing a dynamic, unified development and preservation strategy for the community." The VDCmp project supports an assets-based (versus needs-based) community planning approach - assets are based on three value criteria: human [inhabitation], cultural [place], and infrastructure [ecosystem]. Justice and equity issues were mentioned as a rationale for utilizing the tool, but were not thoroughly investigated on their own.   11. Booth, J. E., Gaston, K. J., & Armsworth, P. R. (2010). Who Benefits from Recreational Use of Protected Areas? Ecology and Society, 15(3), 19. http://www.ecologyandsociety.org/vol15/iss3/art19/  This study investigated the composition of user groups of recreation sites in England in terms of gender, age, socioeconomic level, and ethnicity. It explicitly looked at recreation as an ecosystem service of protected areas, and used recreation because of its "direct-use" nature. Justice is mostly assessed discursively, but also using statistical inference to compare characteristics of site users with local, regional and national populations to establish representativeness. Results showed significant bias among types of users, with fewer women, young people and ethnic minorities using recreation sites. Older, white males were more common site users. The authors note that a minority (about 1/3) of adults in the UK visit the countryside, and argue that it is imperative for recreation policy engage with more people, and more types of people, to enable their access to the benefits of recreation ES.   103  12. Brown, A., Dayal, A., & Rumbaitis Del Rio, C. (2012). From practice to theory: emerging lessons from Asia for building urban climate change resilience. Environment and Urbanization, 24(2), 531–556. http://doi.org/10.1177/0956247812456490  Brown, Dayal and Rumbaitis Del Rio present a paper on the Asian Cities Climate Change Resilience Network (ACCCRN), set up by the Rockefeller Foundation. Three goals for the network are capacity building, development of knowledge networks, expansion and scaling up of climate change mitigation/adaptation actions. The authors want to understand urban vulnerability to climate change, both in general, but also for vulnerable groups, sectors and geographies within cities. The ACCCRN work has highlighted 10 "urban climate change resilience action areas": climate sensitive land use and urban planning; institutional coordination mechanisms and capacity support; drainage, flood and solid waste management; water demand and conservation systems; emergency management and early warning systems; responsive health systems; resilient housing and transport systems; strengthening of ecosystem services; diversification and protection of climate-affected livelihood; and education and capacity building of citizens. A rigorous justice or equity test is not presented, but three questions are posed to structure thinking around climate-related inequalities: Are poor/vulnerable communities considered explicitly? Are there criteria for equity and inclusion? Are the livelihoods of poor communities considered? The authors note that the urban poor are more likely to have their livelihoods negatively impacted by climate change, as well as be more vulnerable to extreme weather events and natural disasters. Health-related absenteeism was also noted as a burden disproportionately borne by the urban poor. Policy responses to these challenges might include financial protection and insurance, economic/employment diversification and even relocation. Increasing the income of poor households can also help, but additionality becomes an issue when climate change is the focus. Some programs, like peri-urban agriculture projects, can be targeted to low income people, however. The authors argue that strong dialogue, mechanisms and transparency policies/actions are required so that what are billed as climate resilience policies are not actually Trojan horses, for e.g. development. Without equity considerations there could be 104  elite capture, with business elites arguing for the protection of business capital, diverting risk to poor communities.   13. Byrne, J., & Wolch, J. (2009). Nature, race, and parks: past research and future directions for geographic research. Progress in Human Geography, 33(6), 743–765. http://doi.org/10.1177/0309132509103156  Byrne and Wolch review geographic perspectives on park use from "… environmental justice, cultural landscape, and political ecology paradigms.” Their aim is to gain a better critical understanding of how parks and park use are shaped by and perpetuate “… historical, socio-ecological, and political-economic processes.” The paper looks at how ideology (class, race, nature) has shaped parks and park creation and the reasons for the well-documented ethno-racial differentiation of park use. They also review how power relations invested in and related to parks/nature spaces are being investigated by geographers. As a result of their review, the authors propose a conceptual model built on four elements: “(1) the socio-demographic characteristics of park users and non-users – as suggested by leisure research; (2) the political ecology and amenities of the park itself – eg [sic], landscape design, vegetation, and facilities, features of surrounding neighborhoods and land uses, management regime; (3) the historical and cultural landscapes of park provision – such as discriminatory land-use practices, philosophy of park design, or politics of development; and (4) individual perceptions of park spaces – eg [sic], accessibility, safety, conviviality, or sense of welcome, all mediated by personal characteristics, and the park’s political ecology, history, and cultural landscape.” Ecosystems services are explicitly mentioned in the review, but mostly rolled into parks and green spaces. The authors highlight many justice and equality issues including how the environmental, cultural and political context of parks may result in non-use and avoidance by people of colour and other vulnerable groups. In general, gender, class and race combine to negatively impact access to urban ES. Also reviewed are the four main reasons for differentiated park use according to leisure theorists: (1) Marginality theory, the theory that "…. people of color face socio-economic barriers that constrain when and how they visit and use parks (e.g. transit-dependent).” (2) Ethnicity theory - different subcultural styles/norms lead to different park use. 105  (3) Acculturation/Assimilation hypotheses asserting that differentiated use arises from unique ethno-cultural heritage "and/or because they have not adjusted to or adopted the dominant values of mainstream society." (4) Discrimination could also be a factor - real or perceived discrimination or hostility could deter people of colour from using parks. These hypotheses all have their faults, nicely summarized on p.750. Byrne and Wolch contend that the combination of all of these and more has led to spatially uneven development of park resources and access, often to the detriment of racialized/poor groups, negatively affecting their well-being. A justice assessment method or framework is not put forward, but the conceptual model presented does aid in understanding park use choices in the context of past and current justice issues. The authors also highlight areas where further research is needed, such as historical racism, gender, climate change and how they impact park use/access. The role that disadvantaged/people of colour play as agents in shaping new parks/access to nature also merits further inquiry.   14. Cen, X., Wu, C., Xing, X., Fang, M., Garang, Z., & Wu, Y. (2015). Coupling Intensive Land Use and Landscape Ecological Security for Urban Sustainability: An Integrated Socioeconomic Data and Spatial Metrics Analysis in Hangzhou City. Sustainability, 7(2), 1459–1482. http://doi.org/10.3390/su7021459  In this modelling paper using Hangzhou, China as a case, Cen et al. show that continual urban expansion can lead to a reduction in landscape level ES provisioning over time (after an initial increase). The overall goal of this study was to improve the understanding of the coupling relationship between intensive land use and landscape ecological security, which represent socioeconomic systems and ecological systems in urban areas, respectively. In addition, the paper aimed to contribute to fundamental knowledge for identifying potential ecological impacts and decision-making toward sustainable urbanization. Justice issues were not a focus of the paper, though the authors did note that trade-offs exist between development and ecological stability, especially in the developing world.   106  15. Chen, W. Y., & Hu, F. Z. Y. (2015). Producing nature for public: Land-based urbanization and provision of public green spaces in China. Applied Geography, 58, 32–40. http://doi.org/10.1016/j.apgeog.2015.01.007  Chen and Hu take an empirical look into how sales of urban land by local governments in China is affecting the provision of urban green space. Municipal land sales have financed a development boom in Chinese cities, generating significant wealth, however there is little data on whether that wealth creation is translating into the provision of public green spaces. A trend of appropriation of common/green spaces by developers and other private interests through the commodification of urban space is highlighted. Global capital flows can aid in this, especially in developing-world cities that may have fewer resources than, say, OECD municipalities. Land sales and development by governments finances much needed social infrastructure, however, the focus of these governments might be on expanding the built environment at the expense of green spaces, which are "chronically underprovided." Authors found that increasing reliance on land sales revenues was negatively correlated with the provision of urban public green space, though less so private green space. Government building regulations require 25-30% green space included in new residential developments, which helps to explain why total green space is not significantly correlated with a reliance on land finance. The negative relationship between land finance and public green spaces was most strongly negative in Eastern and Central China which are more developed, versus Western China, which is still developing. The authors note that residents in larger cities like Shanghai and Huangzhou lack access to parks, and in some cities parks are moved to the urban fringe where access, especially for poorer people, can be difficult. Also highlighted is the dual nature of land finance – generating municipal revenue that could provide green space on the one hand, while incentivizing urban expansion and intensification on the other.  No explicit test of justice or equity is presented. However, the authors do present a quantitative analysis that implies injustices (reduced access to urban public green spaces) are increasing with increased reliance on revenues from land sales in the eastern Chinese cities.  107  16. Cilliers, S., Cilliers, J., Lubbe, R., & Siebert, S. (2012). Ecosystem services of urban green spaces in African countries—perspectives and challenges. Urban Ecosystems, 16(4), 681–702. http://doi.org/10.1007/s11252-012-0254-3  Cilliers et al. pen a hybrid between a literature review and a presentation of related original research. The focus is on urban ecosystem services studies in Africa, with a heavy emphasis on South African studies. In their original research, the authors compare gardens along a socio-economic gradient in Potchefstroom. Results showed that poorer people generally had more useful (provisioning ES) and fewer ornamental plants than higher income people. It was also found that garden species composition changed along a socio-economic gradient (more species and more ornamentals with increasing wealth), but that property values were not related to proximity to green spaces. Fewer regulating services were provided by poorer areas, with smaller and fewer woody species present. Carbon storage levels were also lower in poorer areas, partially due to newer planted trees.  Multiple justice issues were highlighted by the authors, though many were mentioned in passing. In Potchefstroom wealthier areas had more green spaces and tree cover than poorer areas, largely due to the segregationist history of development. This echoed the findings of other researchers that found poorer areas in South Africa can have up to 14 times less green space. Also noted was that significant inequalities in green space provision and access exist even in developed countries. The authors highlighted the importance of considering the views of lower income people through effective stakeholder involvement in planning initiatives (related to which is that different groups value similar ES differently); that multiple ES valuation methods exist; and that ES can have an effect on other values, like property. A lack of African urban ES research was noted, with most African literature based on South African case studies. A more general lack of information on the link between biodiversity and ES for developing countries was also noted. Regarding provisioning ES and urban agriculture, its importance for lower income groups was acknowledged, however so was the fact that there is little evidence that poorer urban residents can engage in urban agriculture on a scale such that it actually supplements incomes.   108  17. Cilliers, S., du Toit, M., Cilliers, J., Drewes, E., & Retief, F. (2014). Sustainable urban landscapes: South African perspectives on transdisciplinary possibilities. Landscape and Urban Planning, 125, 260–270. http://doi.org/10.1016/j.landurbplan.2014.02.009  In this 2014 paper, Cilliers et al. synthesize urban ecology, urban planning and environmental management approaches to discuss how urban areas in South Africa can be managed sustainably. Urban ecology provides the theoretical foundations of how urban form relates to ecological function and can help outline the dynamics of social-ecological systems. Urban planning's key contribution is integrated urban planning as well as prediction and scenario building around land use. The authors note that a key to achieving sustainability in South African cities is to meet basic needs while also redressing historical injustices: “Legacies of spatial inequalities and the composition and quality of the local urban green infrastructure powerfully impact on urban biodiversity and the equal supply of [environmental goods and services] to its residents." Myriad threats to urban sustainability exist in South Africa, including “… water scarcity; safety and security issues; the Aids pandemic; increasing poverty; the tendency of the government to make uninformed decisions to meet deadlines and delivery quotas; the large body of new and complex policies, programmes and guidelines that creates confusion amongst different stakeholders, and the low efficiency in terms of service delivery from certain public service sectors.” The authors also note that wealthier cities and neighbourhoods are embracing a greener, ecosystem-based management, the “Green Agenda,” while poorer urban areas are focusing on a “Brown Agenda” that is concerned with freedom from environmental constraints rather than the enhancement of ES sources and flows. This duality is in part a legacy of colonialism and Apartheid, with significant and persistent segregation based on race and class in South African cities. A further justice issue is raised in a chronic lack of participatory democracy, and how environmental injustices tend to be perpetuated if ES beneficiaries are left out of governance proceedings.   