Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Ground penetrating radar applied : a model for quantifying interpretation of human burials in historical… Daniel, Stephen Edward 2015

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

Item Metadata

Download

Media
24-ubc_2015_may_daniel_stephen.pdf [ 2.53MB ]
Metadata
JSON: 24-1.0166116.json
JSON-LD: 24-1.0166116-ld.json
RDF/XML (Pretty): 24-1.0166116-rdf.xml
RDF/JSON: 24-1.0166116-rdf.json
Turtle: 24-1.0166116-turtle.txt
N-Triples: 24-1.0166116-rdf-ntriples.txt
Original Record: 24-1.0166116-source.json
Full Text
24-1.0166116-fulltext.txt
Citation
24-1.0166116.ris

Full Text

   GROUND PENETRATING RADAR APPLIED: A MODEL FOR QUANTIFYING INTERPRETATION OF HUMAN BURIALS IN HISTORICAL CONTEXTS   by   Stephen Edward Daniel  B.A, The University of British Columbia, 2008      A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  MASTER OF ARTS  in  The Faculty of Graduate and Postdoctoral Studies  (Anthropology)    THE UNIVERSITY OF BRITISH COLUMBIA  (Vancouver)   April 2015    © Stephen Edward Daniel, 2015 ii   Abstract  This thesis explores the applied use of Ground Penetrating Radar (GPR) technology in conducting comprehensive burial survey work in historic period, post-contact cemeteries. These results are based on research conducted from 2008 to 2011 within several First Nations post-contact cemeteries along with work done in other less-defined burial sites in Southwestern British Columbia. My research has been informed by other types of GPR surveys conducted during that period and through 2015, that have added experience in GPR project management, data collection, trace signal analysis, interpretation and reporting. I have developed a recursive, interpretive model that creates a simple, direct and more objective process for evaluating any single or group of potential burial locations situated in a variety of physical contexts. The basic analysis is done by linking GPR signal results obtained from over 300 cases with either prior historical or field-work based knowledge of related ground surface, physical, ethnographic and documentary evidence. The model developed here quantifies the interpretative analysis of these data and develops what I refer to as a Burial Confidence Index (BCI) from a set of parameters or variables that reflects the full extent of our knowledge of any specific location. This allows for testing and statistical comparison of known versus previously unknown locations where GPR evidence is recovered. Other important aspects of GPR-related work in the community are also addressed in brief to provide more complete coverage of the many contexts involved, including professional, academic and social considerations. iii   Preface  This thesis is an original intellectual product of the author, S.Daniel. The fieldwork reported in Chapters 1-4 was conducted primarily within the 2008 to 2010 Musqueam-UBC Archaeological Field Schools led by Dr. Andrew Martindale (UBC Anthropology). The remaining research was extension of that field school environment into projects completed for Kwantlen First Nation. My role in the research program was as the primary technical and project manager, and as lead investigator executing and supervising field and survey work. In addition, I was responsible for the data analysis and reporting phases of the research and for presenting an overview of the results to Musqueam and Kwantlen First Nations.  iv   Table of Contents  Abstract ....................................................................................................................................... ii Preface ........................................................................................................................................ iii Table of Contents ...................................................................................................................... iv List of Tables............................................................................................................................... v List of Figures ............................................................................................................................ vi Acknowledgments .................................................................................................................... vii Dedication ................................................................................................................................ viii  1  Introduction ............................................................................................................................. 1  2  History and Effectiveness of Ground Penetrating Radar ................................................... 6  3  GPR Methodology, Technical Parameters and Research Record ................................... 11  4  Additional Contexts to GPR Survey .................................................................................... 17  5  Burial Confidence Index (BCI) Model .................................................................................. 18  5.1  The Burial Confidence Index Model ........................................................................... 19 5.2  General factors influencing BCI results ...................................................................... 26 5.2.1  Surface characteristics and features .................................................................. 26 5.2.2  GPR signal metrics and morphology .................................................................. 26 5.2.3  GPR signal characteristics ................................................................................. 27 5.3  Surface characteristics influencing BCI results .......................................................... 28 5.3.1  Grave marker presence ...................................................................................... 28 5.3.2  Grave marker anchoring ..................................................................................... 28 5.3.3  Row structure ..................................................................................................... 29 5.3.4  Row spacing ....................................................................................................... 29 5.3.5  Presence of other surface features .................................................................... 30 5.3.6  Availability of other historical records ................................................................. 30 5.4  GPR signal metrics influencing BCI results ................................................................ 31 5.4.1  Signal depth ........................................................................................................ 31 5.4.2  Signal dimensions (area, length and width) ........................................................ 31 5.5  GPR signal characteristics influencing BCI results .................................................... 33 5.5.1  Plan view signal intensity and profile view stratigraphy ...................................... 33 5.6  GPR burial analysis model calculation framework ..................................................... 33  6  BCI Model Results and Analysis ......................................................................................... 40  7  Conclusions .......................................................................................................................... 49  References Cited ...................................................................................................................... 51  Appendices ............................................................................................................................... 53  Appendix A: Coded Burial Location Identification Data .............................................................. 53 Appendix B: Coded Burial Location BCI Results Raw Data ....................................................... 58 Appendix C: Sample GPR Trace Signals ................................................................................... 62 Appendix D: GPR Project Management - Sample Grid Information ........................................... 64 v   List of Tables  Table 5.1     Potential burial location geographical and identification data ........................... 20 Table 5.2     Burial Confidence Index Test Groups – Location Aggregations ....................... 20 Table 5.3     General factors influencing BCI results ............................................................. 26 Table 5.4     Burial Confidence Index component variables .................................................. 27 Table 5.5     Overall weighted contributing factors influencing BCI ...................................... 35 Table 5.6     Weighted surface characteristics influencing BCI ............................................. 35 Table 5.7     Weighted signal metrics influencing BCI ........................................................... 36 Table 5.8     Sample burial locations showing BCI calculations (Excel model) ..................... 38 Table 5.9     Sample burial locations showing variables & BCI weightings ........................... 39 Table 6.1     BCI Model Results Musqueam Known/Marked Locations ................................ 40 Table 6.2     Mean GPR Signal Dimensions: Musqueam Known Locations ......................... 42 Table 6.3     BCI Model Results Kwantlen Known/Marked Locations Comparison .............. 43 Table 6.4     Mean GPR Signal Dimensions: Kwantlen Known Locations ............................ 44 Table 6.5     All BCI Results by Test Data Group .................................................................. 45 Table 6.6     Box and Whisker data for known and unknown burials .................................... 45  vi   List of Figures  Figure 1.1    Graduate students conducting GPR fieldwork, Kaukana, Sicily ......................... 3 Figure 1.2    Kwantlen Main Cemetery, Typical Survey Area ................................................. 3 Figure 3.1    GPR Plan View (Top Down) Output ................................................................. 13 Figure 3.2    GPR Profile View (Side View) Output ............................................................... 13 Figure 3.3    Sensors & Software pulse EKKO Pro GPR Equipment Array .......................... 15 Figure 3.4    GPR field data collection technique .................................................................. 15 Figure 6.1    Box and Whisker plot of Known and Unknown Burials ..................................... 46      vii   Acknowledgments  I am indebted to many who helped me toward concluding this thesis and my UBC research program. First and above all others for his unfailing patience, guidance, encouragement and gentle prodding in the face of my professional non-academic pursuits, I am most grateful to my supervisor, Dr. Andrew Martindale. We are quite simply friends and colleagues I am proud to say. Alongside Andrew, my professors, the essential Dr. Sue Rowley who has fostered my often romantic reasons for doing this and acted as committee member, reviewer and guide, and Dr. Michael Blake who has taught me much. Dr. David Pokotylo acted as statistical mentor and guide and his commentary and ideas were invaluable to completion of my thesis. Together they have endured the many intervals of forced inactivity before reaching the end point. Their patience while I carried out my full-time profession in the professional sports industry satisfying clients who have pushed far harder and with much less understanding, is legend. In addition, Ken Stark of Kwantlen Polytechnic University, Roger Wilson, Bruce Miller and Patrick Moore of UBC encouraged my approach to archaeology at all levels and I am truly thankful for their teachings.  This project has been done with the complete support of Musqueam First Nation Chief and Council, and especially by Treaty Director, Leona Sparrow. Her personal support and trust along with that of many Musqueam community members has been the foundation for my work at UBC. It has always been a privilege to work and live along the Fraser River on Musqueam traditional territory - and to be an inhabitant of Marpole both in space and time. It has been an honour to work with Kwantlen First Nation community members and I am indebted to Chief Marilyn Gabriel, Lekeyten and Cheryl Gabriel for their help.  My research has benefited from the generous input of many Department of Anthropology students and Musqueam community members who participated in UBC-Musqueam Archaeological Field Schools from 2007 to 2010. Marina LaSalle, Wayne Point, Terry Point, Larry Grant, Henry and Stan Charles, Iain McKechnie and Lisa Dojack added immeasurably to the final product with their insights along the way. I am also grateful for the financial support, grants and scholarships provided through the University of British Columbia and other donors.  My thesis and approach to working out in the field has been informed by the humanitarian science of Dr. Carl Sagan, Dr. Richard Feynman and John Romer. Though they did not know it, they taught me that the whole idea is to have adventure … to try to understand the world we live in a little bit better than we do, and to examine our lives both ancient and modern, through very different lenses. My son Michael Daniel continues to remind me of the value of their way of thinking, and Mike’s ideas, love and enthusiasm inspire me. Our shared and collectively odd view of the world is based on an active acceptance of multiple beliefs about or ways of “seeing” things and lies at the core of my academic perspective. And finally, my dear mother would have been proud to see this all happen despite my incredibly slow start and halting attempts at becoming myself. Thankfully, Mom’s acceptance of my differences was unconditional, unwavering and came at no cost.   viii   Dedication  I would like to dedicate this thesis, and the very fact I was able to return to UBC at all, to my amazing wife Carol. Her endless and often undeserved support, understanding and belief in me, along with her steady encouragement and general silliness at just the right times made it bearable to go down parallel, competing professional paths at once.  