18. Cohen, M., Baudoin, R., Palibrk, M., Persyn, N., & Rhein, C. (2012). Urban biodiversity and social inequalities in built-up cities: New evidences, next questions. 109  The example of Paris, France. Landscape and Urban Planning, 106(3), 277–287. http://doi.org/10.1016/j.landurbplan.2012.03.007  Cohen et al. compare the distribution of (semi) public green spaces with socioeconomic data in Paris, France to assess the equity of green space distribution. The authors also seek to answer the question of whether “public semi-natural spaces and green frames serve as democratizing elements for city-dwellers, whilst the gap between the mean incomes of households grouping suggests that Paris is an ‘unjust’ city?” Results showed that the current distribution of green spaces mostly benefit middle- and working-class Parisians, with their neighbourhoods clustered at the edge of Paris, proximal to the most green space. Wealthier residents of Paris generally lived in denser areas of the city with low-medium levels of ES provision from semi-public green space, though they did have greater access to waterways. The authors conceded that their study could not capture the complex determinants of urban inequality (see below). The focus on semi-public green space was especially limiting because many wealthier people in Paris have access to private spaces within and out of the city. They also tend to live in Haussmannian districts that have high built density.  By conducting a spatial statistical analysis of the relationship among urban landscapes, household socio-economic indicators and public semi-natural spaces (with ES values constructed through multi-criteria assessment), the authors were able to provide a quantitative assessment of the justice of ES distribution in Paris. Their results were inconclusive, however, so a strong assessment of the justice of ES distribution was not possible. The authors did acknowledge a number of justice issues, noting that distributional environmental justice issues are often exacerbated by, or occur in addition to, socio-economic inequality. Indeed, the income ratio between the mean incomes of the lowest ten percent and highest ten percent of Parisians is 178. A complex suite of factors determining urban inequalities are noted (“social, economic, cultural, political, spatial and environmental”), as are the common findings of many environmental justice studies that poorer or otherwise more vulnerable populations tend to bear more environmental costs and reap fewer burdens.    110  19. Colding, J., Barthel, S., Bendt, P., Snep, R., van der Knaap, W., & Ernstson, H. (2013). Urban green commons: Insights on urban common property systems. Global Environmental Change, 23(5), 1039–1051. http://doi.org/10.1016/j.gloenvcha.2013.05.006  Building on the work of Elinor Ostrom, Colding et al. investigate urban common property systems and how they can be a successful alternative basis for management compared to private or state property systems. They use cases of urban green-space management in Sweden, Germany and South Africa: community allotments in Stockholm with strong local land management rights; community gardens and public access community gardens in Berlin; and community green space creation in Cape Town. Each city, and in many cases each green space, have unique management rights and institutions that exclude some, include others, and help people organize politically and to make space. The Bottom Road experiment in Cape Town, for instance, led to the successful creation of other community green spaces, sometimes in the face of development pressure. The authors posit that participation in urban green commons may help to counteract alienation from, and therefore a lack of valuation of, nature. Also highlighted is how property rights diversity can lead to greater ecosystem management diversity with beneficial outcomes for biodiversity. The authors note other benefits of urban green commons including "… reducing costs for urban ecosystem management, [and] serving as designs for reconnecting people to the biosphere by offsetting extinction of experience in cities." Urban green commons may also play a part in protecting public land in cities, which may lead to more sustainable social-ecological systems.  Mechanisms for exclusion of urban green commons are often necessary to prevent congestion, therefore they might be best suited to private vacant lands or shared private lands, or public lands that are not adequately managed by governments due to funding issues – the authors concede that giving exclusion rights to a subset of the public on public lands is ethically problematic. Other, perhaps more significant, drivers of environmental inequities noted by the authors are the commodification of land, gentrification and sale of public lands. These drivers have put pressure on community green areas in Berlin to generate real estate market value or face displacement through land use change. Private (though majority state-owned) companies there, 111  like the Liegenschaftsfond, have through their management and marketing of public lands taken over local control of public land from local district authorities.  In Cape Town, land rights are very much still a function of a post-Apartheid reality, with past public or common rights to land non-existent for many South Africans. The community garden at Bottom Road was an explicit attempt to reduce barriers between community members and strengthen social cohesion in a context of previous official racial segregation, creating a common space while rehabilitating fynbos ecosystems.   20. Davis, A. Y., Belaire, J. A., Farfan, M. A., Milz, D., Sweeney, E. R., Loss, S. R., & Minor, E. S. (2012). Green infrastructure and bird diversity across an urban socioeconomic gradient. Retrieved from http://www.esajournals.org/doi/abs/10.1890/ES12-00126.1  This paper is an inquiry into the equitability of ecosystem services distribution in Chicago, IL, USA among socioeconomic groups. Census tracts were the main spatial unit for socioeconomic variables, and through statistical analysis 3 main clusters of census tracts were identified: low-middle income and mostly Hispanic; low-middle income and mostly African American; and middle-high income mixed race (but mostly white). The authors used 4 variables as proxies for ecosystem service abundance: proximity to Lake Michigan, proximity to open space, canopy cover and bird species abundance. Results showed that the mostly Hispanic census tracts had lower values for all 4 ES proxy variables, while mostly-black census tracts had lower median values than the higher incomes tracts for all four variables, but that these differences were not statistically significant. The authors concluded that there is an unequal distribution of ES in Chicago, which they explicitly labelled an environmental justice issue. However, they were also careful to note that the equitability of ES distribution was not assessed. If disadvantaged social groups required more access to ES due to differential need, then even where ES distribution was exactly equal it could still be inequitable. Davis et al. added another caveat to their results, arguing that to “…. determine whether the patterns we see here are equitable, inquiries about procedural justice, or the drivers of the observed patterns, are needed 112  since the location of parks, street trees and open space, as well as affordable housing, is a result of social and institutional forces that go beyond our investigation.”   The justice issues highlighted above were underpinned by the observation that "… green infrastructure and biodiversity are not always distributed evenly among socioeconomic groups.” The authors cited research showing that economically privileged groups generally have more access to ES, making this justice issue explicit by relating ES to quality of life. Reduced ES access may lead to an alienation from nature leading to lower quality of life and acceptance of a reduced baseline of ES leading to further environmental degradation. (Boone et al. 2009).” "... in Chicago, the patterns we report here for low-income African American neighborhoods could also be considered environmental injustice if they are the result of social or political injustices committed in the past."  21. de Oliveira, J. A. P., Doll, C. N. H., Balaban, O., Jiang, P., Dreyfus, M., Suwa, A., … Dirgahayani, P. (2013). Green economy and governance in cities: assessing good governance in key urban economic processes. Journal of Cleaner Production, 58, 138–152. http://doi.org/10.1016/j.jclepro.2013.07.043  De Oliveira et al. examine the urban governance challenges of achieving a ‘green economy.’ The authors identify four areas that can be reformed to align with a green economy: transformation of space; movement; consumption/production; natural, social and ecological services. The necessity of considering the exporting of environmental burdens and of creating incentives to internalize externalities is noted, as is the need for monitoring social-ecological indicators: "Good governance systems for green economy enables resource conservation, increased resilience of the socio-ecological system and improve human well-being, including increase in jobs and income for the poor." The authors also argue that two key components of urban sustainability are considering and managing for the needs and values of the more vulnerable members of society, as well as realizing the disproportionate benefits and costs the cities incur on the environment, both within cities and outside them.  A rigorous justice assessment or investigation is beyond the scope of this paper, however, the authors did present other environmental justice issues: the need for safe cycling and 113  pedestrian routes, lest the poor be excluded from urban life by long commutes; the need for green development to provide safe and affordable housing; affordable public transit; and the need to maintain urban ES which support the green economy, are relied on by the urban poor and are necessary for the reduction of their poverty levels.   22. Dobbs, C., Kendal, D., & Nitschke, C. R. (2014). Multiple ecosystem services and disservices of the urban forest establishing their connections with landscape structure and sociodemographics. Ecological Indicators, 43, 44–55. http://doi.org/10.1016/j.ecolind.2014.02.007  Dobbs, Kendal and Nitschke investigate how ecosystem services and disservices are linked with landscape structure and sociodemographics in the City of Melbourne. The authors used spatially explicit and transferable ES indicators to compare vegetation, ecosystem services and disservices as well as sociodemographics. The goal of the study was to present a quantitative ES meta-model applied through a social-ecological framework explicitly linking ES with human well-being. Results revealed that areas with higher population density and more socially vulnerable populations also had lower ES provision, while the Human Development Index was positively correlated with ES. The authors also found that some parks were locations of low ES clusters, while in and around others were high ES clusters. Though a definitive assessment of justice or equality was not provided, the authors noted that linking socio-demographics, landscape structure and ES allowed environmental inequalities to be related to social inequalities. Also of relevance to justice concerns was a hypothesis that wealthier and more educated people choose to live in areas that already have relatively high levels of ES provision, but also tend to demand more environmental resources which may exacerbate ES inequalities. Trade-offs between density and ES provision were also considered, and the authors warned that maintaining the social and ecological health of cities was imperative.   23. Dobbs, C., Nitschke, C. R., & Kendal, D. (2014). Global Drivers and Tradeoffs of Three Urban Vegetation Ecosystem Services. PLoS ONE, 9(11), e113000. http://doi.org/10.1371/journal.pone.0113000 114   A quantitative global investigation of the drivers of three ecosystem services (carbon storage, recreation potential and habitat potential) for 100 cities is presented by Dobbs, Nitschke and Kendal. The drivers, ES levels themselves and synergies and tradeoffs, as well as "their relation with development, climate and governance" were assessed. Multiple determinants of ES provision were noted, including “…environment conditions, socioeconomics, demographics and politics.” It was also found that “[p]opulation growth, consumption and governance” impacted ES provision, which could impact, in turn, multiple social factors such as “human health, livelihood, culture and equity.” Only about a quarter (26%) of the cities studies had per capita green cover levels meeting the World Health Organization recommendation of 9m2, and only 12% of cities had at least 20m2 of green space per person.   Different sizes of cities tended to provide different levels of ES, with ES provision negatively correlated with population. Developing country cities were also found to have fewer ES than developed country cities, leading the authors to hypothesize that disparities in urban ES provision have global significance. Specifically, “…the association between carbon storage, HDI, DI and temperature, suggests that carbon storage tends to increase in wealthy, educated and democratic cities from cooler climates." Similar results were found for recreation and habitat potential. Within cities urban elites tended to have greater access to ES than poor or other vulnerable groups, such as racialized populations. It was also noted that cities can cause and/or exacerbate already-existing social inequalities in a given society.  Avenues of future research suggested were explicitly justice focused, the authors arguing that they “… should include the addition of indicators that can represent cultural background and the legacy of historic development such as the effects of colonization, wars, ethnic diversity, industrialization, planning regulation and infrastructure development, among others."  24. Ernstson, H. (2013). The social production of ecosystem services: A framework for studying environmental justice and ecological complexity in urbanized landscapes. Landscape and Urban Planning, 109(1), 7–17. http://doi.org/10.1016/j.landurbplan.2012.10.005  115  In his 2013 paper, Ernstson provides one of the few theoretical frameworks published to date that specifically addresses environmental justice and ecosystem services. Based on this literature review and other sources (for example, the excellent urban ecosystem services reviews of  Hubacek and Kronenburg (2013), Luederitz et al. (2015) and Suich, Howe and Mace (2015)), it is a seminal work on the relationship between justice and ecosystem services. Fundamental to the framework is the idea that ES are not objective truths that exist outside socio-political value articulation processes, but that they are contested, socially produced artifacts. The conception of justice employed is made explicit, with Ernstson using a Rawlsian distributive justice definition. The framework is also rooted in ecological complexity and Actor Network Theory (ANT) which allows for the biophysical and social processes that underpin (often unequal) ES distribution patterns to be revealed.  