Carol added that one missing ingredient that resulted in the self-belief I was never quite able to create until she entered my life. Above all others, she simply understands at all times that it is okay to be different when no one else before her stopped to ask or understand. Hopefully, I can return all of that love and inspiration in full someday … but there seems little hope of that.1  1 Introduction  The theme for this project arose from a deep personal and scholarly connection to human burials on First Nations grounds that emerged out of field school work with Musqueam First Nation and UBC’s Laboratory of Archaeology.  This is further grounded in a close, personal attachment to the geographical area within which this thesis project has been conducted – Marpole, Lower Fraser River Delta and the Gulf Islands.  This involvement and came about through application and teaching of ground penetrating radar (GPR) skills as a non-invasive archaeological method; and helped to expand my interactions with several other First Nations including the Kwantlen and Penelakut. The primary technique developed in this thesis is that of a quantitative model that creates a more objective method for evaluating whether GPR signals relate to human burials in historical, post-contact cemeteries. In addition to the model’s results, this project’s relevance lies in providing guidance to cemetery owners and descent communities regarding the present state and future use of any space within cemetery grounds. The methods developed here have utility beyond the formal burial contexts discussed here; these methods can be applied to the analysis of any potential burial location both within and outside formal cemetery contexts. In designing the model, several specific potential applications became apparent, particularly those involving contributions to litigation proceedings and forensic analyses conducted by law enforcement agencies. The model is discursive and tests the sensitivities, factors and assumptions that are most relevant in determining what are the qualities of GPR signals that are most indicative of human burials in cemeteries? The GPR work discussed here has been carried out jointly and co-operatively with First Nation community members in the field with the goal that of a better-understood cemetery 2  or burial ground. In my assessment, non-invasive archaeological techniques such as GPR technology can aid in developing a better overall understanding of cemeteries; not merely the state of sub-surface burials found there, but also their histories and future use. When used as an additional tool beyond documented historical knowledge, i.e., the surface indicators located in the cemetery along with associated maps and diagrams, GPR becomes a significant discovery and planning tool. In my experience working in several local First Nations historic period, post-contact cemeteries, formal burial grounds often contain a range of arrangement patterns (reflecting history of use), the history of which is not clear, in part due to the disappearance of original surface markings. Figure 1 and Figure 2 show examples of spaces where GPR can be used effectively to provide sub-surface information. Figure 1 depicts a survey inside a house structure, and Figure 2 shows a section of open cemetery ground that was a source of the data for this thesis. They are each open with good access and over level ground. Good GPR survey can provide the missing structure and patterns in the grounds as a way of understanding both the history of the space and the future potential use.    Figure 1.1 Graduate students conducting GPR Figure 1.2 Kwantlen Main Cemetery Fieldwork (Kaukana Project: 2009).   Maple Ridge, BC. Typical Survey Area. Photo credit: Carol Wright      (Used with permission, Kwantlen First Nation)  3  There are several essential questions implicit in such GPR research projects, often reflecting the very tasks we are assigned to perform. My approach in surveying cemetery grounds, and the specific spaces found there, has been to pose the following questions:  1)  Does the space, marked or unmarked, have an associated burial? 2)  Is this space available for current or future use? 3)  In any cemetery, is the apparent row or other structure observed during survey original or has it been reconstructed to appear orderly? 4)  Are there un-marked burials in this cemetery and if so where? 5)  How can we know if this a human burial?  Ground penetrating radar is able to begin to answer these questions due to the evolution of the technology itself in recent years.  Alongside the practical aspects of providing professional advice to cemetery owners, in terms of research contribution for GPR, my goal here is to refine an approach to data analysis that accounts for the full suite of available knowledge. This extended approach has been done in an interactive, synergistic way that draws on the local community’s knowledge in a respectful manner, and yet adds in a recursive fashion greater “confidence” in our determinations as to whether any set of signals is in fact or is likely to be a human burial. The research goal of this thesis is to create a model for testing “confidence” in the interpretation of GPR signals of potential human burials within historic period cemeteries and other contexts. It is founded on establishing a set of known measures for burials against which we can test the unknown and make objective comparisons. From a total of more than 300 individual locations surveyed in 2008 and 2009 (supported by additional fieldwork and GPR analysis conducted from 2010 to 2014), I have created an objective way to interpret sub-surface GPR burial imagery by analyzing and quantifying each location across a range of variables covering both surface characteristics (if any) and their unique sub-surface GPR trace signals in multiple dimensions. In this project I compare known 4  burials (those with surface characteristics and GPR signals) to previously unknown burials (those with only GPR signals) as a means of assessing the utility of my model. I conclude that the Burial Confidence Index (BCI) is an effective means of assessing how well a GPR signal represents a human burial. My field use of ground penetrating radar and analysis of data over a wide range of projects and contexts leads me to conclude that GPR is a very effective though not infallible archaeological method in the context of human burial analysis. In my opinion, precise signal trace interpretation is very much a mix of careful observation, pattern recognition and scientific methods, alongside a kind of connoisseurship or evaluative capability based on depth of experience by the GPR operator and data analyst. The technical aspects of GPR are well-understood and supported by a considerable volume of literature and instructive operating guides. That aspect of GPR research is outside the scope of this paper but understanding the technical elements has formed the foundation of my research and field capability in data collection and analysis. GPR hardware and other technical issues aside, generating “good” data is both very straightforward yet challenging at times due to common physical constraints. However, my conclusions on two further aspects of GPR, that of actual project management and signal interpretation, are much more germane to my overall findings. In preparing for fieldwork and data collection, I found that GPR project management had limited documentation; very little existed as to exactly how to go about designing and executing a GPR project in the field. Through an initial process of “trial and error”, conducted over a wide variety of physical settings and time constraints for precious field time, my experience has led to a fairly refined model for data capture that supports post- fieldwork interpretation without the need to return to the survey site. This project management model has been documented in a supporting manual to this thesis. 5  The most important set of conclusions I have made are related to GPR signal interpretation. Having examined thousands of GPR signals in both plan and profile, I have determined that there will always remain an unavoidable ambiguity in drawing precise determinations for a large proportion of GPR signal patterns. I conclude that it is more often the case that we are less than 100% certain of exactly what we are “seeing” from the sub-surface via GPR. This is due to several factors discussed below, and the best we can do is to derive a “level of confidence” rather than an unambiguous “YES” or “NO” to the questions posed initially above. The degree of variability in signal shape, intensity and amplitude, viewed stratigraphically against a background of sub-surface “noise” is a major factor in any identification – thus necessitating development of the core model of this thesis. I conclude that with careful analysis at the lowest level of empirical observation (i.e., a set of detailed variables summed mathematically and indexed), an objective comparison can be made for any set of data to known burial GPR standard traces. GPR interpretation and analysis, therefore, requires quantification to be effective in allowing for confident predictions to be made. This is especially true since it is seldom possible to “ground truth” our results. Finally, I have concluded that the further that GPR research departs from well-structured historical cemetery grounds and out into open, less-defined areas and contexts, the more difficult and challenging our interpretation exercise becomes.  6  2  History and Effectiveness of Ground Penetrating Radar  The use of GPR equipment and its associated software is an emergent archaeological specialty, which has added new, non-invasive methods for exploring the sub-surface and its relationship to ground-based evidence for burials. GPR has been directed primarily toward commercial applications since its development as early as the 1950s when defense radar was found to be useful for measuring depth below surface from aircraft reconnaissance flights (Conyers 1994). To date, its use in construction and the utility industry remains its principal application in the field. However, a wealth of published ground penetrating radar scholarship for use in archaeology now exists and this serves as a guide for effective use of the technology in many specific contexts. GPR has been used as an archaeological tool since the late 1970s (Conyers & Cameron 1997: 18) but in recent years has developed rapidly with the addition of related digital computer applications for data collection and processing. The most consistently identified goal for GPR applications has long remained increasing our understanding of the sub-surface without the need for expensive, potentially damaging and time-consuming excavations. So by definition, GPR is seen as a non-intrusive methodology in its execution in particular, but within my experience from other social contexts, this is not entirely true (Conyers 1997: 3). GPR is generally regarded, however, as the most effective of the geo-physical technologies due to its often clear sub-surface imagery and ease of use (King et al 1993: 4). Most researchers in the field are able to collect data quickly and link the data to GIS and other geographical systems to precisely locate information in space. A review of the literature indicates that the most common objectives for GPR-related archaeological survey ranges from location of large-scale architectural remains to sampling small-scale features and potential burials. Much of the latter involves GPR-related forensic 7  investigations and other buried remains (Schultz and Martin 2010: 64). As a result, ground penetrating radar methods and technology can be applied at differing scales but with equal effectiveness in creating a view of the sub-surface landscape and stratigraphy, essentially going back in time over the same space. During my review of the literature on GPR-related studies, the vast majority of published reports most frequently focus on identifying specific, buried features such as walls, houses, crypts and buildings and very infrequently on conducting forensic studies of the landscape at high resolution which is the emphasis of my thesis (Schultz and Martin 2011: 64-65; Conyers 2004: 18). In my experience, GPR equipment can be applied in quite localized ways to develop interpretations for the likely meaning or source of sub-surface disturbances that stand out within the surrounding soil matrix. This aspect of GPR use has found particular application within some cemetery contexts and to a lesser degree in geographically remote, non-formalized burials. These spaces lack the normal cemetery surface features such as grave markers, row structure, headstones and accompanying ethnographic records.  King et al (1993: 4) state that with a potential array of limitations such as the lack of historical documentation or the loss of original burial markings and other ground surface disturbances, “ a low success rate for finding unmarked graves with GPR is a reality“. France et al (1992: 1458) take opposite more positive view, concluding “GPR offers the most useful tool to delineate possible graves”. Doolittle and Bellantoni (2009: 941) suggest “performance shortcomings and unproductive field time have produced some cynicism toward the use of GPR in forensic investigations”. Overall, Conyers (2006: 71) points to two specific features from historic graves as being particularly appropriate for GPR analysis: reflection hyperbolas from burial containers and shaft truncation planes. GPR analysis of depth profiles can be very effective in highlighting these features as highly reflective anomalies are easily identified in the imagery. 8  The body of published research is, therefore, somewhat divided on GPR’s usefulness due primarily to the extensive influencing factors and the effectiveness of the technology. Conyers (2010:1) suggests that often very limited data collection and post-processing options are applied along with the absence of post-survey verification by ground-truthing (i.e., excavating). Conyers & Leckebusch (2010: 2) add that “GPR failures were quickly forgotten … and in contrast, successes were considered an important reason to publish”. Conyers (2004: 168) comments further: Many in the archaeological community may continue to employ GPR only as an “anomaly-finding device” to locate possible features that can later be excavated … in the future GPR’s maximum effectiveness will be when it can be integrated with detailed archaeological and geological information collected from excavations and stratigraphic studies.    