The frame work has three analytical foci – ecosystem services generation, distribution and articulation – which are assessed in two main stages of analysis: the creation of a social-ecological network model, and the analysis of the ES value articulation process using methods informed by ANT. The first step is to use network analysis to identify locations (nodes) of ecosystem services and their connections to other locations. The nodes represent biophysical locations and the connections represent flows of ecological connectivity and functioning. Social aspects are also integrated into the network, with each node having a protective and management capacity associated with it. Each node is assessed for how competing value-articulation narratives create ES values. Analyzing the actor-networks that are engaged in values articulation at each node illuminates the structure of the value articulation process, which actors are involved and which have more or less power. The network and ANT (value-articulation) analyses can highlight where ES are generated, their distribution on the landscape, who has helped to create ES value, and who has access to ES, as well as how well ES values on the landscape can be managed and protected (an issue of competing narratives and power structures). Is should be noted, however, that the framework as deployed doesn’t have a definitive test or threshold for what is just or not. The framework allows for a more nuanced understanding of ES distribution and production, refuting stances that are “overly objectivist” as well as those that focus on “finding the right trade off.” The latter is “… often simplified into a consensual process, or a rational 116  choice game between actors with fixed interests (so called stakeholders) that can be steered/guided by economic incentives," a process which cannot represent reality if ES are produced through a contested, political value articulation process. Assigning protective and management capacity to network nodes has meaning for management too, allowing for the assessment of the nodes most vulnerable to ES loss or optimization through development or ecological enhancement actions. This can enable community groups, city managers and other political actors to engage in a conversation about resilience at the local and city-wide level, managing for the maintenance of benefits of social-ecological systems, for certain groups of people, accounting for dynamic system conditions.  Throughout the paper multiple justice conceptions and challenges are raised, including those that are left out the framework. Ernstson is explicit that his frameworks employs a "...Rawlsian notion of distributive justice [not]... other dimensions of environmental justice and urban political ecology, including the politics of participation and recognition, debates concerning power and knowledge; and class, race and systemic oppression." The paper also doesn't address "... the criticism that has been launched against the ecosystem services approach, for instance how it strengthens market-based and depoliticizing paradigms of decision-making..." Ernstson notes that strong economic incentives to compete for land use impacts the spatial distribution of environmental benefits and costs in the city and also tends to lead to landuse heterogeneity that can have ecological consequences.  There are also trade-offs between enhancing ES and enhancing the equality of ES distribution when, for instance, increasing forest connectivity would result in planting more trees in higher income areas. The framework, and ANT in particular, assumes the knowledge is a construct and therefore contested, which might lead to certain knowledges to be silenced. Understanding this, network actors can better understand how social-ecological system management can be biased towards certain ES in certain places. Indeed, the actors biasing ES provision and distribution could have enough power that they are able to maintain highly unjust social structures that are nonetheless resilient in a system-dynamics sense.  Resilience itself, then, is not necessarily an end in itself, a realization that may help to encourage “… creative and radical thinking on political actions to transform the system towards both increased ecological resilience and social justice." 117   25. Ernstson, H., & Sörlin, S. (2013). Ecosystem services as technology of globalization: On articulating values in urban nature. Ecological Economics, 86, 274–284. http://doi.org/10.1016/j.ecolecon.2012.09.012  Ernstson and Sörlin criticize ecosystem services as a conceptual force that undermines other (especially local) forms of place-creation by enforcing managerial, depoliticizing value frames. In their words, "[o]ne of the key points in this article is to demonstrate that when ecosystem services appear as objects of calculated value — guided by the ambition to attain influence in decision-making — they cannot be viewed as reflecting an objective biophysical reality, but should be understood and researched as a social practice to articulate value." The ESS approach is criticized as part of a globalized, universalist, new management that uses a language of economics and monetization. This type of management leads to the depoliticization of complex social issues, and social context and history being left out of ES valuation methods. Collective decision-making might then suffer at the hands of standardized, expert-determined criteria.  The authors’ solution "… is to provide space for critical ethnographies that traces how the ESS approach is enacted in-place, in various cities and locales, rather than yet another article that repackages the gesture of objectivity and universality in trying to come up with the ultimate ecosystem services framework."  An “ESS approach” that is not grounded in social value articulation has implications for distributional justice, social diversity and the equity of ES access. ES can’t be valued uniformly, and their value depends on numerous sociocultural factors such as “…location, income, livelihood, gender, culture.” A lack of self-scrutiny in the ESS approach can lead to it being privileged over other articulations of place and value, erasing those past efforts. Other articulations of value are not, for instance, included in the TEEB Manual for cities, “… which casts doubt on its usefulness; how could it be proven best practice, which is the claim it makes, if alternative value articulation is not evaluated?" Cities are, in part, defined by contested uses of space, and, the authors argue, an ES approach must consider that. Two cases studies, in Stockholm and Cape Town, are employed to show value articulation and place-making in action.  The Stockholm case showed that the values articulated by park activists eventually stabilized 118  into a new frame that explained the park landscape. A similar value articulation process leading to the stabilization of new place narratives occurred in Princess Vlei in Cape Town, a place with pre- and Apartheid era significance for people of colour. Princess Vlei was saved from development by a local community organization that successfully argued for its value in the face of significant pressure for property development.   26. Ernstson, H., Sörlin, S., & Elmqvist, T. (2009). Social Movements and Ecosystem Services — the Role of Social Network Structure in Protecting and Managing Urban Green Areas in Stockholm. Ecology and Society, 13(2), 39. http://doi.org/10.1002/pad  This paper presents a social network analysis of the social movement(s) that helped to create and protect Ecopark in Stockholm, Sweden, starting in the early 90's and continuing at time of writing. By conducting the social network analysis, Ernstson Sörlin and Elmqvist sought to gain a better understand of the social forces at work in urban social-ecological systems, and how those forces impact ES abundance and distribution. The Ecopark movement created a new narrative and vision for the Ecopark area and moved the discussion from the municipal to the national stage. Forcing municipal collaboration, and the involvement of country level planning, the Ecopark movement significantly impacted the governance of the area.  The diversity of social movement organizations, and the structure of the social network that brings them together, determines the protective capacity of the Ecopark. There was a high density of links between the core and periphery of the network, denoting strong associations. Core actors function as information brokers, and have a lot of power, because "... core and semi-core actors shape to a greater extent than others — whether consciously or not — the collective action that unfolds." With the network characterized as having a “core-periphery structure”, the authors contend that it "…has mutually reinforced two crucial processes, that of protecting the park from direct exploitation, and that of creating its identity through framing its values." An example of how values, as well as knowledge are important for collaborative ecosystem management.  119  An important justice consideration from the network analysis is that social movement structures can both “facilitate and constrain collective action.” Though the core-periphery structure helped to increase the park's protective capacity in the face of development in general, it reduced the ability of the social network to engage in collaborative management. For example, the authors found that user groups, such as allotment gardeners, tended to be marginal actors in the Ecopark movement. So, "...the domination of some actors — state agencies and/or civil-society organizations — can come to work as a conservative force by locking certain landscapes into a certain identity that could hinder experimentation and decrease adaptive capacity..." The authors also noted the necessity of having landscape and regional perspectives of ecosystem management, because the protective capacity of Ecopark might displace development pressure to other areas with lower protective capacity.  27. Estoque, R. C., & Murayama, Y. (2014). Measuring Sustainability Based Upon Various Perspectives: A Case Study of a Hill Station in Southeast Asia. AMBIO, 43(7), 943–956. http://doi.org/10.1007/s13280-014-0498-7  Estoque and Marayama sought to understand how the sustainability of urbanization can be assessed for a Baguio City, Philippines using a quantitative, triple bottom line approach. Their results showed that while the social and economic sustainability components improved from 2000 to 2010, the environmental component degraded over the same period. Sustainability might therefore be improving if each component of the triple bottom line were weighted equally, or declining if the environment component was weighted more heavily. The authors also found that the "... increasing trend of the EF [environmental footprint] of Baguio City indicates that the potential external environmental impact (leakage effect) of its urbanization has been increasing." Justice issues highlighted by the authors include persistent urban inequality in Asian cities despite rapid development, as well as the challenges inherent in balancing “… economic growth and environmental sustainability; economic sustainability and poverty reduction; and inequity and exclusion.” The authors also noted that lower data availability in developing countries requires different analysis techniques, like the use of proxies and indicators from other 120  studies. Further, appearances of sustainability might be belied by an externalized ecological footprint, the impact for which must be accounted for a holistic understanding of sustainability.   28. Filatova, T., Voinov, A., & van der Veen, A. (2011). Land market mechanisms for preservation of space for coastal ecosystems: An agent-based analysis. Environmental Modelling & Software, 26(2), 179–190. http://doi.org/10.1016/j.envsoft.2010.08.001  Filatova, Voinov and van der Veen seek to understand how urban expansion can affect ecosystem services in coastal areas by employing an Agent-Based Model (ABM) to assess how land markets and market instruments might affect the preservation of open space in a coastal urban area. Imposing a coastal development tax reduces property tax revenue for the town, but the tax will help recoup some of that revenue, and people may relocate away from coastal areas, which might be in line with some higher level conservation goals. There are landscape level impacts from actions taken in individual cities, however: "while total property value along the coast may go down, it will go up in some other municipality elsewhere," which may lead to development pressures in those locations.  The authors note that agents with higher incomes can better afford waterfront property, and might enter the market from outside the town if local people are priced out by a coastal development tax. Either way, a coastal development tax could exacerbate wealth inequality. If lower income people are priced out of waterfront property, it maybe prevents them “…from enjoying coastal amenities leaving the latter as a privilege for high-income group only." The authors argue that it might be easier to prohibit development in coastal zones rather than rely on market instruments, because raising the cost of land will disproportionately affect those with lower incomes.   29. Freitas, C. M. de, Schutz, G. E., & Oliveira, S. G. de. (2007). Environmental sustainability and human well-being indicators from the ecosystem perspective in the Middle Paraiba Region, Rio de Janeiro State, Brazil. Cadernos de Saude Publica, 23, 513–528. Retrieved from 121  https://apps.webofknowledge.com/full_record.do?product=UA&search_mode=AdvancedSearch&qid=15&SID=2Bw5iVlCLEa8hfJJc65&page=9&doc=90  Freitas et al. assess the environmental sustainability and well-being of the Middle Paraiba Region (MPR) in Rio de Janeiro State, Brazil. Human well-being is based on UN Millennium Ecosystem Assessment criteria, comprised of being endowed with basic materials, health, security, good social relations, freedom of choice and action. The study is primarily based on interviews with environmental managers in the MPR, and their assessment of how environmental impacts have led to losses in human-well-being. Results of the study point to a lack of planning and organization having led to both unsustainable natural resource appropriation and land occupation that have negatively affected well-being. For example, lack of access to basic sanitation has led to water pollution and potential disease exposure, while open-air dumps, devoid of the environmental or public health standards of landfills, were found to be used for 25% of solid waste. Though the institutional structure for environmental governance was found to be weak in the MPR, the authors noted that it still might be stronger than in other Brazilian regions. Despite governance issues, the majority of environmental conflicts were deemed to stem from legal and illegal industrial waste – challenging environmental impacts to mitigate given that industry is also the backbone of the regional economy.  Keeping with the “… principle of equity, it follows that the greater the social exclusion and the less the human development (affecting the components of human well-being), the more the population... become vulnerable to environmental and health problems.” The authors note that in a “vicious circle” environmental degradation tends to lead to social vulnerability which tends to further exacerbate social vulnerability. For instance, mortality due to respiratory illness increased both from unequal access to healthcare and slashing and burning air pollution, often practiced by poor farmers, which is also a driver of ecological degradation. Thus there are moral and ecological reasons for striving for an environmental sustainability that incorporates human well-being.    