This sentiment echoes the mode in which my research has been constrained in that for the most part it has been limited to “anomaly-finding” only and subsequently analyzing what can be observed in the plan (top down) and profile (side-view) imagery. However, though we are clearly limited in this way, this thesis challenges Conyers’ assertion to a degree and seeks to refine our interpretive skill set through statistical analysis of an extensive local cemetery dataset (Conyers 2010: 83).  Here I integrate all the results from my own research to create an analytical tool for wider use – a template for interpreting any set of GPR signals with specific numerical values, and with a rigorously produced and ultimately defendable “confidence” value. The nature of the context in which this research takes place, that of the survey of formal cemeteries, naturally limits follow up verification through more traditional archaeological techniques. Interpretation and decision-making can, therefore, be tested not by ground-truthing, as may be expected from less personal or contentious settings than cemeteries, but by comparing results to a wide range of signal traces from several hundred locations of 9  both known and potential burial signals. It is thus possible to create a compilation of GPR observations associated with known burials as a template for analysis that adds objectivity to our assessments of unknown cases. The evaluation of the latter in the context of the former can act as a recursive test to the algorithm of compilation. This allows for a more explicit assessment of reliability in GPR analysis of burials. Many other GPR-related considerations need to be addressed, and this is further support for using a “confidence level” estimate based on a composite of observations rather than a single judgmental measure. For example, burials occur at varying depths, many different types of containers having different GPR signatures are used, and they often have degraded GPR signals due to the length of time since the burial occurred. Even though a location may be well-marked at the surface or fit within a row structure or surface order, GPR results can be ambiguous at times. Sub-surface rocky formations and root masses may appear to present themselves in a similar aspect and as well defined as burial containers. These natural formations often produce an ambiguous signal through scattering, especially when using frequencies of 400 MHz or greater (Novo et al 2011: 137). In conducting post-data collection analysis and interpretation, my research has shown that a wide range of GPR signal types and subtleties appear and this requires a considerable level of interpretative skills. These are often subjectively applied and this influences the effectiveness of the tool in the field and assessment of the results. In the available GPR burial literature, only mixed results have been achieved by GPR methodology for burials and many researchers conclude that their detection cannot be guaranteed (Doolittle and Bell 2009:942; King et al 1993). This is especially true when taking into account the standard ground penetrating radar application procedures as part of a larger field project. Many archaeological projects that use GPR tend to use radar survey as a prospection tool followed by a set of selective excavations. That methodology is clearly 10  unavailable for most cemetery contexts and as such, a better method is needed to help answer the fundamental question in using GPR for identification of human burials: how can we know it is a human burial only from the GPR signal? GPR survey data and analysis routines are at their best when integrated with other techniques. In order to develop sub-surface GPR images of sufficient quality to analyze historic period cemetery burials, my experience has shown that antenna frequencies of 400 MHz or greater yield the best results. This assertion has been derived from more than 60 field days of GPR grid placement and analysis of the trace results derived from those surveys. All of the data obtained in this thesis project were collected with a 500 MHz antenna set and thus has allowed for clear image creation down to just over 2.0 meters. This is well suited for use in cemeteries; particularly those whose soil content is primarily alluvial fill. One mitigating factor however, is that with this higher resolution (than say 250 MHz), other in-ground features such as rocks and tree roots appear more prominently and add additional complexity to the analysis of results (Stanger and Roe 2007: 47). Ground penetrating technology works by sending high-energy waves into the ground and upon return to the receiver measures their return intensity and elapsed time (Conyers and Cameron 1998). With a properly calibrated setup, accurate calculation for the depth, size and nature of the anomaly against the background matrix becomes possible. For many archaeological projects the next step is often to excavate based on these initial GPR findings, however, in cemetery contexts this is naturally limited or inappropriate and constrains what we can discover and thus necessitates a modeling approach. GPR results provide for generalized identification only of burials and little detail as to their nature or condition. In this regard, the interpretive process can be very subjective. Thus, this thesis project is designed to integrate all that can be known and quantify it in terms of probability estimates of the likelihood of a GPR signal pattern representing a human burial. 11  3  GPR Methodology, Technical Parameters and Research Record  Theory behind the use and effectiveness of GPR technology has been dealt with at length in the archaeological literature since the 1980s and most clearly by Conyers & Cameron (1998) and Conyers (2004), and other authors for specific, project-related field applications (e.g., Conyers 2011: S13). Those theoretical, technical discussions fall outside of the scope of this thesis, which is focused on applied use in the context of community needs, and refining the interpretive methods and measures used to report results. A brief description of how GPR functions is important however, to describe the kind of work that is involved and the associated field procedures. Ground Penetrating Radar (GPR) is based on the same technology as surface radar systems used in navigation (Conyers & Goodman 1997: 23). A source antenna sends out an electromagnetic wave signal and a receiver antenna collects return signals.  Variation from the background signal creates anomalies or “targets”, a form of echolocation.  GPR systems send a signal into the ground and can map out subsurface variation by locating areas of difference or high contrast between the background soil matrix and signal anomalies.  Such devices are used in industry to locate point targets (such as buried objects), line targets (such as pipes), area targets (such as groundwater contaminations) and stratigraphic patterns (such as bedrock surfaces).  Archaeologists have found that GPR equipment is useful for mapping buried heritage data, and locating human burials though opinion varies as to its effectiveness.  Burials make good targets for GPR because they represent large patterns of difference from the background soil and fill matrices.  In addition, burials often contain metal or other hard surface objects and they are areas of soil disturbance, with higher water concentration and voids (areas with no water). Interments often represent relatively planar excavations for 12  burial shafts that create stratigraphic disjunctions. Each of these qualities can create GPR signal abnormalities when compared to the background. GPR data collection proceeds by sweeping an area with the GPR and creating a map of electromagnetic (EM) anomalies.  In cemeteries, rectangular anomalies found about at 0.5-2.0 m below the ground surface, and of approximately 0.80 to 1.50 m in dimension can correspond to burials.  The two main principles of GPR are that: 1) returned electromagnetic (EM) signals reflect strongly off of some materials (such as metal or salt water) and, 2) the returned signal is slowed by some materials (such as freshwater). Thus, areas in the ground that contain highly reflective materials or those that contain higher or lower concentrations of water than the background soil will produce a visibly anomalous signal.  In essence, the GPR is a very sensitive clock that measures minute time differences (nanoseconds or billionths of a second) between the EM signal that comes back to the antenna quickly by reflection or more slowly by passing through materials, which impede conductivity. Depths are estimated by knowing the velocity of the EM signal through the ground (which can be calibrated) so that the time of the return signal can be converted into distance below ground (Conyers & Goodman 1997: 108). GPR Data collection is done by mapping and covering small rectangular sections (called grids) of a cemetery one small area at a time. In order to collect enough data to locate burials, GPR readings are taken in lines (called traces) usually every 25 cm within the grid.  The project management techniques developed for this thesis tested this line interval, specifically finding that at 50 cm spacing some critical data is lost, while at 10 cm intervals little additional resolution is gained. Therefore, line interval is an essential design component influencing the resolution of sub-surface images.  We collect grid data in perpendicular directions and then use a software mapping program (© Ekko Mapper) to extrapolate the data into a plan view map of the GPR anomalies in the grid.  Grid sizes vary 13  but are about 10 x 10 m square, so a number of grids are required to cover an entire cemetery.  When grids are linked together, a sub-surface map of all the GPR anomalies that represent potential unmarked burials in the entire cemetery can be produced.  When GPR trace signals are found in an unmarked area of the cemetery, an initial finding may be derived for the location of a potential unmarked burial. However, given the complex sources of GPR signal anomalies, positive identification of a burial is usually an informed opinion based on multiple lines of evidence. This kind of interpretation lends itself to application of the Burial Confidence Index Model (BCI) technique for analyzing all of the objective attributes that enable identification and interpretation as either a burial or not.  Figures 3.1 and 3.2 show: 1) plan (top-down) GPR view at variable depths below surface and 2) profile (side view) across a range of depths. GPR imagery is also available in a more "advanced" form as rotatable 3-dimensional "shapes". (See Appendix 3 and Appendix 4).       Figure 3.1  GPR Plan (Top Down) View Output. Figure 3.2.  GPR Profile (Side View) Output.  Notable in Figure 3.1 are rectangles that highlight both clear and obvious GPR trace signals, as well an area in the upper right that is less clear when captured in the print out, but which appears more strongly in on-screen analysis by toggling the depth control. These 14  detailed images provide information on GPR findings for length and width of objects, their approximate depth, shape, degree of reflectivity and much more. The quality of GPR signals is also affected by a variety of factors that influence image intensity, subtleties in observed patterns and strength of signal. This suite of influences contributes substantially to signal attenuation and includes:  Soil properties and moisture content  Overall contrast between buried features, stratigraphic discontinuities and the surrounding soil matrix  Degree of contrast between disturbed and undisturbed soil margins  Burial shaft delineation versus the surrounding matrix  Size and shape of the original burial  Burial age and state of preservation  Original container material (if any)  Original depth of the interment and subsequent surface erosion or soil addition  Signal scattering by rocks, roots or other sub-surface features  GPR equipment survey setting defaults and variables  Many of these conditions are either partially known or entirely unknown, and can cause substantial variations in the intensity of GPR trace signals from any location. These inherent factors, combined with the age of many burials (which are more difficult to locate due to breakdown over time), necessitate development of a more objective means of identification and interpretation of sub-surface signals. The surface geographical and physical context surrounding each site also plays a large role in any GPR-based investigation and this data is also considered in the model designed to test probabilities for any location. Thus far I have discussed subsurface GPR signal qualities for individual potential burials. However, two additional types of observation can influence burial identification: surface features, including memorial markers, and alignment patterns, both of which can be physically present or documented. I discuss these in more detail below. 15  The GPR equipment array used in this research was the © pulseEKKO Pro model developed by Sensors & Software of Mississauga, Ontario and purchased by the UBC Laboratory of Archaeology via TLEF (The Learning and Education Fund) Grant in 2008. Equipment setting defaults were generally set at: 50 ns time window; 0.10 ns sampling interval; 500 points per trace; 0.02 m step size; and 4 stacks. Data collection parameters for all grids placed included: a 100 ns time window; 0.10 ns sampling interval; 1,000 points per trace; 0.05 m step size; and 4 stacks (Dojack 2012). The normal data editing processes were done prior to performing GPR analysis and included line orientation (supported by field note references), line reversals and repositions when required. These procedures along with line gridding were done in © Sensors & Software’s EKKO View Deluxe and © GFP Edit software. Data processing for amplitude slices was done in © EKKO Mapper, and included technical elements dewow (signal smoothing), background subtraction, attribute analysis (envelope), amplitude equalization, and depth conversion. Figures 3.3 and 3.4 represent examples of the standard © Smartcart setup included with the control unit for ease of movement across the ground this serves to speed up the data collection process substantially. An array of this type was used to conduct the research in this thesis.          