30. Heynen, N. C. (2003). The Scalar Production of Injustice within the Urban Forest. Antipode, 35(5), 980–998. http://doi.org/10.1111/j.1467-8330.2003.00367.x 122   Heynen, a regularly cited author with respect to urban environmental justice, presents an investigation of how scale affects production of injustice with regard to the provision of benefits from urban forests. The crux of Heynen’s argument is that what is just or not changes at different scales, therefore a multiscalar approach to justice assessment is required to assess justice trade-offs.  To illustrate the argument, a case study of Indianapolis’ urban forest is used. As with many urban areas, lower-income areas have fewer trees in Indianapolis, suggesting that those areas would be prime locations for expansion of the urban forest. However, an equitable distribution of trees may not yield the most efficient delivery of ecosystem services benefits to the most people. The ecological efficiency of trees reduces in the long term, and it makes sense to plant trees near existing stands for habitat contiguity and connectivity. However, this might often mean planting trees in already relatively wealthy neighbourhoods. Furthermore, how should one account for the global vs. local benefits of urban forests? If the focus is on local equity with an acceptance of reduced ecological efficiency, this can be a concern at the global level, for instance with carbon sequestration and storage ES mitigating anthropogenic climate change. Heynen writes that "... the externality effects brought about by the presence of urban trees fall between local-scale issues of social justice and larger-scale issues that contribute to ecological resilience." The scales of analysis chosen are arbitrary, but considering them still yields a greater understanding of how the justice of ES distribution can vary.   Heynen does not offer a definitive test of justice, his approach being more discursive, but the author does address many other urban ES justice issues. Other than observing that urban ES distributions tend to be unequally distributed along socio-economic gradients, Heynen also discusses how existing power structures tend to reinforce uneven urban environments. Those that have the most control tend also hold the lion’s share of capital, while "... those individuals and groups who lack access to recourses and the ability to have any meaningful control over capitalist production, consumption, and exchange tend to suffer environmental injustices that result in an overall lower quality of life."   123  31. Ibes, D. C. (2015). A multi-dimensional classification and equity analysis of an urban park system: A novel methodology and case study application. Landscape and Urban Planning, 137, 122–137. http://doi.org/10.1016/j.landurbplan.2014.12.014  Ibes introduces park classification and equity analysis methodologies, testing them on the case city of Phoenix, AZ, USA. The park classification is ‘multi-dimensional’ and quantitative using distance from the city centre, park size, land cover, and the built and social contexts of the park as classification criteria. Equity was assessed by statistically clustering park types based on quantitative attributes then relating those park types to the socio-economic context in which they were situated. Significant relationships between the five park types classified and neighbourhood social variables were found. One park type, with many amenities, was quite evenly distributed around the city. Large, more natural parks were associated with higher income areas which could either be because wealthier neighbourhoods were developed near them, or as a result of "white flight" from the city centre, or both. Lower access to these large parks could be an equity issue, but different ethno-cultural preferences mean it could not be. Ibes did not label the distribution as just or unjust, however, and only highlighted which park types were most associated with which socioeconomic groups. Indeed, the authors concedes that the equity analysis presented "… does not constitute a comprehensive analysis of urban park equity." It does allow for a relative ranking of more or less environmentally just areas in Phoenix, however. Population density was highlighted as an important equity indicator – the higher the density, the more public outdoor spaces may be necessary.  The author notes that research on the equity of urban park accessibility has increased in recent decades. Some studies “… reveal that disadvantaged groups have access to fewer park spaces, while others reveal that these disadvantaged populations have higher access to more parks, in number, but access to less park acreage and smaller spaces." Differing park preferences among ethno-cultural groups are also mentioned, with preferences based on “... the amenities and facilities present, landscaping features, geographic context, and other social, environmental, and built characteristics represent other important measures of quality park space, beyond size.” Further, Ibes notes that socio-demographics alone do not determine park use, but that visitation is 124  also dependant on “(1) the political, social, historical, and economic context of park spaces, (2) park amenities and environmental characteristics (e.g. landscaping, facilities, surrounding land uses), and (3) differing perceptions with regards to park accessibility, safety, and convenience." Phoenix itself, with a 40.8% Hispanic population, has a history of racial exclusion and segregation that has contributed to current social and environmental injustices.   32. Jenerette, G. D., Harlan, S. L., Stefanov, W. L., & Martin, C. A. (2011). Ecosystem services and urban heat riskscape moderation: water, green spaces, and social inequality in Phoenix, USA. Ecological Applications, 21(7), 2637–2651. http://doi.org/10.1890/10-1493.1  Jenerette et al. examine the spatio-temporal distribution of the cooling effect of vegetation in Phoenix, AZ, USA, and compare this with neighbourhood socioeconomic characteristics. Phoenix has a pronounced urban heat island (UHI) effect that has been exacerbated in recent years because of conversion of irrigated agricultural land to urban land covers. With high temperatures a potential risk to human health, and the burden of those temperatures possibly being born unjustly, the authors employed a spatially explicit risk assessment using ‘riskscapes’ of the distribution of a critical surface temperature threshold of 50oC. The amount of water required to sustain vegetation was used as a way to investigate ecosystem services trade-offs – in this case between a cooling ES and water consumption in an arid climate. The cooling effect of vegetation was modeled from weather and remote sensing data (surface temperature and NDVI), while the Gini coefficient was used to assess the variability in income and vegetation in the study area. The study results showed that higher income areas of Phoenix had lower critical temperatures and more cooling vegetation. NDVI and income in Phoenix both rose from 1970 to 2000 in general, but cooling vegetation became more concentrated in wealthier areas.  The authors assert that “[d]ecision-making about ecosystem services should also include an explicit treatment of associated costs and resource use trade-offs… as well as social equity considerations in the distribution of services.” Historical racial and ethnic segregation has impacted the current socio-economic distribution in Phoenix, with low-income and minority 125  groups clustered in the city centre – where temperatures are also highest – still segregated from wealthier, white populations in the suburbs. Increasing concentrations of cooling ES in wealthier areas was not a phenomenon before 1970, though the trend increased considerably since then, pointing to a growing disparity in wealth and equitable outcomes. Increasing green spaces in under-served areas would help to redress the unjust distribution of cooling ES in Phoenix, however this would also require a large increase in the amount of water used for irrigation of public green spaces. However, in order to reduce the overall public health risk and environmental vulnerability in Phoenix, increasing green spaces in neighbourhoods with lower median income will be necessary.  33. Jenerette, G. D., Miller, G., Buyantuev, A., Pataki, D. E., Gillespie, T. W., & Pincetl, S. (2013). Urban vegetation and income segregation in drylands: a synthesis of seven metropolitan regions in the southwestern United States. Environmental Research Letters, 8(4), 044001. http://doi.org/10.1088/1748-9326/8/4/044001 An investigation into how vegetation greenness differs between urbanized and 'natural' areas in the Southwestern US, Jenerette et al. also looked at how vegetation greenness varied with income within and among the 7 cities studied. The authors found that urban areas were generally greener than surrounding wildlands, and that higher greenness values were correlated with higher incomes. Furthermore, the “… socioeconomic effect was independent of overall differences between urban and wildland vegetation patterns—cities lacking such differences still showed socioeconomic effects.” The authors also note the trade-off between water use and vegetation abundance, especially in hotter climates, and that as precipitation increases the disparity of vegetation abundance along an income gradient shrinks.   34. Jennings, V., & Gaither, C. J. (2015). Approaching environmental health disparities and green spaces: an ecosystem services perspective. International Journal of Environmental Research and Public Health, 12(2), 1952–68. http://doi.org/10.3390/ijerph120201952  126  Jennings and Gaither present a review of how disparities in green space provision and access to ES affects health outcomes in the U.S. They posit that "… inequitable access to urban green spaces is an environmental justice issue that links characteristics of the natural environment to the development of some health disparities." They note that just having access to green spaces improves health outcomes, with recreation opportunities adding to this effect. The psychological benefits of green spaces were also noted.  A common pattern in the U.S. of less green space in low income and minority areas was found. Possible mechanisms for the disparity include higher densities in low-income urban cores, less funding for street tree maintenance, and more trees on the private property of the wealthy. The authors also noted that uneven green space distribution might be due to “… power differentials between socially marginal groups and middle/upper income whites." One upshot of green space inequality is that socially marginal groups are more vulnerable to climate change. Other studies of individual cities bear this out, with ethnic minorities and low income people in the US disproportionately more susceptible to heat stress - with higher heat hazards and lower levels of temperature mitigating vegetation. The psychological health of vulnerable groups is also often lower with lower self-reported psychological well-being and higher suicide rates.   35. Kabisch, N., & Haase, D. (2014). Green justice or just green? Provision of urban green spaces in Berlin, Germany. Landscape and Urban Planning, 122, 129–139. http://doi.org/10.1016/j.landurbplan.2013.11.016  Kabisch and Haase investigate the possible injustice of urban green space provision in Berlin, Germany. Urban green space (UGS) is defined as "forests, parks, cemeteries, allotment gardens and brownfields with vegetation. Per capita UGS is compared with the three "… socio-demographic indicators of population density, immigrant status and age." A traditional EJ definition is given in the one used by the EU, namely; “... equal access to a clean environment and equal protection from possible environmental harm irrespective of race, income, class or any other differentiating feature of socio-economic status.” However, the authors employ an alternative EJ framework “… combining the presented definition of environmental justice with the social justice concept developed by the anthropologist Low (2013).” Low argues that 127  addressing the injustice of public spaces requires an accounting of distributive, procedural and interactional justice: “While distributive justice focusses on the fair allocation of public spaces and related resources for all social groups, procedural justice relates to fair integration of all affected groups into the planning and decision process of a public space. Finally, interactional justice is about the quality of interpersonal relations in a specific place and if people interact safely without, e.g. discriminant behavior.” There are two spatial scales of analysis, a city-wide scale, as well as a case study of the Tempelhof airport park. Specific social groups, such as immigrants and the elderly are also separated out in order to see how UGS provision varies among populations.  The minimum threshold for UGS in Berlin is 6m2 per capita, which most sub-districts met. Overall, there was a "... negative relationship between the provision of UGS and population density.” Cluster analysis was used to categorize the variance in UGS and socio-economic variable with three main clusters identified – a densely populated, low-vegetation cluster in the city centre with a high proportion of immigrants; a low density, high-vegetation cluster on the city periphery; and an intermediate cluster, where the means for all variables tested were below the city-wide average. The Gini index was used to quantify the distribution of UGS by district, with the finding that UGS distribution was highly unequal, constituting distributive injustice. This was especially the case for immigrants who had a Gini value of 0.84, compared with values of 0.69 and 0.65 for all residents and people over 65, respectively.  The neighbourhoods surrounding Tempelhof had high immigrant populations as well as individuals aged over 65. These demographics were not represented in park user data, however, with the majority of park users being younger Germans (only 9% were immigrants). Minority groups and the elderly may have desired different park amenities than existed at Tempelhof at the time of study, when there was little in the way of developed park facilities. The findings at Tempelhof were distinct from the pattern of distributive injustice found at the city scale, and may more closely relate to procedural justice concerns about effective consultation with nearby residents on appropriate park amenities. A nod is also given to interactional justice, with the assertion that "...interactional justice, UGS should allow for all visitor groups – regardless of age and cultural background – to interact freely and safely."   