Figure 3.3  Sensors & Software pulse  Figure 3.4 GPR field data collection techniques. EKKO Pro GPR Equipment Array.   (Photo credit: S.Daniel 2008) (Source: UBC Laboratory of Archaeology   16  Licensed © EKKO Pro Manual, 2008)    My initial GPR research program was established for the 2008 Musqueam-UBC Archaeological Field School. The curriculum was designed to teach basic GPR skills and theory to students, contribute to academic development of the topic, and if possible produce real research results for Band leadership. These objectives were modest with hopes for merely learning how to use the equipment at a basic level and analyzing a small part of the grounds. The 2008 GPR module developed rapidly, however, well beyond the initial vision into development of a comprehensive teaching template/project management guide for conducting large-scale GPR projects with immediate results. This led to a greater understanding of formal burial grounds on the Reserve and the creation of an extensive signal, surface and related ethnographic information database. These results were well beyond our initial expectations. The work was carried out in a collaborative way with Band members helping to collect data and instruct students. These data were then linked to parallel processes for mapping and describing the grounds within existing Band initiatives for GIS and other spatial information systems. Subsequently, the 2008 season’s GPR results became an essential component of the final Field School Reports (Martindale & Daniel 2008a). Other First Nations then became aware of the successful use of GPR authorized by Musqueam First Nation and expressed interest in similar work on their cemetery grounds (Martindale 2009, 2010 & 2011, Martindale & Daniel 2008b).    17  4  Additional Contexts to GPR Survey Work  In my experience, GPR fieldwork is usually a tightly focused task with the time available to gather data often limited by protocol considerations, work permit constraints, poor weather, the need to teach skills in field school or community-based settings, and basic equipment problem-solving. I have found that many other aspects of the work tend to take precedence above basic data collection. In presenting the results from this work at academic conferences and Band meetings, the theme that I have chosen consistently is re-connection between ancestors and the descent community with GPR as the catalyst. The relevance and importance of these results to First Nations descent communities can therefore not be underestimated. With the visibility of the researcher in the field being quite obvious to Band members and others, this suggests that GPR work is anything but non-invasive archaeology when viewed in total. This adds another important consideration beyond why the work is being done – that of how it is being conducted in the community. As a guide to this work, there is a clear obligation to each First Nation and its families for maintaining privacy and respect in both the data collection and reporting phases. Respect for community members and the grounds on which the work is performed are always paramount considerations. For myself, the most satisfying outcome has been to earn the trust and friendship of the community at all points through dialogue and development of an initially clear set of project objectives, methods and collaboration. Finally, it is recognized that permission has been required and obtained to use these data and results without identifying individuals from the burial population.    18  5  Burial Confidence Index (BCI) Model  The GPR statistical model on which this thesis is based is designed not to invent an entirely new way of looking at burials, nor is it to create any especially innovative types of data inspection beyond what is already apparent. The goal is to standardize and quantify each disparate observation into a collective process to achieve a more objective, aggregated evaluation of the qualities that best reflect human burials in historic cemeteries.  Effectively, this exercise expresses what we can know on an easily understood scale acting as an empirical generalization of our thought processes in evaluating a complex set of information and history of physical changes over time. It is an effort to make explicit what is assumed in the interpretive process. In GPR research, we are always limited by physical and social constraints; therefore, the goal is also to create an adaptable and repeatable process for understanding the data presented at any location and across a pool of results.  By applying the same standards in each case, the analysis technique becomes a formalized and consistent approach. Beyond basic cemetery data collection, normally limited to surface manifestations such as markers and rows, GPR software-based signal interpretation can be a critical element of any project where the technology has been applied to supplement or replace traditional archaeological techniques. Good GPR field procedures will collect many other types of surface information in addition to GPR grid-based signal data, as these data types can support and confirm one another.  The contextual data from surface features and memorial markers can assist in the interpretation of sub-surface imagery and wave patterns, particularly for burials of significant age. Indeed, investigating age-related signal degradation effects is possible from such information, though I do not pursue that analysis here.  Surface data may be supplemented by existing historical reference maps and other 19  written records as well as first-hand accounts and oral histories related to any cemetery. It can be the key to creating a more confident interpretation of what appears in the GPR imagery and suggests that prior knowledge and GPR findings can be correlated to create a better understanding (Martindale and Supernant 2009: 203).  5.1 The Burial Confidence Index model (BCI)  In order to evaluate the entire suite of known information about any discrete location, this thesis is based on a model that produces a composite “index” value expressing the likelihood that any single location surveyed is in fact a human burial.  The Burial Confidence Index (BCI) Model was developed on an initial population of data from 288 discrete potential burial locations located within two separate First Nations Reserves (Musqueam and Kwantlen First Nations) and their three major cemeteries. The raw data used in the model is backed up by hundreds of hours of interpretive analysis to determine the GPR-related variable results. In fact, the essential by-product of this analytical component has been to enhance the most difficult aspect of GPR work, one that is not regularly included in the parts list – the capability of understanding and interpreting accurately the meaning of graphic GPR outputs. If there is a single finding that is most apparent after eight years of my work in this area, it is that merely knowing how to operate the equipment is not enough to guarantee a clear set of results and recommendations, nor a comprehensive understanding of the cemetery grounds. Table 5.1 shows the surface data collected for each of 288 different locations at Musqueam and Kwantlen cemeteries. For purposes of confidentiality however, this information is not presented here, but is relevant for assessing their likelihood of being associated with a burial locations.  20  Table 5.1 Potential burial location geographical and identification data.  Location Identification Data Information Included Index # Discrete location ID / confidential code Cemetery Site name GPR Grid ID # Specific gridded area of source GPR data Cemetery/grid coordinates – X/Y In-grid surface location Marker dates If available Burial type Single / multiple Interment date If available from marker (can differ from above)  At the three cemeteries surveyed, known burials were found with only two exceptions to be aligned East-West in recognizable rows. This orientation helped to focus GPR survey efforts and allowed for coverage of larger cemetery areas. The only diversions from that burial alignment pattern were found to be areas of likely reburial after significant surface damage and two specific burials with differing cultural affiliation confirmed by community members. These areas were not covered in the GPR data collection surveys for logistical and other reasons. The survey data and relevant identification information collected for each site data was divided into three major data groups based on their primary location/major surface characteristics noting the presence or absence of GPR signal results:  Table 5.2 Burial Confidence Index Test Groups – Location Aggregations.  Test    Locations GPR Group Cemetery Sites Designation No Marked? Signal? 1a Musqueam “Known”  94 Yes Yes 1b Kwantlen “Known”  35 Yes Yes 2 Musqueam and Kwantlen “Unknown” 159 No Yes 3 Musqueam and Kwantlen “No GPR signal” 15 Yes No   To perform initial testing of the model, the first two major groupings are referred to as “Known” locations, supported by clear surface indicators and/or historical information producing significant and recognizable GPR trace signals. This larger category (see Table 5.2) is sub-divided into 1a) Musqueam locations, and 1b) Kwantlen locations for comparison 21  purposes. Group 2 represents locations found within our surveys that lacked any historical or surface indications but that produced GPR trace signals. These are referred to as “UNKNOWN” locations identified at Musqueam and Kwantlen (from the perspective of what was visible on the ground surface during survey) but were candidates using GPR software analysis. It must be noted that the identification of specific “UNKNOWN” spaces was not random but guided by the fundamental organization of the cemetery grounds including spaces and other gaps in recognizable rows. Specific examples are discrete spaces that fit within a defined row of markers but are not adjacent to other marked locations on either side. These spaces give all the appearance of potential burial locations, but through GPR analysis are determined to be absent of trace signal. A third group of locations at Musqueam and Kwantlen (Table 5.2, Group 3) is listed in Table 5.2 but were not analyzed; those locations where surface markers were present but, with laboratory interpretation, did not show any indication of an associated sub-surface GPR signal. Group 1a therefore includes only marked Musqueam locations previously known or assumed to be burials due to the existence of headstones, enclosures and other markers as well as evident row structure common to historic period First Nations cemeteries. Group 1b includes similar locations within the Kwantlen Main Cemetery in Maple Ridge, BC. GPR software trace analysis was performed on all of these locations referenced in Table 5.2 to ensure that they did not fall into Group 3, those at Musqueam and Kwantlen with a clear surface marker but without an associated sub-surface component of any kind.  For clarification, Group 3 locations are those that are very clearly marked at the surface, typically with small rectangular ground level marker or even large standing monument, but for which no sub-surface GPR traces were found after very detailed survey and analysis at high resolution.  Many of these marked spaces represent memorials only where no associated interment was made, or may be cenotaphs. This was confirmed where possible 22  with community members through on-location discussions. More critically it also covers those markers replaced following acts of vandalism or other damage done in specific areas on the grounds; this was a particular issue at Kwantlen. These markers are frequently unanchored and were often re-placed in a similar row structure to recreate the original appearance of the grounds but without specific knowledge of the original burial space. Some specific Group 3 examples were either known beforehand (through the marker data itself) or identified through the actual survey work done in each cemetery. As such, this necessitated surface and GPR analysis of the entire grounds over several field days and extended the field time required for full site coverage. Small point sources or clearly unrelated GPR traces from other naturally occurring phenomena have already been excluded by initial GPR analysis in the original field reports. Supporting the surface and GPR evidence for confirmation as a burial are other written historical records, rough maps or grid plans along with oral histories provided by community members and Band leadership. Group 1a (Musqueam) is the control dataset against which the fit of any random sample within the group can be tested. The initial step was to compare these data against marked Group 1b Kwantlen burials across the 12 different variables, data for which was assembled by careful project management and data collection techniques within the context of two UBC-Musqueam Field Schools and a separate project conducted for Kwantlen First Nation (Daniel 2009, Martindale 2008, Martindale 2009). Once that was completed, the control data was compared against Musqueam and Kwantlen locations (Table 5.2, Group 2) associated with no surface markers but which showed significant GPR signatures obtained in our surveys. Group 2 locations, those unmarked spaces identified in our surveys but that showed recognizable GPR signals suggesting potential identification as burials, represent the major subset to be analyzed and tested against standards created for Group 1 that would reflect a 23  burial with high probability.  These analytical techniques seek to answer the following question – based on what is known from the qualities and parameters coming from the GPR signatures displayed by Group 1 known locations, what can we conclude from a similar evaluation of any unique group of unmarked locations? This analysis may be performed against any random sample from Group 2 or from a single cemetery or other subset of the data. The BCI algorithm may also be applied to data obtained from other cemeteries to assess the likelihood that any other set of signals represents a human burial. In order to express the BCI index in an easily understood way and conduct basic statistical testing, a scale of 0 to 1 has been used, with a zero result being no likelihood that a location represents a burial, and approaching 100% certainty with a value of 1.00. The BCI values anticipated for Groups 1a and 1b, those that are marked and referred to as “Known” burials, would therefore be assumed to approach 1.00 or 100% certainty.  However, it is vital to recognize that designation must also account for GPR signal attenuation and degradation over time. These natural processes are due to differences in original burial container types and their reflectivity, the natural breakdown of ancestral remains with time, as well as the effect of signal scattering by the soil matrix that varies from space to space. This natural process will result in a range of BCI values obtained across known burial locations below the 1.00 maximum value, but an overall set of parameters may be derived that reflects the full population and which in my opinion would be expected to be well above 0.80 or 80% value, particularly for recent interments. These resulting values may then be compared to any subset of unmarked burial locations as well as tested for the influence of any single factor, for example spacing, monument markings, or fit within observed row structure. In fact, burial age information is available for most known locations and an analysis is possible for testing GPR signal qualities versus any variable.  24  The primary analytical technique is to compare GPR signal metrics and morphology from known, marked burials against the GPR signatures of a group of unmarked, previously unknown locations. These data related to potential burial locations were reported on in 2008 and 2009 following Field School and Band-directed research surveys at four separate First Nations historic period cemeteries (Martindale 2008, Martindale 2009). By developing a profile of known burials quantified over a range of variables that describe their inherent characteristics and diagnostic signatures, a standard can be established to which other sets of GPR signal data can be tested for comparison.  The BCI algorithm compiles both surface and sub-surface data and then aggregates that information into a standard results scale for comparison. By evaluating and comparing the GPR signal qualities that appear from previously unknown locations, we may identify any individual signal set as potentially a burial, and create a value distribution for any sample drawn from the dataset. In cases of known burials, surface and contextual data are both relevant and combined in the BCI calculation (although the contributing effects of GPR and surface data can be parsed). In unknown context, the BCI can be converted to result exclusively from GPR data. The mathematical process of conducting these tests of known versus potential burials is to assign relative weighting for each measurable parameter or GPR signal evaluation. Assessing the relative importance of any single factor is by nature an arbitrary process, but one that may be varied to examine the influence of one versus another. For example, in conducting signal analysis what weighting should be given to the degree of stratigraphic compaction, the depth at which signals are encountered, the intensity of GPR imagery in plan (top-down) view or for the overall impression conveyed by the signals? For each set of GPR signals surveyed, the same evaluative approach has been taken to ensure 25  comparability but this judgmental assessment is a necessary part of the analysis to derive an overall “confidence” measure (BCI). With that in mind, the BCI metrics are made explicit here, to allow for modification of the weighting by other researchers and in the context of new data. To summarize, this GPR analytical tool focuses on those elements I consider to be reliable identifiers for burials based on GPR signal quality. For unmarked locations where some physical contour evidence exists at the surface, such as mounding or depressions, that information too is taken into account in creating an estimate of BCI for comparison. GPR analysis in isolation can never be 100% certain that any location represents a burial. Since ground-truthing via excavation is not an option, however, by testing known sites against other sets of less-supported but evident signals, our level of “confidence” in describing them may be improved. A model of this kind, where selected factors can be highlighted, and that reflects the limits of what we can know and measure, may become an effective method for increasing our interpretive capabilities. Re-iterating the objectives for this thesis provided in the Overview, these methods can be applied to the analysis of any potential burial location in varying social contexts. The model is designed to examine and test what are the qualities of GPR signals that are most indicative of human burials? In Table 5.3, each variable has been grouped into three broad sets of factors reflecting typical GPR data collection processes for characterizing each location. This allows for a greater understanding of the contributing factors in each case to the overall BCI measure and for top-down manipulation of the decision criteria. These three categories cover the following factors:      26  Table 5.3 General factors influencing BCI results.  Factors Evaluative nature Surface characteristics and features Physical conditions and contextual information GPR signal metrics and morphology GPR signal measurement indicators GPR signal characteristics Signal content evaluation  5.2 General factors influencing BCI results  5.2.1 Surface characteristics and features  The first category or broad factor that needs consideration covers contextual data and physical observations made during GPR grid data collection. This can include features extant at each location such as grave markers, variations in ground conditions such as grassy mounds and other physical observations, row structure fit and spacing to adjacent burials. In addition, the interment date and other information are equally important observations from the physical structures placed at site. Included with this information are other surface descriptions, plus the precise geographical location and GPR grid identification information. It allows for direct connection of surface and sub-surface factors incorporating the influence of nearby physical features such as trees and rocks, and is essential in linking surface context to GPR signal results.  5.2.2 GPR signal metrics and morphology  The second factor covers specific measurement of GPR display outputs. Roemer and Cowling (2006: 519) note that GPR displays can often be very difficult to interpret especially in plan view, however, measurements in profile are much more accessible and allow for target location probabilities to be developed. Constituent measures include average depth, length and width of signal anomaly in profile, and a calculation of the total area covered. For example, in a recent burial made in a highly reflective container with high contrast to the surrounding soil (such as metal), the size of the GPR image displayed will match closely the dimensions of the actual container. For burials of significant age that have broken down, or 27  that were made of less reflective materials such as wood or cloth, contrast to the soil matrix will be much reduced. This effect is likely exacerbated in older interments.   5.2.3 GPR signal characteristics  The third important factor covers the interpretation and evaluation of GPR display content in both plan and profile view outputs. The analysis tends to be more intuitive and is highly experiential, and seeks to evaluate the strength and subtleties observed within individual GPR images matched to precise surface locations. The key areas to be evaluated for each case are signal intensity, and the degree to which the image in profile indicates stratigraphic compaction and contrasting density from the surrounding soil matrix. Table 5.4 elaborates on these three broad of cemetery data factors available to us and breaks them down into 12 discrete, measurable variables shown in Table 5.3. These factors form the fundamental variables in the BCI Model and their relevance as key contributors to the determination that any location is in fact a burial:  Table 5.4 Burial Confidence Index component variables.  Surface characteristics High likelihood Low likelihood Grave marker presence Present Absent Grave marker anchoring Anchored and immobile Removable Row structure Adjoining aligned graves Scattered Row spacing Between 1.0 and 2.0 m. Closely packed markers Presence of other surface features Mounds, grave outlines None Availability of other historical records Existing written and oral None    GPR signal metrics High likelihood Low likelihood Signal depth Depths of > .90 m. Depth of < .30 m. Signal area Value > 1.40 m. Value < 0.40 m. Signal length: (W-E) Length > 1.60 m. Length < 0.50 m. Signal width: (N-S) Width > 0.80 m. Width < 0.30 m.    GPR signal characteristics High likelihood Low likelihood GPR Plan view signal intensity Bright yellow to red Absent to pale blue/yellow GPR Profile view stratigraphy Bright alternating layers Discontinuous layers  The first six variables cover surface characteristics and observations made before and during GPR surveys. These observations have a direct effect on our ability to analyze the 28  GPR signals as they point to specific spaces worthy of closer examination. Their value in providing a more complete analysis of any space cannot be underestimated.  5.3 Surface characteristics influencing BCI results  5.3.1 Grave Marker Presence  Though self-evident, there is likely no more diagnostic burial variable than the presence of a grave marker. However, only 96 (33.3%) of the 288 locations surveyed and analyzed were marked with a permanent structure with named occupant. The grave marker names themselves can also contribute to identification of locations as burials as some areas within post-contact cemeteries are dedicated to particular groups or families. There were a further 33 locations with markers (making up the 129 total marked locations shown in Table 5.4 (1a and 1b) for Musqueam and Kwantlen) but these lacked names or other identification. This key factor emphasizes the value implicit in the use of GPR survey in identifying in-cemetery burial locations. Evaluation criteria: present or not (Yes/No).  5.3.2 Grave marker anchoring  The second variable associated with burial determination is the security of the actual marker itself. Over time, through vandalism, weathering, grounds restructuring/re-organization and other natural change factors, grave markers can be dislodged. In my survey dataset, this phenomenon was particularly noticeable at the Kwantlen Main Cemetery (Daniel & Martindale 2009). There were two entire rows in the Southeast part of the grounds (Grids #6 & 7) with markers that were mobile and that appeared to have (re)placed within a defined row structure. In my opinion, this was done likely for appearance sake and to present more orderly grounds. Though being anchored is not necessarily an absolute assurance that the stone is associated with the burial beneath, it is certainly a 29  strong indicator and worthy of inclusion as a variable in the BCI Model.  Evaluation criteria: mobile or not (Yes/No).  5.3.3 Row structure  This factor in the model is essential in that it points to high priority areas for survey and location of unmarked burials (or “Unkown” as referred to in Table 5.3). My survey techniques included placing specific GPR grids at the end of rows to check for their extension beyond the mere surface appearance. At times, certain rows of grave markers would be interrupted and resume some yards farther along, or be offset by a meter or less and continue. This variable is certainly the key to finding those areas that lack specific surface indications but with a high probability of containing burial signals. In effect, this variable provides a rough set of coordinates for grid placement and among the 288 locations surveyed was responsible for identification of approximately 40% of the unmarked locations presenting strong GRP traces. Evaluation criteria: in row or isolated (Yes/No).  5.3.4 Row Spacing  As an extension of the row structure variable, significant gaps within each row become prime targets for GPR grid analysis. Any GPR survey that does not cover these apparently “empty” spaces between adjacent graves falls short of the goal of being as comprehensive as possible. It is also essential in practical terms for identifying potential space for contemporary, new burials. The centre point at the headstone of each burial was found to be approximately 2.0 to 2.2 meters distant from the next burial in a row. Significant variations from this typical spacing became automatic targets for more in-depth analysis of the signals.  Evaluation criteria: Spacing of 2.0-2.2 meters or other spacing (Yes/No).  30  5.3.5 Presence of other surface features  This variable was designed to account for apparent grave sites that were not accompanied by a specific headstone or marker, but had the appearance of a highly disturbed soil or surface area. This often appeared as a visible outline of normal grave dimensions (approximately 1.2 by 2.5 meters) with variations in the surface vegetation or clear blank space where a marker was likely to have been located at one time. It can also be as obvious as a recent burial mound that has not yet had final memorials or markers placed there.  Evaluation criteria: presence or absence of these other indicators (Yes/No).  