128  36. Lakes, T., Brückner, M., & Krämer, A. (2014). Development of an environmental justice index to determine socio-economic disparities of noise pollution and green space in residential areas in Berlin. Journal of Environmental Planning and Management, 57(4), 538–556. http://doi.org/10.1080/09640568.2012.755461  Lakes, Brückner, and Krämer “… study the relationship between socio-economic inequality and environmental burdens and benefits” in Berlin’s residential areas using an environmental justice index (EJI). The study is explicitly characterized as an environmental justice study assessing distributive justice. Using average NDVI and average noise (dB) values for each residential planning unit an environmental indicator (EI) was calculated. The EI thus utilized an indicator each of environmental benefits and burdens. A ready-made social development indicator (SI) was normalized and combined with the normalized EI to create an environmental justice index. Using the EJI, 'hotspots' and 'coldspots' of EJ can be identified, allowing for a relative spatial ranking of justice in a city. In Berlin "… the planning units with double burdens of low environmental and socio-economic characteristics, the hotspots of environmental inequality, are located in the central parts of Berlin," though there were two clusters of lower socio-economic status in the city’s outskirts. The authors found that there were 22 (5%) of 434 planning units that had both low EI and SI, although 130 planning units had burdens attributed to one indicator or the other. EJ ‘coldspots’, of which 47 were identified, were mostly closer to the edge of the city, corresponding to areas of higher income and more abundant green space. Interestingly the EJI was not found to vary along a distinct gradient in Berlin, showing a more fragmented pattern. The authors hypothesize that wealthier inner city areas that have been subject to recent gentrification may have contributed to this fragmented effect. Regardless, environmental quality was correlated with socio-economic status.  37. Lal, R. (2007). Soil Science and the Carbon Civilization. Soil Science Society of America Journal, 71(5), 1425. http://doi.org/10.2136/sssaj2007.0001  129  A summary and review of the challenges and opportunities for soil science and soil conservation in the 21st century. Increasing demands on soil resources, accelerating rapidly in the latter 20th century, are summarized, with 10 types of challenges noted: global food security, water scarcity, water management, soil biodiversity, desertification control, climate change, genetic diversity and abundance, planetary and human history, energy needs and biofuel, and finally, urban soils and drastically disturbed lands. Few justice issues are mentioned by Lal, though yield gaps and food insecurity in the developing world are highlighted, with the Millennium Development Goal of cutting world hunger by 50% by 2015 was not likely to be met at time of writing. In light of that, the use of soil to produce biofuel must be carefully weighed against displacement of food crops, water and energy needs, with the author arguing that full LCAs are required to compare alternatives.  38. Larson, K. L., Wiek, A., & Withycombe Keeler, L. (2013). A comprehensive sustainability appraisal of water governance in Phoenix, AZ. Journal of Environmental Management, 116, 58–71. http://doi.org/10.1016/j.jenvman.2012.11.016  Larson, Wiek and Keeler present assess the sustainability of the water system in Phoenix, including the resource supply, infrastructure, rule-creating institutions, actors, types of water demand and water-related activities.  The authors also present a comprehensive definition of water sustainability for Phoenix with 7 principles: social-ecological system integrity, resource efficiency and maintenance, livelihood sufficiency and opportunity, civil engagement and democratic governance, inter-generational and intra-generational equity, interconnectivity from local to global scales, and precaution and adaptability. Sustainable water governance is defined “… as the decision processes of stakeholders who influence and are impacted by activities involving water supplies, deliveries, uses, and outflows in ways that ensure a sufficient and equitable level of social and economic welfare without compromising the viability and integrity of supporting hydro-ecosystems now and into the future."  The sustainability analysis uncovers numerous justice issues. Despite water rights being based on the 'first in time, first in right' principle, local aboriginal peoples were excluded from 130  holding water rights by settlers and continue to fight for water rights. There is little participatory decision making, with few powerful interests controlling decision-making processes. This may have contributed to urban and industrial water uses being favoured over agricultural uses. Low-income neighbourhoods are more exposed to pollutants in ground and surface water, for example from the industrial pollution of a Motorola company site. The quality of water that Mexico, just downstream from Phoenix, receives is also very poor, a distributive and procedural justice problem. The authors also found that future generations receive little consideration through formal governance, with planning time frames extending to a maximum of 100 years, or just a few generations.  39. Lwasa, S., Mugagga, F., Wahab, B., Simon, D., Connors, J. P., & Griffith, C. (2015). A meta-analysis of urban and peri-urban agriculture and forestry in mediating climate change. Current Opinion in Environmental Sustainability, 13, 68–73. http://doi.org/10.1016/j.cosust.2015.02.003  Lwasa et al. conduct a literature review focusing on how urban and peri-urban agriculture and forestry (UPAF) can help urban areas mitigate and adapt to climate change. They focus on "… eight East and West African cities in a range of coastal, inland and mountainous ecosystems." Rapid urbanization in Africa is part of the justification of the need for this work, with many African cities facing increased extreme weather frequency and falling levels of ES.  Rapid growth often leads to unplanned development, which is “… associated with increased inequality and vulnerability of urban populations to climate change impacts." UPAF can help to mitigate this, and the authors argue that there is significant potential for UPAF to help transform African cities to become more inclusive and sustainable. UPAF was found to have a well-documented positive effect on livelihoods, helping to reduce food insecurity and poverty as well as sensitivity to environmental stresses. The potential benefits of UPAF from the reviewed literature are clear, but scaling up of UPAF initiatives faces three challenges: development deficit (lack of green infrastructure), institutional reform (to support multifunctional urban landscapes), and knowledge sharing. The majority of reviewed literature treated adaptation and mitigation separately, with a much greater focus on adaptation.  131   40. Marques-Perez, I., Segura, B., & Maroto, C. (2014). Evaluating the functionality of agricultural systems: social preferences for multifunctional peri-urban agriculture. The “Huerta de Valencia” as case study. Spanish Journal of Agricultural Research, 12(4), 889. http://doi.org/10.5424/sjar/2014124-6061  This paper presents an analysis and characterization of the "agricultural functional systems" of Valencia, Spain using a multifunctional agriculture theoretical framework. The aim is policy recommendation and ecosystem services identification and valuation. Consumer demand for and value of ES is established by applying the Analytical Hierarchy Process (AHP). Authors "... created a descriptive approach for the multifunctionality of agricultural systems by grouping their various functions and the goods and services that they provide, according to their economic (E), social (S) and environmental (EN) dime [sic]." The AHP was used to calculate a social utility function of Valencia’s agricultural functional system, with aim of maximizing the utility provided by the agricultural system to society. The main data used in the study were survey responses, however, the authors focused more on the theoretical underpinnings of the hierarchy and social utility function than on the survey results, or what those results mean for Valencia and for policy makers in Valencia.  41. McDonald, R. I., Forman, R. T. T., & Kareiva, P. (2010). Open space loss and land inequality in United States’ cities, 1990-2000. PloS One, 5(3), e9509. http://doi.org/10.1371/journal.pone.0009509  McDonald, Forman and Kareiva “… measure the open space lost from urban growth in all 274 metropolitan statistical areas (MSAs) in the contiguous United States from 1990 to 2000." Their analysis went beyond a simple urban expansion study, however, by looking at the equity of land consumption for each MSA on a per capita basis, the impact of zoning regulations (“…classified as either ‘Traditional’, ‘Exclusion’, ‘Reform’, or ‘Wild Texas’ zoning”) on open space loss, as well as whether conservation funding in an MSA tends to mitigate loss. The analysis was conducted at the MSA level, as well as a finer scale of census blocks, the latter 132  allowing them to incorporate housing density data. Using this data, the authors also assessed the land inequality of a city (few houses using more land is more unequal) using the Gini Coefficient. The authors found that most metropolitan areas grew in population from 1990-2000 and that open space loss was strongly correlated with population growth. Results also showed that land consumption per capita varied by an order of magnitude among MSAs, with smaller land consumption per capita generally in larger cities and vice versa. Developed land consumption per person generally decreased over the study period with some exceptions. Greater conservation funding was related to a decrease in per capita land consumption (though the significance of this relationship varied), as was "reform" zoning as compared to traditional zoning. Bigger cities tend to have denser housing, and the denser the city the more likely there is higher conservation funding. Though interestingly, "In contrast to the correlation between conservation funding and density, zoning type is not significantly correlated with the proportion of housing in dense neighborhoods." Midwest and East-coast cities had some residential density reductions, while West Coast cities became denser. Cities that became more residentially dense also had a greater decrease in per capita land consumption. Accounting for relationships that one might expect to be positive, the authors found that the "… change in proportion of houses in high density neighborhoods was not correlated with the median house value, the city size, or the degree of conservation funding." The authors also note that contemporary and historical factors impact urban growth. Contemporary factors like conservation funding, reform zoning and population growth can lead to higher density cities and lower land consumption per capita while also correlating with a small increase in land inequality. Historical factors include the sprawling of cities with automobile-oriented development leading to lower population densities and more consumption of land per capita. The main equity concern of the paper was the measurement of the land inequality using the Gini index. Higher values denote more land consumed by fewer households and vice versa (note that this is not the same as per capita land consumption). Results showed that, in general, relatively few residents tend to consume most of the land at a low density, with a national average Gini value of 0.63 and a range of 0.35 to 0.93. The most equal cities were those in the Rust Belt with declining population, while East Coast cities were generally more equal than 133  West Coast ones. The Gini coefficient was fairly static over the decade but cities that saw an increase in the proportion of high-density residential areas, and/or had reduced per capita land consumption saw an increase in the Gini coefficient, and inequality. Cities that spent more than $100 per person on conservation also saw an increase in land inequality. City size, zoning category and median house value, however, were not related to changes in Gini values. The upshot of these results is that “… the preferences and economic choices of a relatively small number of urban residents are associated with much of the natural land-cover lost to development." The authors recommend that future policy efforts may need to target "… this subset of residents causing most natural habitat loss in order to limit the impact of urban development on natural systems."  42. Newell, J. P., Seymour, M., Yee, T., Renteria, J., Longcore, T., Wolch, J. R., & Shishkovsky, A. (2013). Green Alley Programs: Planning for a sustainable urban infrastructure? Cities, 31, 144–155. http://doi.org/10.1016/j.cities.2012.07.004  Newell et al. consider “… eight alley greening programs in seven US cities,” to assess how the programs create green infrastructure in support of urban sustainability planning.  A triple-bottom line definition of sustainability is used and green infrastructure is identified based on three 'core ideas': connectivity, multifunctionality, and green. Stormwater management is the principle ES of the green alley programs surveyed, though Chicago and Los Angeles’ programs also have the goals of mitigating the urban heat island effect and mitigating light pollution. Chicago also seeks to enhance energy conservation through green alley programs, and Los Angeles seeks to reuse storm water. Despite mentioning seven US cities, the paper mainly focuses on green alley programs in Los Angeles.  The authors are concerned with both the distributional and procedural equity of green infrastructure, arguing that “… the production of green infrastructure space must involve ensuring democratic participation in decision-making processes about the qualities of that infrastructure." They find that their concerns are echoed in the green alley programs themselves, with some elements of the Los Angeles program seeking “… to build community and to empower residents to improve their neighborhoods, such as through the involvement of residents 134  in the alley design process.” Baltimore's Gating and Greening program must also be accepted by 80-100% of residents adjacent to a project area, and residents have a voice in articulating what "greening" means in their alleyways. Despite this, economic development and social equity goals are mostly lacking in green alleyway programs, with the exception of the programs in LA. It was also found that green alley programs are mostly implemented where there is already less access to parks and green spaces, with the implication that equity considerations are especially important to redress both distributional and procedural inequalities. The neighbourhoods where green alley programs were being implemented in LA, for instance, had lower incomes, more ethnic minorities, lower health outcomes, higher flood risks, and fewer parks and green spaces than the LA average. By building green alleys that provided ES in the form of stormwater mitigation, recreation and natural beauty, among others, multiple distributional inequalities were addressed. Procedural inequities were also addressed in LA, with green alley program agencies proposing “… to survey residents on their needs and priorities related to alleys and to use the results as a design template."  43. Norman, L. M., Villarreal, M. L., Lara-Valencia, F., Yuan, Y., Nie, W., Wilson, S., … Sleeter, R. (2012). Mapping socio-environmentally vulnerable populations access and exposure to ecosystem services at the U.S.