5.3.6 Availability of other historical records  The last of the surface informational variables covers other documentary evidence of the existence of grave sites, many of which may have been lost over time. At Musqueam, we were presented with a very rough grid diagram of the Main Cemetery complete with names, but in no way geo-referenced. This provided a very rough row structure and geographic location and turned out to be extraordinarily valuable in the search for presently unmarked locations. At Kwantlen Main Cemetery, considerable historical documentation was provided before beginning our surveys there. These data ranged from row-by-row identifications and names, to oral histories of the grounds and specific areas within. Discussions with community members and their offers of input, often rendered while survey work was being conducted, pointed frequently to certain spaces and led to requests to use our GPR array there. This is a broad category of data and yet a variable that can lead to a more comprehensive coverage of the grounds.  Evaluation criteria: available or not available (Yes/No). It should be noted here that the degree to which historical information supports any location can vary widely in quality and quantity but for purposes of this thesis and its relatively low weighting, is sufficient to be a Yes/No decision. 31  5.4 GPR signal metrics influencing BCI results  5.4.1 Signal depth  This is one of the most important criteria in determining burial identifications, particularly for those GPR “anomalies” that present very near the surface. The likelihood of these signals being rocks and roots is higher near the surface, especially at the edge of cemetery grounds bordered by trees and brush. The data values for signal average depth used in the BCI model reflect actual in-the-ground measurements from GPR profile view output. Evaluation criteria/scale: For depths of less than 0.3 m. it is assumed for this study that the anomaly would not be representative of a burial, especially in apparently undisturbed contexts. For depths below 0.3 m., a linear scale of increasing likelihood that a signal represents a burial has been applied. Starting at just below the surface down to 0.3 m., a 0% likelihood was assigned for this variable. From 0.3 m. to 1.0 m. depth, the likelihood that a GPR signal represents a burial increases down to the optimal depth encountered consistently across the majority of locations at 1.0 m or more (assigned a 100% score). It should be noted for the signal depth variable that post-interment histories at some sites indicate much change across their surfaces over time. This is due to periodic fill removal or additions, along with the natural processes of erosion or sudden falls down adjacent embankments. This latter event was described to us at two of the three sites in the study along with a fourth site (Kwantlen Upper Cemetery) that was not analyzed due to the extreme level of disturbance to the grounds.     5.4.2 Signal dimensions (area, length and width)  Typical GPR profile (side) view output allows for accurate measurement along two grid axes (X & Y) and this reflects the size of each anomaly effectively. These measurements 32  may be cross-checked with the top down or plan view for concurrence. Since many GPR-identified anomalies are located close to or adjacent to others, it can be a process of distinguishing one from the other with reference to original sketch maps and GPR grid photos taken during survey. With normal attenuation of signal quality with depth and due to the degrading factors listed above, the relationship between burial outlines at the surface and as modeled via GPR shrink by a ratio of 60 to 70% on average for strong signals and more for moderate or weaker ones. From these dimensions, the area can then be calculated, which provides an additional important factor for calculating BCI. The importance of area as a variable is that it tends to recognize a shape that approximates a burial (i.e., rectangular with sides that do not vary widely from one another). With respect to the absolute and length and width of GPR signals as indicated in plan (top down) views, these match the actual surface burial space (comparing signal area to surface area as recorded in the associated data for each location). For example, on the surface any potential burial location may show as 1.5 by 2.8 meters. The GPR signals associated with it sub-surface will match that set of dimensions but as mentioned above, be slightly smaller due to signal attenuation or the assumed natural break down of the burial container. It should be noted that an absolute match is not testable since we are of course unable (nor is it allowed or advisable) to carry out test excavations. Evaluation criteria/scale: For profile view signal length and width values given the inherent reduction in the size of reflection data, Table 5.4 provides a set of evaluation criteria for each measure that indicate different likelihood rank-levels that an anomaly may represent a burial. Those burials that are known and well understood, (e.g., contemporary interments with specific boundaries), the most frequent dimensions are approximately 1.5 by 2.8 meters with further space notable at times for more extensive memorials.   33  5.5 GPR signal characteristics influencing BCI results  5.5.1 Plan view signal intensity and profile view stratigraphy  GPR signal quality is by far the most subjective element in the model but is supported for plan view intensity by distinctly coloured output scales within the Ekko Mapper software. The least intense signals are barely visible or light blue in appearance (similar to the background would be without any apparent GPR trace) and range to deep red, similar to isothermic temperature maps. Each location may be assessed in plan view in this way and with an increasing intensity scale.  GPR signal strength and quality is addressed more easily in profile, through analysis of the stratigraphic layering apparent for each anomaly. Stronger signal sets appear as highly contrasting, thicker stratigraphic layering black and white with alternately layers marked at their edges by sharp lines delineating burial shaft/container margins. Evaluation criteria/scale: Valuations for each of the 288 locations were performed across a scale from “None” in the absence of any signal strength, through “weak”, “moderate” and up to “strong” signals.   5.6 GPR burial analysis model calculation framework (BCI)  An important design criteria for the BCI Model was first to take a “top down approach” assigning a general level of importance to each broad category of contributing factors or traits (shown in Table 5.3). At the highest level, this asks:  What information is of the highest value or should be the greatest contributor in determining whether a location represents a burial or not?  Based on the three groups of survey data in Table 5.3, and comprised of twelve specific measurable variables (Table 5.4), through calculation we are able to determine their potential for contribution to the calculation of the BCI index. The next step before producing 34  a BCI value is to evaluate their overall contribution to BCI at the macro-level through careful weighting. In this study, the GPR signal analysis is weighted more heavily as that is the emphasis here; 40% of BCI has been allocated toward surface indications (Group 1: Variables 1-6). The GPR-related variables (Groups 2-3, variables #7-12) are assumed to carry 60% of the weighting overall. For example, to what extent should the presence of a grave marker and other definitive historical records for an interment be favoured over the presence or absence of GPR signals?  In my experience, I have seen perfectly standard and substantial monumental “burials” turn out to be completely absent of any GPR signal. Conversely, some of the strongest and most unambiguous signals have been obtained from absolutely unmarked locations, many of which we confirmed via historical documentation and other ethnographic evidence, were indeed burials. How precisely to make this split was therefore an “unanswerable” question, but for purposes of this analysis was varied and the results compared. Recursively, the effect of each competing probability (say between a marked, known burial lacking a GPR signal, and an unmarked burial with a very strong, recognizable burial trace) can be explored separately and varied to produce different results that can be subjected to review. The model’s results can be iterated through testing thus steering the model’s “fit” for any location toward potentially a more realistic result. In this way, the influence of any one factor can be tested as well as the creation of the index itself. For example, if there is a burial that is recent, well-marked, witnessed and recorded, then a 100% BCI value is the only appropriate result. However, in fact the real usefulness of this BCI model is directed toward locations where those data are not fully available. In fact, this acted as a guide in prioritizing areas at the time of data collection. Given this caveat, for 35  these larger questions in this model, I have taken the following approach as laid out in Table 5.5:  Table 5.5 Overall weighted contributing factors influencing BCI.  Factor Contribution to BCI Surface characteristics and features 40% GPR signal metrics and morphology 20% GPR signal characteristics 40%    The weightings for each component were coded into an Excel-based spreadsheet model (see calculations and data in Appendix 1) and run against the values derived in assessing each of the 288 locations across 12 criteria. The model was created in Microsoft Office Excel (2010) and with sufficient flexibility to vary any of these overall parameters at either the highest level or in any detailed weighting % assigned. Table 5.5 indicates the relative weighting for each broad data category producing an overall assessment of how much they should contribute to BCI. In Tables 5.6 and 5.7, the relative influence of each constituent factor in the model is assigned. In Table 5.7, the rationale for the range intervals is based on increasing likelihood that a GPR signal represents a burial as the depth increases below 0.3 meters.  Table 5.6  Weighted surface characteristics influencing BCI.  Surface Characteristics Contribution to BCI Grave marker present? 20% Marker anchored? 2% In a row structure? 10% Row spacing typical? 2% Other surface indications? 3% Supporting historical records? 3%           36  Table 5.7  Weighted signal metrics influencing BCI.  GPR Signal Metrics Range of Results and Assigned Score Signal average depth  (5%) Data <.30 0.30-0.59 0.60-0.79 0.80-0.99 >1.00 (meters) Score .00 .25 .50 .75 1.00        Profile view trace length (5%) Data <.50 0.50-0.89 0.90-1.29 1.30-1.59 >1.60 (meters) Score .00 .25 .50 .75 1.00        Profile view trace width   (5%) Data <.30 0.30-0.44 0.45-0.59 0.60-0.74 >0.75 (meters) Score .00 .25 .50 .75 1.00        Trace area calculation  (5%) Data <.40 0.40-0.79 0.80-1.09 1.10-1.39 >1.40 (sq. meters) Score .00 .25 .50 .75 1.00       GPR Signal Content Range of Results and Assigned Score Plan view intensity     (20%) Range None Weak Moderate Strong  Score .00 .33 .67 1.00       Profile view intensity  20%) Range None Weak Moderate Strong  Score .00 .33 .67 1.00       In terms of method, for all 288 locations, their individual assessments across each variable were performed, with the presence or absence of each factor assessed either from the original data recorded in the field (and as reported in each Project Report to the client along with supporting field notes and sketch maps) or from GPR signal analysis. This documentation became especially important for unmarked locations where the original headstone may have been lost, broken, removed, stolen or decayed away and the location coordinates were not immediately obvious on-site.  As such, results derived from the GPR signal evaluation in Table 5.7 fell into two types, the first of which was the absolute metric values as produced in GPR displays for both top down and profile views. The second set was primarily interpretive for each contributing factor and relied on the specific measures, qualities or evaluation criteria observed and described above. Calculating BCI Values: Field-recorded data for each surface-related variable was evaluated by presence or absence and then assigned a value of one or zero independently as either Yes/No. This result was then multiplied by the assigned weighting and a raw score created. More complex calculations were required for the signal analysis variables as they are factors measured across a range of expected values from GPR signal displays for all 37  locations sampled. For these measurements, their absolute data values as revealed in the GPR signal displays were assigned a scale of relevance or the degree to which they may be indicative of burials (Scales shown in Table 5.7. Values for all variables across all locations were then compiled into 288 individual BCI results. In producing “scores” for locations, the results were compiled into sub-indices reflecting the three groups of contributing factors. Each sub-index was analyzed to assess the major factor contributing most or least to the overall BCI, the assigned likelihood that a specific anomaly may reflect an adult human burial in extended position of approximately 1.5 by 2.8 meters. Results were compiled and evaluated across similar BCI results in each data category and correlated back to any single variable. A BCI result of zero would therefore represent no likelihood of burial presence, while higher scores represent greater probability up to 1.00 and relative certainty. Though not performed here, BCI results may be correlated to interment dates for locations where markers provided that information.  The model was designed to easily adjust weightings and test internally the assumptions made at the outset for weighting. It was created in a way that laid out the basic algorithm in simple easily understood fashion, along with other features and flexibilities for in-depth study of the data. What was required for each potential burial location was to analyze each site across all variables that make up the index and their related signal trace data going through all GPR grid displays, assessing content and signal quality. For each variable, the optimum range of expected values received a score of 1.00; at the low end a score of zero was applied indicating that a burial was likely not present (example: depth of 5 cm. or an anomaly indicative of a point source less than .4 m. square). The anomalies or exceptions here would be child burials, small containers or adult flex burials. These variations would be challenging for any analyst as they vary substantially from the vast majority of the data that reflects adult burials in extended position. 38  Tables 5.8 and 5.9 show sample BCI calculations and decisions for several locations and depicts graphically a small part of the actual Excel-based BCI model (detailed in Appendix A and Appendix B). The lettering shown in some columns is part of the coding that allows for privacy of the data but with clear matching to the particular cemetery or location with the confidential key.  The calculated BCIs are a direct measure of how likely a burial is present and is more rigorous than producing a single, intuitive evaluation such as “weak”, “moderate” or “strong” as the sole indication. The BCI model could also allow for an element of “overall” judgment (13th variable or assessment) as part of the interpretive process akin to connoisseurship, but that approach was not employed in favour of the more rigorous process across 12 variables/measures. The GPR signal evaluation variables do require some qualitative evaluation, however, the colour intensity scale and clarity of profile curves in the GPR software output render that a relatively straightforward and objective process.  Table 5.8  Sample burial locations showing BCI calculations (Excel model).             39  Table 5.9  Sample burial locations showing variables & BCI weightings.   Summarizing the fundamental modeling technique, the process was based in part on methodology used effectively by Martindale and Supernant (2009:196) calculating a defensive index across multiple sites in Southwestern British Columbia. Looking forward, the model is designed to become a useful template for organizing and interpreting burial data from any historic period cemetery site.  40  6  BCI Model Results & Analysis  To evaluate the Burial Confidence Index Model, the 288 locations initially identified in Table 5.3 were isolated into the three separate groups. The Control Group was referred to as “1. Musqueam (Known/Marked)” and contained 94 individual geographic locations within the Main Musqueam Cemetery. Data for these burials was then compared across the 12 different criteria/variables with their relative weightings and then aggregated to produce a Total BCI for each space. Results for this first group are shown in Table 6.1:   Table 6.1  BCI Model Results Musqueam Known/Marked Locations.  Test Locations GPR Tot   Data Group Type / Marked Sig? Tot % BCI Surface GPR         Study Total Weighting  288 100 1.000 0.400 0.600                  1.  MUSQUEAM Known (marked) Yes 94 33 0.898 0.394 0.504  Results: % of max.  90.0% 98.5% 84.0%        In the calculation, each BCI result is shown as two components – Surface (40% weighting) and GPR signal values (60% weighting) – reflecting their contribution to the overall result.  Re-iterating, the split of 40/60 for weighting was selected due to the emphasis on GPR signal results, and this sub-division in the BCI enabled a deeper look into what factors were influencing the value the most. The 94 Group 1 marked and known Musqueam locations produced a pooled BCI with a mean value of 0.90 out of 1.000 and may be interpreted as having a mean that approaches 90% of the ideal, absolutely certain value. The Surface variable contribution toward the 0.90 overall result was 0.394 or 99%, while the GPR result was 0.504 or 84% of maximum. This means that there is a very strong match between known historical burials supported by both historical knowledge and tangible burial memorial structures, and their related GPR signal traces. 41  The difference between the 90% result for surface characteristics and what may be expected to be a perfect one-to-one correlation is that some of these 94 locations lacked elements such as anchoring of the headstone, were irregularly spaced or were not included in older Band-owned maps and diagrams of the Main Musqueam Cemetery. Among these 94 marked locations, the spread showed a median value of .934 BCI, a mean value of .898 and indicated 76 locations with a BCI of least .800. None were lower than a .600 BCI and that was the result primarily of weak GPR signals associated with well-marked graves. Other graves were placed not in a given row but appeared to be randomly placed, though this is of course outside our ability to know. Maps prepared for the 2008 and 2009 UBC-Musqueam Field Reports (Martindale 2008, Martindale 2009), included all of these locations and their precise cultural identity and geographic coordinates. The difference between the .504 (84%) result obtained for GPR signals and the maximum result of .600 is a more complex relationship, but may be attributed to signal attenuation with burial depth and variation in soil matrices, along with the burial aging factors referred to in the description of the model. In the Musqueam Main Cemetery, there are a high proportion of locations with interment dates in excess of 40 years. Therefore, in my view, based on examination of thousands of sets of GPR trace signals, an 84% result is an extraordinarily high correspondence between the hard found evidence at the surface, and the GPR graphic display associated with it. To further examine the 84% comparison of GPR signal quality to the maximum attainable value and test that quality within the model, the means of the 94 associated GPR signal dimensions at Musqueam were calculated (see Table 6.2) These mean GPR signal dimensions have a display size of 1.09 by 1.80 meters (length:width ratio = 1.65).   42  Table 6.2  Mean GPR Signal Dimensions: Musqueam Known Locations  Test Avg Avg Avg Avg Data Group Tot % Area Width Length Depth         1.  MUSQUEAM (Known, marked) 94 33         Mean GPR signal dimensions 2.03 1.09 1.80 1.18                   COMPARISON TO SURFACE:         Mean surface location dimensions 2.50 1.15 2.17                     The length to width ratio of 1.65:1 recorded in GPR signal displays matches very well with the ratio of average burial location size on the ground of 1.89:1 (measured as grave sites of approximately 1.2 by 2.2 meters). The reduction in average area from the ground extent (4.48 m2) to the size at depth (2.03 m2) is also easily justified in my experience and a good match.   As a final test of the data for Musqueam Known burials, more than 91% (86 of 94) of the 94 locations run through the BCI model, contributed to BCI at .360 or higher (60% of the maximum value). Only six locations exhibited relatively weak GPR signals contributing to BCI at less than half of maximum of .400 and therefore provide a much lower level of confidence in their interpretations as human burials. The central tendency as measured by the mean BCI values is indicative of a dataset that appears tightly packed around the mean with few outliers, and thus is similar both in value and quality. Summarizing the control group run initially through the BCI model, the results indicate strongly that there is a clear correlation between GPR signals and their associated surface physical features and characteristics. GPR software appears to be able to identify with some reliability and “confidence” human burials in the historical contexts tested in this study. Having established that the BCI model is able to match between ground and sub-surface burial manifestations at Musqueam, the next step was to conduct the first 43  comparative test and assess if the fit is as strong for burials at a completely separate location. Table 6.3 provides BCI results for 35 known and marked locations from the Kwantlen Main Cemetery ( “2. KWANTLEN (Known (marked)”) along with comparable results from Musqueam known and marked burials for comparison (from Table 6.2)   Table 6.3 BCI Model Results Kwantlen Known/Marked Locations - Results Comparison  Test Locations GPR TOT   Data Group Type/Marked Sig? Tot % BCI Surface GPR          STUDY TOTAL WEIGHTING  288 100 1.000 0.400 0.600                  2.  KWANTLEN Known (marked) Yes 35 12 0.846 0.400 0.446  Results: % of max.  85% 100% 74%                1.  MUSQUEAM Known (marked) Yes 94 33 0.898 0.394 0.504  Results: % of max.  90% 99% 84%         The 35 Group 2 marked and known Kwantlen locations produced a pooled BCI with a mean value of 0.846 out of 1.000 or having a central tendency of 85.6% of the ideal, absolutely certain value. The Surface variable contribution toward the 0.846 overall result was 0.400 ( 100% fit), while the GPR result iwas 0.446 or 74.3% of maximum. Interpretation of this 100% BCI Model result for surface features was not surprising given that the Kwantlen Cemetery is an exceptionally ordered site, with a strongly defined row structure and supported by a wealth of paper historical records, maps, drawings, death certificates and other accounts. In my informal discussions with Band members who assisted in data collection on-site, Kwantlen Main Cemetery had recently undergone a substantial refurbishment and its history was well known and understood. Kwantlen Band Council provided us with much historical documentation and these data closely matched grave sites on the ground. The GPR signal results were more problematic given the lesser .446 out of .600 contribution to BCI (74% fit). Reasons for this are likely that the age of the burials found 44  there in many cases exceeded more than 60 to 70 years with many interments as long ago as 1876. Longer time frames likely imply a more substantial breakdown of the burial itself, and provide less of a target for GPR equipment to locate and return information. Beyond that, since we do not know the nature of the burial containers at either cemetery site. Given those conditions, a .446 GPR signal result (out of .600) provided me with the assurance that again, the BCI model indicated a strong relationship between surface and sub-surface indicators for human burials. The ratio of Kwantlen GPR signal dimensions (Table 6.4) to surface location extent is fairly consistent with that found at Musqueam and acts as a further test of signal size reduction in a wholly different location and cultural association. The main observation and assertion to be made in this comparison is that within this dataset of 129 locations, ground penetrating radar analysis appears capable of identifying human burials equally well regardless of cultural association. However, there will always be factors which will limit our ability to be certain, and that have the potential to significantly affect BCI results both in absolute and comparative terms. Not knowing the natural variations in the type of burial container and original depth of the burial (as opposed to the depth at which GPR signals are first observed), along with weathering and other changes at the surface, radical ground level changes such as that which occurred at Kwantlen, soil matrix and water content levels, land falls and climate impacts necessitate some caution.  Table 6.4  Mean GPR Signal Dimensions: Kwantlen Known Locations  Test Avg Avg Avg Avg Data Group Tot % Area Width Length Depth         2.  KWANTLEN (Known, marked) 35 12         Mean GPR signal dimensions 1.18 0.80 1.30 0.82     Ratio Length to width  1.63:1                     COMPARISON TO SURFACE:         Mean surface location dimensions 2.50 1.10 2.05      Ratio Length to width  1.86:1           45  GPR statistical analysis was done to compare BCI results from the two known groups at Musqueam and Kwantlen to the remaining 159 previously unknown burial locations. I found the absolute number of unmarked, and therefore “unknown”, burials to be well beyond my expectations and was 55% of the 288 total locations identified. These spaces lacked any headstone or marker but in many cases had at least some surface indications. Table 6.5 shows BCI values for all three groups but discussion here focuses on results derived from running the model against all unmarked sites “3. ALL OTHERS”:  Table 6.5  All BCI Results by Test Data Group  Test Locations GPR Tot   Data Group Type / marked Sig? Tot % BCI Surf GPR          STUDY TOTAL WEIGHTING  1.000 0.400 0.600          1.  MUSQUEAM Known (marked) Yes 94 33 0.898 0.394 0.504  Results: % of max.  90% 99% 84%         2.  KWANTLEN Known (marked) Yes 35 12 0.846 0.400 0.446  Results: % of max.  85% 100% 74%         3.  ALL OTHERS Unknown (unmarked) Yes 159 55 0.560 0.094 0.466  Results: % of max.  56% 23% 78%  STUDY TOTAL All locations  288 100 0.71 .229 .476  Results: % of max.  71% 23% 78%   The contribution of the surface features to BCI was very low at .094 of a possible .400 – only a 23.5% match. As a further test of the BCI results data, a box and whisker analysis (Table 6.6 and Figure 6.1) was performed to check the quality of data in terms of spread. Table 6.6  Box and Whisker Data for Known and Unknown Burials.   ----- Known Burials ----- ---- Unknown Burials ----  Tot Surf GPR Tot Surf GPR Statistic BCI BCI BCI BCI BCI BCI        Minimum value 0.544 0.280 0.145 0.195 0.000 0.157 Quartile 1 0.816 0.400 0.429 0.434 0.120 0.365 Median 0.925 0.400 0.534 0.588 0.120 0.525 Quartile 3 0.988 0.400 0.600 0.695 0.120 0.575 Maximum value 1.000 0.400 0.600 0.934 0.400 0.600 Mean 0.884 0.396 0.488 0.560 0.094 0.466        Standard deviation lower 0.864 0.392 0.473 0.515 0.092 0.460 Standard deviation higher 0.986 0.408 0.595 0.661 0.148 0.590     46    Figure 6.1 Box and Whisker plot of Known and Unknown Burials.   The box and whisker analysis performed in Table 6.6 and illustrated in Figure 6.1 confirms that the BCI results shown in the summary Table 6.5 were very tightly packed around the median values with few outliers and a comparatively small spread. These patterns reveal that 1) as expected, known BCI values are tightly clustered for Surface, GPR and Total values, and 2) GPR data for unknowns has similar distribution and median values to the known sample. This suggests that BCI is a reasonable tool for approximating confidence of human burial identification in unknown contexts.   