–Mexico borderlands. Applied Geography, 34, 413–424. http://doi.org/10.1016/j.apgeog.2012.01.006  Norman et al. strive to map ES, specifically flood and erosion control regulating services, and the degree of access that socio-environmentally vulnerable populations have to them. They do this for the Santa Cruz watershed, which straddles the Mexico-U.S. border. The authors couple a biophysical model (The Soil and Water Assessment Tool (SWAT)) and a vulnerability index (the Modified Socio-Environmental Vulnerability Index (M-SEVI)) to achieve their goal. The results show that flood and erosion control ES were declining in most of the populated locations within the study area, likely from urban expansion. This pattern mostly held for socio-environmentally vulnerable (M-SEVI > 0.22) populations, though they didn't perform any statistical test to show whether there was a significant difference between vulnerable and more resilient neighbourhoods. A broad discussion about how landuse/landcover trends was not 135  undertaken, which did not enable them to ascertain whether socio-environmentally vulnerable populations had fewer ES than more resilient populations.  Regardless, the authors did highlight numerous justice issues in the Santa Cruz watershed. They define environmental justice as “…the concept that environmental burdens and benefits should be equally distributed to all people to ensure a safe, healthy environment for all…” and that the EJ movement has not always integrated well with local sustainability initiatives concerned with social equity.  It is asserted that both global and local processes impact vulnerability in the study area, “… as well as class, ethnicity, age, and gender.” Indeed, Hispanic people are undercounted in the US census at a rate of 7:1. Colonias, informal settlements of Mexican migrants near the border, are a particular focus. Colonia residents often lack “…the resources to fund infrastructure improvements needed to minimize environmental degradation associated with development,” and are often employed in low-skilled service jobs that lack security. 12 of the 15 vulnerable locations identified in the study area were colonias. Finally, Norman et al. highlight the need for those managing the land to understand the geography of ES and socio-environmental vulnerability in order to effectively consider human well-being in their management decisions.   44. Pasquini, L., Ziervogel, G., Cowling, R. M., & Shearing, C. (2014). What enables local governments to mainstream climate change adaptation? Lessons learned from two municipal case studies in the Western Cape, South Africa. Climate and Development, 7(1), 60–70. http://doi.org/10.1080/17565529.2014.886994  Pasquini et al. “... propose a set of enabling factors” that they argue enables cities to more easily engage in climate change adaptation policies. They use Cape Town and Hessequa, relatively larger and smaller municipalities in South Africa, as case studies. There is a growing use of ecosystem-based management where biodiversity and ecosystem services are understood to provide value and reduce costs associated with climate change. There are key knowledge gaps, however, including how "institutional size, and the social network dynamics of a municipality” will affect the ability of institutions to engage in adaptation activities. There is also an issue “… the issue of political stability and party politics effects on local government functioning,” the 136  destabilizing effects of which require further scrutiny. Few justice issues were mentioned in the study, except for the need to address climate-related food insecurity with the establishment of urban garden in "previously disadvantaged areas."  45. Pattanayak, S. K., Ross, M. T., Depro, B. M., Bauch, S. C., Timmins, C., Wendland, K. J., & Alger, K. (2009). Climate Change and Conservation in Brazil: CGE Evaluation of Health and Wealth. The B.E. Journal of Economic Analysis & Policy, 9(2), Article 6.  Pattanayak et al. use a computable general equilibrium (CGE) econometric model to estimate “health-related ecosystem values.” Specifically, they look at how forest cover relates to mosquito-borne disease prevalence in Brazil in a spatially explicit way. They find that as deforestation rates increase so do the incidence rates of malaria and dengue fever in a statistically significant correlation. Their model results also show, however, that a Brazilian government policy to maintain more forest cover on the landscape actually has slightly negative economic consequences “… with the decline in output and GDP from restricting deforestation slightly more than counterbalancing the positive economic effects of improving rural health." Multiple justice issues are raised by the authors, especially in the interpretation of their results. They note that the trade-off between economic efficiency and equity is a common one that requires answering the question “… how much efficiency (negative GDP and other macro indicators) are we willing to give up so we can improve the wellbeing of rural laborers and indigenous people of the Amazon region?” The poor, and the rural poor especially, do not have the same political power to advocate for their position, which complicates ES value estimation and leads to unequal ES provision. The authors note that drivers at multiple scale effect ES provision, too, with global environmental changes driving "...losses in biodiversity and ecosystem functions – the latter being a direct contributor to the flow of socially valuable goods and services, particularly for the poor in the developing world…"   46. Peng, J., Du, Y., Ma, J., Liu, Z., Liu, Y., & Wei, H. (2015). Sustainability evaluation of natural capital utilization based on 3DEF model: A case study in Beijing City, 137  China. Ecological Indicators, 58, 254–266. http://doi.org/10.1016/j.ecolind.2015.06.002  Peng et al. conduct an analysis of the sustainability of Beijing, China, using an ecological footprint (EF) model. A chief goal of the study was to integrate “…ecology, equity, and efficiency” by measuring the depth and size of the EF of the 16 districts of Beijing. The Gini coefficient is used to “…measure equity in natural capital utilization” and “the efficiency of natural capital utilization” is measured using the HDI and similar indices.  Results showed obvious disparities in HDI and EF in Beijing, with intensity of natural capital use and HDI highest in the centre and decreasing towards peri-urban areas. However, the authors argue that when looking at Beijing as a whole that the EF is balanced. Curiously, inequality uncovered in the results (see varying Gini coefficients and coefficients of variation below) are dismissed by aggregating the results to the city scale. The distributive justice of resource use was a primary concern addressed by the authors. They argue that “[m]easuring the distribution of natural resource use is necessary in achieving justice and efficiency,” an important step to surpassing the limited equity considerations of neoclassical economics. Inter- and intragenerational equity were also noted as important temporal and spatial components of a full accounting of equity. ES were not explicitly investigated, though the justice of their distribution was assessed by proxy using “… the Gini coefficient of EFsize per capita (G) and the variation coefficient of EFdepth per capita (C.V) to characterize the equality of natural-capital-flow consumption and natural-capital-stock occupation." After dividing Beijing into four “functional zones” it was found that the value of G within each zone was quite low – below 0.2 – indicating relative equality. A value of G of 0.583 was found when comparing the four zones, however, indicative of “… the high imbalance of capital flow occupation among the different functional zones and districts (i.e., the intra-generational equity of natural capital spending is low)."  47. Perkins, H. A. (2011). Gramsci in green: Neoliberal hegemony through urban forestry and the potential for a political ecology of praxis. Geoforum, 42(5), 558–566. http://doi.org/10.1016/j.geoforum.2011.05.001 138   Perkins argues from a Gramscian perspective that urban trees and forests have multiple, contested perceptions, and that particular, neoliberal conceptions of urban tress are championed by the state and civil society actors in service of the state. The core of the argument is that an ecosystem services narrative of urban trees is often used in service of a neoliberal agenda: the ES of trees are part of a pedagogy that creates consent for the commodification of trees which allows for the welfare state's role in environmental stewardship to be diminished and replaced by civil society organizations in the interest of generating profit. Another ramification of valuing urban trees using ES within a capitalist market is that a division of labour privileges the role of “state-sanctioned ‘experts’” over the population at large, with implications for procedural justice. Furthermore, markets “… are inherently discriminatory in their constitution based on factors of race, gender, socio-economic status, and the like.”   The focus of the paper is on how urban tree pedagogy is used to coerce and entice populations into consenting to neoliberal hegemony. The urban tree "... pedagogy is not just about the trees as it appears on the surface; it is about the ‘proper’ market relations for producing the trees within a capitalist economy." A coercive pedagogy is required because of the multiplicity of perceptions regarding the urban forest.  The state uses its own intellectuals to educate prominent actors in civil society on the ES of urban trees, the involvement of whom add legitimacy to the ES narrative. Perkins argues that this “…dialectical process limits, yet potentially gives rise to, an alternative political ecology of praxis that can build socio-natural hegemony without capitalist, market ideology." In other words, though the ES narrative championed by state and state-allied actors in Wisconsin is in service of a neoliberal agenda to commodify the urban forest, it also enhances the profile of urban ecology in the public consciousness, allowing for the articulation of alternative narratives of ecological valuation.    48. Pincetl, S. (2012). Nature, urban development and sustainability – What new elements are needed for a more comprehensive understanding? Cities, 29, S32–S37. http://doi.org/10.1016/j.cities.2012.06.009  139  Pincetl authors a review paper looking at how to conceptualize urban nature through ecosystem services, urban metabolism and urban political ecology. The review synthesizes these three perspectives in order to build a theoretical framework that includes the built environment of cities into an urban ecology that is explicit in its treatment of cities as novel ecosystems. Urban metabolism studies can provide data on resource flows for political ecologists to understand what systems and power structures drive resource flows and how. Pincetl asserts that there is a difference between ecology in the city and ecology of the city. Ecology in the city sometimes suffering from a "...tendency in ecosystem science to naturalize cities, using models and concepts from ecology... [which can leave] little room for human agency, for the ways in which humans build the cities, (re)place nature, the functioning of nature in the city itself, and the impacts on far flung ecosystems and social systems for urban needs."  Equitable ES provision is a concern of the author, noting an increase in interest on the topic since the early 2000s and that the maintenance and management of urban ES have equity impacts. The author notes that equity is a key concern of a sustainable city, and that “… it is necessary to focus on the political economic processes and relations that bring about urban environmental change and social inequality. This is because, though nature provides a foundation for human civilization, social processes “…produce nature’s and society’s history.’’  49. Porter, J. R., Dyball, R., Dumaresq, D., Deutsch, L., & Matsuda, H. (2014). Feeding capitals: urban food security and self-provisioning in Canberra, Copenhagen and Tokyo. Global Food Security, 3, 1–7. Retrieved from https://apps.webofknowledge.com/full_record.do?product= UA&search_ mode=AdvancedSearch&qid=15&SID=2Bw5iVlCLEa8hfJJc65&page=11&doc=102  Porter et al. assess the global land area needed to provide the food security of three wealthy capital regions (CRs) – Canberra, Australia; Copenhagen, Denmark; and Tokyo, Japan – from 1965 to 2005. The authors used regional and national food commodity data to address four key areas of inquiry: “… (i) how the CRs' food systems have changed over time, (ii) the origins of each CR's food in terms of the geographical location of local or distant food supplying landscapes (iii) the size of the land areas that the CRs sequester for production of their consumed 140  food and (iv) the degree of each CR's food security and self-provisioning capacity." The analysis compared the extent of local and domestic food provisioning with the volume of imported food, and how that relationship changed through time. An increasingly global market for food has led to individual countries focusing on food commodities that maximize their comparative advantage. The authors caution, however, that in the face of increased pressure on food supply from demographic (e.g. population growth and changing diets) and environmental factors (e.g. climate-change induced drought or flooding) a reassessment of food security might be necessary. For the authors this requires a balancing of market forces and free-market doctrine with an increasing valuation of ecosystem services that might lead to better valuation of the ecosystem processes underlying food production and a shift towards more domestic-oriented food production strategies.  The authors raised few justice concerns, and offered no definitive test or assessment of justice. Justice issues are raised by the authors through the concern that global trade in food can come “… at the expense of the source nation's capacity to feed its own population.” Receiver nations, therefore, might need to depend on food, and the land management decisions that enable its production, in places where they have no jurisdiction. Further, exporting nations’ pursuit of cash crops may lead to degrading land use practices that lead to reduced agricultural productivity through time. The authors advocate a tracking of ecosystem services that support global food trade would help decision-makers better understand trade-offs between local and global food production, meeting the food security of their populations while mitigating risk from socio-economic and environmental sources.   50. Safransky, S. (2014). Greening the urban frontier: Race, property, and resettlement in Detroit. Geoforum, 56, 237–248. http://doi.org/10.1016/j.geoforum.2014.06.003  Safransky uses the historical perspective of colonialism to investigate how the redevelopment of Detroit is being undertaken using narratives and tactics that can be traced to a long history of US settler colonialism. Safransky’s stated goal is to connect neoliberalism and gentrification to a racist history of settlement, and to “… make visible the often-invisible geographies of settler colonialism so that we may then consider what it would mean to 141  decolonize Detroit and other places and stand in solidarity with those already working to do so." In recent decades the population of Detroit, MI has fallen considerably, leading to foreclosures and many vacant properties. In the midst of a fiscal crisis ‘green’ redevelopment is being touted as the way forward in Detroit by city managers. Expanding the role of green infrastructure and ecosystem services is seen as a method to replace built infrastructure and to limit service extension in the city, thus shrinking the role of the municipal government. Detroit city managers commissioned a study to “…rationalize investment and disinvestment in technical rather than political terms and to allocate scarce resources with more geographical precision”. This study evolved into the “… Detroit Future City plan and deploys green infrastructure as the key strategy by which these disconnected spaces will be repurposed." Detroit planners emphasized vacancy, which allowed them to avoid talking about removing people from the land, rather repurposing a “…purportedly neutral landscape removed from the deeply political conditions of its production." The underlying justice issue of green redevelopment is that “… the costs and benefits of green redevelopment are distributed unevenly within the context of gentrification and bankruptcy."  A key method by which land is opened for development is by categorizing it as empty. Treating land as empty can erase values held by those still attached to or using the land, such as community gardeners, or displace people themselves. Abetting an empty land narrative is one of urban decay that “… taps into post-apocalyptic cultural imaginaries in a way that can deemphasize the ongoing struggle of the city’s hundreds of thousands of human residents." Safransky notes, too, that those residents often have low-incomes (Detroit has a 50% unemployment rate) and are often visible minorities (83% of the residents of Detroit proper are African-American), and so are already subject to economic and social power imbalances. Considering land empty paves the way for settlement and the assignment of private property rights to settlers who are, in the case of Detroit’s gentrifiers, often middle-class and white. A particularly problematic case of private property rights assigned in the name of ‘green’ redevelopment is the case of Hantz Woodland, a large land acquisition by a single individual to create an urban tree farm in the face of significant community opposition.  Uncovering settler colonialism at work, Safransky argues, is an important part of decolonizing spaces, and of articulating alternative narratives of value: "As planners and 142  development boosters champion market-based greening, austerity, and gentrification, social movements contest the values embedded in this territorial restructuring, particularly its inattention to racial inequality." There is also an emerging literature critical of ‘green gentrification’ “… which has argued that seemingly benevolent or benign urban greening and ecological restoration projects perpetuate inequities (rising rents, displacement, unequal access)."   51. Shackleton, S., Chinyimba, A., Hebinck, P., Shackleton, C., & Kaoma, H. (2015). Multiple benefits and values of trees in urban landscapes in two towns in northern South Africa. Landscape and Urban Planning, 136, 76–86. http://doi.org/10.1016/j.landurbplan.2014.12.004  Shackleton et al. investigate the ES of trees in three types of neighbourhood – townships, ‘RDP areas’, and informal settlements – in two different South African cities. Each neighbourhood type tends to have a different vegetative composition providing different sets of tangible and intangible benefits to its residents. Each neighbourhood type also has residents of different socio-economic status that the authors hypothesized may value the services provided by trees differently. Residents of RDP areas, planned social housing areas that often had little vegetation, tended to value trees highly. Residents of informal settlements also valued trees for cooling, food and fuel provisioning and cultural and spiritual reasons, a suite of important uses that the authors noted were often not acknowledged by planners. Residents of townships that were generally more well off tended to value the ES provided by trees less (though still valued them), especially desiring their aesthetic services over cooling or food provisioning. Trees in public and private spaces were also valued differently, with public trees sometimes seen as reducing street safety.  The authors considered a number of justice issues in their analysis. They assert that an understanding of the valuation, benefits of burdens of the urban forest is “… critical for urban policy and planning that promotes social justice, equity, well-being and sustainability." Each of the neighbourhood types chosen have been neglected in terms of vegetation planning and management and also have a strongly racialized history deeply intertwined with the history of Apartheid. Blacks were excluded from urban areas, except townships, which have "high density 143  housing, poor services, limited commercial opportunities, few recreational green spaces or aesthetic features like street trees and other plantings, and widespread poverty." There were considerable post-apartheid social housing developments, but these "‘RDP areas'... like township development in the past, generally lack planning attention to recreational green spaces and visually appealing elements." Trees, therefore, become a vital part of the landscape for the residents of the neighbourhood types studied. Trees are sometimes rare, however, and tree planting faces numerous barriers including small plot sizes, and “… poor water supply, restrictions on the use of water, insecure tenure in informal settlements, and policy and legal systems that generally preclude public investment on private land and in personal spaces." Park spaces were sometimes seen by residents as unsafe, especially those characterized by poor maintenance, which the authors note could be a self-reinforcing hold-over from the Apartheid era.    52. Steenberg, J. W. N., Millward, A. A., Duinker, P. N., Nowak, D. J., & Robinson, P. J. (2015). Neighbourhood-scale urban forest ecosystem classification. Journal of Environmental Management, 163, 134–145. http://doi.org/10.1016/j.jenvman.2015.08.008  Steenberg et al. present an urban forest ecosystem classification (UFEC) methodology and apply it to the case city of Toronto, Canada using a hierarchical cluster analysis. The classification framework “… integrates 12 ecosystem components that characterize the biophysical landscape, built environment, and human population,” while its application in Toronto “… used 27 spatially-explicit variables to quantify the ecosystem components.” The cluster analysis identified 12 ecosystem classes in Toronto. In general, “… canopy cover was positively related to economic wealth, especially income,” though was not always consistently positively correlated with home ownership or level of education. Open green space and tree density (stocking) were “… more closely related to population density, building intensity, and land use,” than income or other socio-economic and demographic indicators. The authors argue that the UFEC can help urban forest managers to better prioritize management actions, accounting for whole ecosystem functioning rather than maximizing individual ES.  144  Socio-demographic data are included in the UFEC explicitly to “… aid in the consideration of social equity as far as urban tree canopy distribution is concerned.” The authors note that urban green space and ES are often unequally distributed with regards to income and ethnicity, an assertion broadly supported by the results of the Toronto case study. It is also acknowledged that “… there are certainly instances of different ethnocultural groups preferring an absence of tree cover on their property…” which is still an issue of equity in terms of urban forest distribution. Given the unequal access to urban ES provided by the urban forest, the authors argue that prioritizing limited municipal investment into urban forests for the benefit of those currently underserved is a key priority, and one that can be aided by employing UFEC.   53. Strohbach, M. W., Lerman, S. B., & Warren, P. S. (2013). Are small greening areas enhancing bird diversity? Insights from community-driven greening projects in Boston. Landscape and Urban Planning, 114, 69–79. http://doi.org/10.1016/j.landurbplan.2013.02.007  Strohbach, Lerman and Warren investigated whether small, community-driven green space benefitted avian diversity in Boston, MA. The community-run green areas of interest were compared with random sites nearby, as well as larger, wooded green areas. The authors found that the small greening projects that most benefitted avian diversity tended to connect to or expand on existing green space. Because urban green space has an unequal distribution in Boston, the authors noted that preserving biodiversity might represent a trade-off with environmental justice.   54. Unnikrishnan, H., & Nagendra, H. (2014). Privatizing the commons: impact on ecosystem services in Bangalore’s lakes. Urban Ecosystems, 18(2), 613-632. http://doi.org/10.1007/s11252-014-0401-0  Unnikrishnan and Nagendra explore how restricting lake access in Bangalore, India affects ES access and provision. Local authorities granted management and exclusion authority over some of Bangalore’s lakes to private companies based on the perception that nature was 145  being overexploited. The companies were also empowered to “… develop the lakes into profit-making recreational facilities for the paying public.” The authors compare how ES provided by privatized lakes compare to those of public lakes. The authors visited the lakes, mapping lake users (and evidence of their use) to determine ES usage and provision. They determined that private and public lakes did have different levels of ES provision, with public lakes providing more as well as different types of ES users.  The authors argue that by privatizing urban commons “… there has been an alienation of both actors and social networks from the ecological landscape.” An especial focus of the authors is the assessment of cultural ES, observing that “ecosystem services derived from a resource are a product of many experiences and interactions people have with the ecosystem… and hence may not fit neatly into specific categories.” Privatization of the lakes was accompanied by a “discursive shift” from abundance to scarcity, leading to the commodification of lakes and the ES they provide. The authors also posit that exclusionary discourses that lead to exclusion from, and commodification of, commons might most affect cultural ES precisely because they are a product of complex social-ecological interactions. The authors argue that the alienation “… of both actors and social networks from the ecological landscape” has serious justice implications for both provisioning and cultural ES. Public private partnerships (PPPs) have benefits, in the form of private capital investments, however, the cost of PPPs can be commoditization of ES, which can lead to price increases that exacerbate existing inequalities. The most socio-economically vulnerable are likely to be the first to be priced out. Indeed, those “… dependent on traditional lake-associated livelihoods and on lakes for domestic and subsistence use often belong to already marginalized communities of village inhabitants and migrant workers.” Commoditization can lead to individual ES being valued over ES bundles and can also change the “cultural imaginary” of resources, both of which can negatively impact the provision of cultural ES. In the long run, exclusion of those most dependent on lake ES, both for livelihoods and culture, might reduce social-ecological resiliency in Bangalore.   55. While, A., Jonas, A. E. G., & Gibbs, D. (2010). From sustainable development to carbon control: eco-state restructuring and the politics of urban and regional development. 146  Transactions of the Institute of British Geographers, 35(1), 76–93. http://doi.org/10.1111/j.1475-5661.2009.00362.x  While, Jonas and Gibbs propose a conceptual framework that understands state environmental policy and action arising out of a continual restructuring process that incorporates select environmental goals. Eco-state restructuring (ESR), as it is called, is then illustrated by examining carbon governance – in particular, how countries’ Kyoto obligations have resulted in a process of ‘downscaling’ national regulatory commitments to lower levels of government. The authors maintain that "... ‘downscaling’ of carbon management raises a series of issues about the distribution of responsibility for carbon reduction within and across existing jurisdictional boundaries, the capacity and willingness of different authorities to respond, and the extent to which prevailing governance norms enable and constrain different types of urban and regional response.” Four main effects of carbon control on regional and urban development are highlighted: (1) compliance with low carbon regulation; (2) pressure on sub-national governments to invest in low-carbon infrastructure (both social and physical); (3) local carbon budgets set by higher government levels needing to be met; (4) engagement in carbon trading markets. The role of local governments will therefore increase with increasing carbon control, but that regulatory prominence will bring with it new pressures to achieve regulatory targets with potentially limited budgets.  The justice and equity implications of the ESR of carbon governance are wide ranging. The authors begin by situation carbon governance and ESR in the context of capitalism and neoliberalism. They note that that “[i]f all human societies depend on nature, the production process in capitalism can have devastating consequences for the workers and communities that depend on the particular products of nature for access to basic human needs.” Because the costs and benefit of capitalism accrue unevenly, so do the costs and benefits of environmental damage: “…uneven development is central to the context and form in which social labour appropriates nature.” Furthermore, resource-destructive activities create “…obstacles in particular places and regions for further rounds of capital accumulation, and opens up opportunities to externalise social costs and externalities onto others (e.g. workers, residents and taxpayers in other cities and regions).” Distributional injustices in space and time are an inherent part of capitalist production, 147  say the authors, therefore addressing social and environmental justice issues explicitly in policy is paramount. All the more reason, then, to understand how carbon governance has been coopted into neoliberal policy-making and the effects that has on governance and equity.  Specific to carbon governance and Kyoto, by assigning Annex One countries legally binding emissions targets and others not (as part of ‘contract and convergence’ policies), notions of global equity are built in to the Kyoto accord. Understanding that the most vulnerable are the most exposed to climate change impacts, there is a moral imperative established, and so regulating and pricing carbon can help to redistribute costs and benefits. This redistribution might even challenge the neoliberal market economy itself. However, some policies, such as use of nuclear power or championing biofuels might have costs deemed too high to bear. Offsets, such as those in the Clean Development Mechanism, might allow rich countries to effectively outsource their emissions to developing nations, potentially “… narrowing development pathways in the process.” There is also concern that in the pursuit of emissions reductions goals that equity considerations will be addressed in name only, or that authoritarian governance and state surveillance will be used to reward compliant carbon users, while punishing others. At root, there is a distributional equity problem that must be considered when national governments downscale climate governance responsibility to other jurisdictions that has three elements: “1. the distribution of responsibility for emissions reduction through instruments such as territorial carbon budgets, 2. the strategic and spatial selectivity of investment and resource allocation linked to climate policy and 3. the regulatory and fiscal mechanisms employed to regulate carbon use." The authors caution that "… as the politics of carbon control intersects more closely with the politics of urban and regional development, questions of social and spatial justice are likely to come forcefully into play."  56. Wilkinson, C., Saarne, T., Peterson, G. D., & Colding, J. (2013). Strategic Spatial Planning and the Ecosystem Services Concept- An Historical Exploration. Ecology and Society, 18(1). http://doi.org/10.5751/ES-05368-180137  Wilkinson et al. compare strategic spatial plans from 1929-2010 for both Melbourne and Stockholm to compare how ES have been treated in each city through time. The authors 148  investigated both “…the types of ES taken into account, and how human-nature relations and the valuation and trade-off discussions regarding ES were framed." An ES coding scheme, based on various Millennium Ecosystem Assessment categories, was used to record ES in each plan. By using an ES list created before assessing the planning documents, the authors could assess which ES were included and, importantly, which were left out. A modified version of this same scheme was used to conduct a post-hoc recoding of ES in this literature review to provide a common baseline for comparison of the journal articles.  The authors found that only “… two ES (of the 39 ES coded for) are addressed in every plan for both cities, namely “freshwater” (B4) and “recreation” (D6)... [though] for Melbourne “aesthetic values” (D4) is also referenced across all plans." Four ES were not addressed in any of the strategic plans: primary production, disease regulation, inspirational and spiritual and religious values. About two-thirds of all ES are included in at least one plan. Monetary valuations of ES are not developed for ES, but the trade-offs arising from land use conflicts are, dating back to some of the earliest plans. Spatial planning in each city does not consistently address various ES, and could benefit from new tools approaches such as: "the consideration of bundles of ecosystem services based on social-ecological factors… and tools such as InVEST (A tool for Integrated Valuation of Ecosystem Services and Tradeoffs) which allow users to fairly easily compare the ES produced in different landscapes.” The lack of consistent ES consideration through time is an example of how individual services can be prioritized over others, as well as how time-scale mismatches are not adequately addressed in planning documents.  The planning documents did frame ES in terms of inter- and intragenerational equity, especially with regard to recreation and aesthetic ES. Melbourne plans also identified that poorer people have differential ES needs than wealthier citizens even in the plans’ earliest incarnation in 1929. The 2002 plan also addressed distributional justice issues noting that parkland is unequally distributed in Melbourne and giving underserved locations investment priority for new open space. The plan also notes that addressing distributional equality is “…. not a case of trade-offs between ES but rather a matter of redressing historical inequities of access." In Stockholm, which has more equally distributed green space, the focus in more on intergenerational equity and “…the means of access to green spaces rather than their geographical distribution per se."  149  57. Wolch, J. R., Byrne, J., & Newell, J. P. (2014). Urban green space, public health, and environmental justice: The challenge of making cities “just green enough.” Landscape and Urban Planning, 125, 234–244. http://doi.org/10.1016/j.landurbplan.2014.01.017  Wolch, Byrne and Newell conduct a literature review to compare how U.S. and Chinese cities engage in urban greening, the environmental justice implications of urban greening initiatives and how various jurisdictions account for them. The authors note that most urban green space studies reveal unequal green space distribution, with the more affluent or otherwise advantaged (for instance due to race), generally reaping a disproportionate share of urban green space benefits. A key finding of their review is that efforts aimed at redressing green space injustices may have the opposite effect due to gentrification and displacement. Green space can increase property values and rents in previously less-desirable communities and displace the very people the green space was meant to help – potentially pushing them into neighbourhoods with fewer ES benefit flows and more health risks.  The proposed cure for this paradox is to “… focus on urban green space strategies that are ‘just green enough’ and that explicitly protect social as well as ecological sustainability." In order to do so, urban planners and green space designers must consider procedural equity – in other words, to “… design green space projects that are explicitly shaped by community concerns, needs, and desires rather than either conventional urban design formulae or eco-logical restoration approaches." A definite test or suite of tests for justice were not put forward, nor was a comprehensive framework for assessing ES justice. However, myriad justice issues arose out of the literature review, many of which have been touched on or investigated by the articles above. Current environmental injustices in the US were noted to stem from historical ethno-racial injustices and oppression. Health risks of lack of green space and of the need to travel to green space were also highlighted, including the negative health correlations of displacement and precarious housing that green gentrification can exacerbate, as well as pollution exposure in areas where active forms of transportation are advocated without “…commensurate efforts to reduce levels of air pollution.” In China, environmental injustice was also related to socio-economic status as well as 150  challenges inherent with rapid urbanization and high population density. There were other concerns too, including the lack of benefits for non-official residents of Chinese cities and the lack of procedural justice for Chinese citizens. The authors note that “…citizen participation in decision-making is limited, as are avenues for raising formal complaints about environmental protection and management… residents also fear that complaints will bring reprisals or persecution.” Park use is also highlighted as a justice issue, with the needs of local residents needing to be factored in to green space design and siting.   58. Wong, T. H. F., & Brown, R. R. (2009). The water sensitive city: principles for practice. Water Science & Technology, 60(3), 673. http://doi.org/10.2166/wst.2009.436  Wong and Brown describe the ‘hydro-social contract’ that is needed for implementation of the ‘Water Sensitive City.’ The contract “... is a term used to describe the pervading values and often implicit agreements between communities, governments and business on how water should be managed." The contract is shaped by cultural frames and narratives, the historical context of water values, institutions, regulation and the physical infrastructure of water provision. The authors also note that artifact creation, as described by Actor Network Theory, plays in role in how the hydro-social contract arises: "... research revealed that the development of [water sensitive urban design] across Melbourne has been the result of a complex and sophisticated interplay between key champions (or change agents) and important local context variables." The authors describe a Water Sensitive City as a resilient system, where disturbances are not only accommodated but used as opportunities for innovation.  Environmental and social justice are not a focus of Wong and Brown’s paper. They include intergenerational equity in a figure of the elements of the Water Sensitive City. They also allude to intergenerational equity when they assert that “… the major challenge is to develop governance systems that make it possible to relate to environmental assets in fashion that secures their capacity to support societal development for a long time into the future.” However they also advocate for a “socio-technical perspective,” rooted in the physical realm “… as the most promising for addressing the need for resiliency and advancing sustainable development.” This is 151  a position which is almost in opposition to many of the justice and equity perspectives outlined above by seeming to propose an expert-led, positivist version of science-based management.     59. Yao, L., Liu, J., Wang, R., Yin, K., & Han, B. (2014). Effective green equivalent—A measure of public green spaces for cities. Ecological Indicators, 47, 123–127. http://doi.org/10.1016/j.ecolind.2014.07.009  Yao et al. propose “… a metric of effective green equivalent (EGE), which is defined as the area of green space [in a city] multiplied by corrected coefficients of quality and accessibility.” The authors develop two city-wide indicators from the EGE: the average EGE for all urban residents, and the inequality coefficient of the EGE, measuring the equality of EGE distribution in an urban area. The authors incorporate distance to green space into the EGE and use NDVI values to assess green space quality. The inequality coefficient is based on the Gini coefficient, with income replaced by EGE.   The EGE and its two indicators are applied to the case study city of Beijing, China. Their results show that EGE values per person are normally distributed, with an average value of 355.49ha per person, and an inequality coefficient of 0.24. Private green spaces are rare in Beijing, and so resident rely on public green space provision. Despite this, it was found that "...68.2% of city residents in Beijing can access a moderate level of green space." The authors used the inequality coefficient to apply a quantitative limit on equity, with values “… under 0.4 approaching 0… considered increasingly fair.” It was noted that quality and accessibility affect the ES actually accessed by residents, with the EGE incorporating these elements to develop a more comprehensive picture of how green space is distributed in cities.   60. Yin, H., Song, Y., Kong, F., & Qi, Y. (2007). Measuring spatial accessibility of urban parks: A case study of Qingdao City, China. In Geoinformatics 2007 (p. 67531L–67531L). International Society for Optics and Photonics. Retrieved from https://apps.webofknowledge.com/full_record.do?product= UA&search_mode=AdvancedSearch&qid=15&SID=2Bw5iVlCLEa8hfJJc65&page=13&doc=127 152   Yin et al. measure park accessibility using geospatial techniques in Qingdao, China. The authors argue that the “… Level of access to urban parks is an important indicator of the effectiveness of their equitable provision.” Container and minimum distance methods were used to measure park accessibility. The container method yielded poorer (less accessible green space) results than the minimum (Euclidean) distance method. The container method was also less accurate. The authors noted the important of identifying park entrance points and not merely boundaries, however, it was not clear if the authors based their minimum distance results on entry points or not. A 1000m threshold between good and poorer access was used, calculated for each household in the study area. Percentages of very good (<500m), good (500-1000m), poor (1000-200m) and very poor (>2000m) access were then reported. The thresholds used were arbitrary, and considering the substantial difference in green space standards among countries, don't tell the reader much about the amount or quality of green space that residents have access to.  The basic standard for per capita green space in China is 6-8m2, whereas in the UK and US, common standards are 24 and 41m2 respectively. However, the vast majority of homes had very good access (64.04%) or good access (28.11%), with substantially fewer residents having Poor (7.69%) or very poor (0.16%) park access. The authors assert that their results show that there is a fairly equitable distribution of parks in Qingdao, with good access to green space at the household level.  Despite Qingdao’s relatively equality, Yin et al. note that there is an unequal distribution of parks worldwide, in part due to the use of per capita allocation standards without regard to parks’ spatial distribution. This unequal distribution is an environmental justice issue, impacting “quality of life” for urban residents. The authors note that accessibility studies had not been done for Chinese urban parks at the time of writing, nor indeed was it common practice to evaluate the accessibility of most other urban amenities, including schools and hospitals. This lack of accounting for distributional justice by Chinese planners, the authors argue, aggravates social inequities.   153  Appendix C  Object features exported for each type of object used for classification. Object features of Tree Canopy and Unclassified objects were predictor variables for RF. Predictor variables were separated into RapidEye-derived (spectral), LiDAR-derived (LiDAR), and geometric (including some contextual metrics) predictors. Class key: Buildings = 1, Trees = 2, and Unclassified objects = 3. Information on individual eCognition metrics can be found in the eCognition Reference Book (Trimble 2012a, 10–18) and on the eCognition Community Wiki (eCognition Community Members 2009).  Object Features Export Classes Formula/Description Spectral   GLCM Contrast NIR 1, 3  GLCM Homogeneity NIR 1, 3  Image Brightness 1, 2, 3 ([Mean NIR]+[Mean Red]+[Mean Green])/3 Mean Blue 1, 2, 3  Mean Green 1, 2, 3  Mean NIR 1, 2, 3  Mean RE 1, 2, 3  Mean Red 1, 2, 3  NDRE 1, 2, 3 ([Mean NIR]-[Mean Red Edge])/([Mean NIR]+[Mean Red Edge]) NDVI 1, 2, 3 ([Mean NIR]-[Mean Red])/([Mean NIR]+[Mean Red]) NDVIRE 1, 2, 3 ([Mean Red Edge]-[Mean Red])/([Mean Red Edge]+[Mean Red]) NIR:RE 1, 2, 3 [Mean NIR]/[Mean Red Edge] SAVI 1, 2, 3 ((1+0.5))*(([Mean NIR]-[Mean Red])/([Mean NIR]+[Mean Red]+0.5)) Standard Deviation Blue 1, 2, 3  Standard Deviation Green 1, 2, 3  Standard Deviation NIR 1, 2, 3  Standard Deviation RE 1, 2, 3  Standard Deviation Red 1, 2, 3  Visible brightness 1, 2, 3 ([Mean Blue]+[Mean Green]+[Mean Red])/3   154  Object Features Export Classes Formula/Description LiDAR   GLCM Contrast nDSM 1, 3  GLCM Homogeneity nDSM 1, 3  Coefficient of Variation nDSM 1, 2, 3 [Standard deviation nDSM]/[Mean nDSM] Max Height - Min Height 1, 2, 3 [Max. pixel value nDSM]-[Min. pixel value nDSM] Mean nDSM 1, 2, 3  Mean nDSM slope 1, 2, 3  Mean zDev 1, 2, 3  Standard Deviation nDSM 1, 2, 3  Standard Deviation nDSM slope 1, 2, 3  Standard Deviation zDev 1, 2, 3  Geometric   Relative Border to building 1, 2, 3  Relative Border to trees 1, 2, 3  Relative Border to unclass 1, 2, 3  Border index 1, 2, 3  Compactness 1, 2, 3  Density 1, 2, 3  Elliptic Fit 1, 2, 3  Length: Width 1, 2, 3  Rectangular Fit 1, 2, 3  Roundness 1, 2, 3     

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0363005/manifest

Comment

Related Items