47  In terms of their actual geographic locations, many of the Unknown (Group 3) GPR signal sets came from clear gaps in rows or were spaced in such a way as to suggest strongly that a grave was there. In many cases, the marker had simply degraded (as in wooden crosses), had been removed by vandalism (per Kwantlen and Musqueam community members), or was subject to other natural processes. The most culturally valuable locations among these “unknowns” occurred on two specific occasions at Musqueam following exchanges initiated by community members during our survey. Two individuals approached me asking for us to examine an isolated area where only one unanchored wooden cross lay on the ground but contained two burials of close family members. Way beyond GPR research considerations, this sort of historical knowledge contributes far more to the obviously much higher descent community goals than is the focus of this paper. After testing in the field with the software that day, two distinct intense sets of GPR presented themselves clearly, almost entirely absent of any substantial surface indication.  The real test of the BCI model however is, of course, whether we can find indications of burials using ground penetrating radar that exhibit all of the same qualities but for previously unknown locations. From a project management perspective, this represents an extraordinary challenge as in order to be assured one way or the other, full ground coverage at each cemetery site was required. In field school settings where a teaching component is essential, and where weather can be a factor (GPR in the rain is very difficult) it is often hard to go quickly.  Table 6.5 provides BCI results for previously unknown burials pooled across 159 sets of data run against our weightings, a .466 out of .600 (78%) sub-surface GPR-related contribution to BCI emerged. This result fell in between the Musqueam control group (.504 of .600, 84%) and the smaller dataset compared to that from Kwantlen Main Cemetery (.446 48  of .600, 74%). This 78% GPR signal relationship for previously unknown burials compares very closely to the aggregated value for BCI for all 129 known and marked burials of .488 or 81%. In my assessment this 81% BCI value and strong correlation suggests with a high degree of “confidence” that these locations are indeed likely to be human burials. 49  7  Conclusions  The foundation for this thesis is a highly detailed ground penetrating radar study of approximately 12,000 m2 of historic period cemetery grounds over a period of three field seasons in direct collaboration with Musqueam and Kwantlen First Nations. The work was done often in Field School or instruction mode as well as by direct engagement, and always with well-defined protocols and community-reporting goals. The methodology relied primarily on construction, testing and application of a Burial Confidence Index across an array of almost 300 separate spaces and many, many more that did not meet the criteria for study. The technique was quite simply to organize and evaluate all we can know about any particular location and quantify our findings into an easily understood quantification of our assessments. The results derived from testing and running the model lead me to conclude that given very similar absolute BCI values for previously known and unknown burials, along with consideration given to the observed quality of the data in terms of spread and a tightly packed nature of the BCI results around the mean, that GPR analysis appears to be able to identify burials with a high level of confidence. Beyond the actual statistics presented, the quality and intensity of the graphics presented in GPR plan and profile displays for both previously known and unknown burials were virtually indistinguishable. Indeed, it is apparent that strong sets of GPR signals that evoke the basic shape and dimensions of burials at the surface can act as a proxy for often absent or limited ground features and indications. Though the BCI results show that we are less confident in the unknowns, the difference between mean BCI’s of .488 and .466 is not a substantial difference  As a final set of conclusions, in my view, the strength of this study lies not merely in the development of a standardized model or organized GPR management process but in 50  having done the actual data collection and analysis phases of the work over many hours. Identifying 159 previously unknown burials means a lot of survey time spent on grounds that lack both surface indications such as markers, and any significant GPR signals. GPR survey using this BCI model allows researchers to overcome substantially the inability to be 100% certain, and to develop a higher level of “confidence” or likelihood that a location is or appears to be a human burial. With that in mind however, it is a certainty that to be effective and make reliable assessments, critical experience with the GPR software and developing the ability to “read” GPR graphic curves and shapes and their messages is essential. This is a particularly important next step for researchers interested in taking on practical engagements where GPR results are a key tool for advising clients in determining the use of any space within their cemetery grounds. 51  REFERENCES CITED Buck, Caitlin E., William G. Cavanagh, and Cliff Litton 1996  Bayesian Approach to Interpreting Archaeological Data. Wiley, Mississauga.  Conyers, Lawrence B. 2011 Discovery, mapping and interpretation of buried cultural resources non-invasively with ground penetrating radar. Journal of geophysics and engineering 8:S13-S22.  2010  Ground-penetrating radar for anthropological research. Antiquity 84:175-184.  2009  Ground-penetrating radar for landscape archaeology: Method and applications. In Seeing the Unseen, edited by Stefano Campana and Salvatore Piro, pp. 245-255. Taylor and Francis Group, London.  2006 Ground-Penetrating Radar Techniques to Discover and Map Historic Graves. Historical Archaeology 40(3):64-73.  2004  Ground-Penetrating Radar for Archaeology. AltaMira Press, Lanham.  Conyers, Lawrence B., and Catherine M. Cameron 1998  Ground Penetrating Radar Techniques and Three-Dimensional Computer Mapping in the American Southwest. Journal of Field Archaeology 25(4):417-430.  Conyers, Lawrence B., and Dean Goodman 1997  Ground-Penetrating Radar for Archaeology. An Introduction for Arachaeologists. AltaMira Press, London.  Conyers, Lawrence B., and Juerg Leckebusch 2010  Geophysical Archaeology Research Agendas for the Future: Some Ground penetrating Radar Examples. Archaeological Prospection 17(2):117-123.  Goodman, Dean, Kent Schneider, Salvatore Piro, Yasushi Nishimura and A.G. Pantel 2007  Ground-penetrating radar advances in sub-surface imaging for archaeology.   Daniel, Stephen E. 2009 The Application of Ground Penetrating Radar: A View from Above and Below at Kaukana, Sicily. Unpublished report on file with the Laboratory of Archaeology, Department of Anthropology, University of British Columbia, Vancouver.  Dojack, Lisa 2012  Ground Penetrating Radar Theory, Data Collection, Processing, and Interpretation: A Guide for Archaeologists. Available online at UBC’s digital information repository (cIRcle – https://circle.ubc.ca/handle/2429/41983).  Doolittle, James A., and Nicholas F. Bellantoni 2010  The search for graves with ground-penetrating radar in Connecticut. Journal of Archaeological Science 37:941-949.   52  France, D.L. et al 1992  A multi-disciplinary approach to the detection of clandestine graves. Journal of Forensic Science 37:6, 1445-1458.  King, J.A., B.W. Bevan, and R.J. Hurry 1993  The reliability of geophysical surveys at historic period cemeteries: an example from the Plains Cemetery, Mechanicsville, Maryland. Historical Archaeology 27:4-16.  Martindale, Andrew 2011 2010 Musqueam-UBC Field School – ANTH 306 Field School Final Report. Unpublished report on file with the Laboratory of Archaeology, Department of Anthropology, University of British Columbia, Vancouver.  Martindale, Andrew 2010 2009 Musqueam-UBC Field School – ANTH 306 Field School Final Report. Unpublished report on file with the Laboratory of Archaeology, Department of Anthropology, University of British Columbia, Vancouver.  2009  2008 Musqueam-UBC Field School – ANTH 306 Field School Final Report. Unpublished report on file with the Laboratory of Archaeology, Department of Anthropology, University of British Columbia, Vancouver.  Martindale, Andrew, and Steve Daniel 2008b  Kwantlen First Nation Cemeteries Project: Final Report Report. Unpublished report on file with the Laboratory of Archaeology, Department of Anthropology, University of British Columbia, Vancouver.  2008a  2007 Musqueam-UBC Field School – ANTH 306 Field School Final Report. Unpublished report on file with the Laboratory of Archaeology, Department of Anthropology, University of British Columbia, Vancouver.  Martindale, Andrew, and Kisha Supernant 2009  Quantifying the Defensiveness of Defended Sites on the Northwest Coast of North America. Jornal of Anthropological Archaeology 28(2):191-204.  Novo, Alexandre, Henrique Lorenzo, Fernando I.Rial, Mercedes Solla 2011  3D GPR in Forensics: Finding a clandestine grave in a mountainous environment. Forensic Science International 204:134-138.  Roemer, Louis, and David Cowling 2006  Bayesian Methods in Ground Penetrating Radar. AIP Conference Proceedings 872(1):519-524.  Schultz, John J., and Michael M. Martin 2011  Controlled GPR grave research: Comparison of reflection profiles between 500 and 250 MHz antennae. Forensic Science International 209:64-69.  Stanger, Ross and David Roe 2007  Geophysical Surveys at the West End Cemetery, Townsville: An Application of Three Techniques. Australian Archaeology 65:44-50. 53  Appendix A: Coded Burial Location Identification Information  Coded  CODED LOCATION DATA     GPR GRID COORDINATES: Report Coded Loc. GPR Data   Grid Grid From From ID# Site Mark? Sign? Grp     Refer Row E  N  R18 M Y Y 1 NE 3 8.4 1.5R19 M Y Y 1 NE 3 8.5 3.8R20 M Y Y 1 NE 3 9.1 5.5R21 M Y Y 1 NE 3 9.1 7.0R22 M Y Y 1 NE 3 9.0 8.3R31 M Y Y 1 NE 4 12.0 15.9R53 M Y Y 1 4 6 22.0 13.3R54 M Y Y 1 4 6 21.4 15.8R55 M Y Y 1 4 6 21.4 17.8R56 M Y Y 1 4 6 21.0 19.5R52 M Y Y 1 5 6 21.4 10.4R59 M Y Y 1 5 7 24.5 9.4R60 M Y Y 1 5 7 26.2 11.4R63 M Y Y 1 6 7 22.8 26.8R129 M Y Y 1 7 14 44.5 18.7R130 M Y Y 1 7 14 44.5 20.2R131 M Y Y 1 7 14 44.5 21.9R132 M Y Y 1 7 14 44.5 23.0R133 M Y Y 1 7 14 44.5 24.0R134 M Y Y 1 7 14 44.5 25.4R135 M Y Y 1 7 14 44.5 27.1R146 M Y Y 1 7 15 49.7 19.6R147 M Y Y 1 7 15 49.7 22.4R148 M Y Y 1 7 15 49.4 24.4R149 M Y Y 1 7 15 49.6 26.0R46 M Y Y 1 8 5 15.2 22.6R47 M Y Y 1 8 5 15.2 24.1R48 M Y Y 1 8 5 14.9 25.4R28 M Y Y 1 10 4 14.9 2.6R29 M Y Y 1 10 4 14.4 5.5R30 M Y Y 1 10 4 14.4 8.5R37 M Y Y 1 11 5 18.5 2.6R38 M Y Y 1 11 5 18.7 5.6R39 M Y Y 1 11 5 18.2 7.5R40 M Y Y 1 11 5 17.7 10.5R41 M Y Y 1 11 5 17.7 11.4R42 M Y Y 1 11 5 17.7 12.7R34 M Y Y 1 13 4 12.0 27.0R35 M Y Y 1 13 4 11.7 28.1R36 M Y Y 1 13 4 11.5 29.6R49 M Y Y 1 13 5 14.5 26.9R50 M Y Y 1 13 5 14.6 28.4R51 M Y Y 1 13 5 13.8 29.5R27 M Y Y 1 14 3 9.9 27.0R23 M Y Y 1 15 3 9.1 10.1R24 M Y Y 1 15 3 8.8 12.9R66 M Y Y 1 16 8 29.2 6.6R67 M Y Y 1 16 8 29.2 8.2R68 M Y Y 1 16 8 28.7 11.4R69 M Y Y 1 16 8 28.3 14.7R70 M Y Y 1 16 8 28.4 16.4R71 M Y Y 1 16 8 28.5 17.9R72 M Y Y 1 16 8 28.8 19.2R76 M Y Y 1 16 9 32.1 9.3R77 M Y Y 1 16 9 32.0 10.9R78 M Y Y 1 16 9 32.3 12.1R79 M Y Y 1 16 9 31.8 13.8R80 M Y Y 1 16 9 31.0 16.5R81 M Y Y 1 16 9 29.2 18.5R100 M Y Y 1 16 11 35.8 10.7R108 M Y Y 1 17 12 39.8 7.8R109 M Y Y 1 17 12 39.2 12.1R110 M Y Y 1 17 12 39.2 13.1R111 M Y Y 1 17 12 38.4 15.5R112 M Y Y 1 17 12 38.2 17.7R113 M Y Y 1 17 12 37.4 20.0R118 M Y Y 1 17 13 42.6 12.754   Coded  CODED LOCATION DATA     GPR GRID COORDINATES: Report Coded Loc. GPR Data   Grid Grid From From ID# Site Mark? Sign? Grp     Refer Row E  N  R119 M Y Y 1 17 13 42.4 15.3R120 M Y Y 1 17 13 42.4 17.5R124 M Y Y 1 17 14 45.4 8.0R125 M Y Y 1 17 14 45.4 11.5R126 M Y Y 1 17 14 45.4 14.0R127 M Y Y 1 17 14 45.4 15.8R128 M Y Y 1 17 14 45.2 17.4R99 M Y Y 1 18 11 35.8 5.4R123 M Y Y 1 18 14 45.4 5.8C01 MC Y Y 1 3 1 4.0 5.5C02 MC Y Y 1 3 1 4.0 7.5C03 MC Y Y 1 3 1 4.4 9.9C04 MC Y Y 1 3 1 4.7 12.1C05 MC Y Y 1 3 1 4.2 17.1C06 MC Y Y 1 3 1 4.2 17.1C07 MC Y Y 1 3 2 7.3 12.1C08 MC Y Y 1 3 2 7.5 15.6C09 MC Y Y 1 3 2 6.8 17.5C10 MC Y Y 1 3 2 7.2 19.5C11 MC Y Y 1 4/3 3 10.8 4.1C12 MC Y Y 1 4/3 3 11.0 7.1C13 MC Y Y 1 4/3 3 11.0 10.8C14 MC Y Y 1 4/3 3 11.0 12.1C15 MC Y Y 1 4/3 3 11.0 13.3C17 MC Y Y 1 4 4 15.2 7.2C18 MC Y Y 1 4 4 14.7 8.9C16 MC Y Y 1 5/3 3 9.7 18.1K01 K Y Y 3 1 1 7.1 6.3K02 K Y Y 3 1 1 7.0 8.5K03 K Y Y 3 1 1 6.9 10.3K04 K Y Y 3 1 1 6.9 12.3K05 K Y Y 3 1 1 6.7 14.6K06 K Y Y 3 6 1 6.5 17.0K07 K Y Y 3 6 2 10.1 15.4K08 K Y Y 3 6 2 9.9 17.2K09 K Y Y 3 6 2 9.8 19.0K10 K Y Y 3 1 3 14.4 1.0K11 K Y Y 3 1 3 14.3 2.2K12 K Y Y 3 1 3 14.1 3.0K13 K Y Y 3 1 3 14.0 3.8K14 K Y Y 3 1 3 13.9 5.0K15 K Y Y 3 1 3 13.8 6.0K16 K Y Y 3 1 3 13.8 8.0K17 K Y Y 3 1 3 13.7 10.0K18 K Y Y 3 1 3 13.5 11.5K19 K Y Y 3 1 3 13.4 13.5K20 K Y Y 3 6 3 13.3 15.8K22 K Y Y 3 7 4 21.7 23.6K25 K Y Y 3 3 5 29.6 0.8K28 K Y Y 3 3 5 29.6 7.2K31 K Y Y 3 3 6 33.5 1.5K32 K Y Y 3 3 6 33.5 2.8K35 K Y Y 3 4 7 37.2 1.5K36 K Y Y 3 4 7 37.2 2.9K37 K Y Y 3 4 7 37.2 4.3K40 K Y Y 3 4 7 37.2 9.8K43 K Y Y 3 4 8 41.0 4.5K47 K Y Y 3 5 9 44.7 1.4K48 K Y Y 3 5 9 44.7 3.5K49 K Y Y 3 5 9 44.7 4.7K50 K Y Y 3 5 10 48.3 1.8K51 K Y Y 3 5 10 48.3 4.2 55   Coded  CODED LOCATION DATA     GPR GRID COORDINATES: Report Coded Loc. GPR Data   Grid Grid From From ID# Site Mark? Sign? Grp     Refer Row E  N  M31 M N Y 2 4 7 26.5 13.5M32 M N Y 2 4 7 26.5 15.0M33 M N Y 2 4 7 26.0 16.5M34 M N Y 2 4 7 25.0 19.7M25 M N Y 2 5 6 21.0 5.0M26 M N Y 2 5 6 21.0 6.3M27 M N Y 2 5 6 21.0 7.4M28 M N Y 2 5 6 21.0 8.5M19 M N Y 2 5 7 26.5 5.0M20 M N Y 2 5 7 24.0 3.5M21 M N Y 2 5 7 24.0 5.0R58 M N Y 2 5 7 24.5 6.9M22 M N Y 2 5 7 24.5 7.5M23 M N Y 2 5 7 24.0 10.5M46 M N Y 2 6 7 22.0 23.3M47 M N Y 2 6 7 22.0 24.5M48 M N Y 2 6 6 20.0 24.5M49 M N Y 2 6 6 20.0 28.5M09 M N Y 2 7 13 42.0 20.0M10 M N Y 2 7 13 42.0 22.0M11 M N Y 2 7 13 42.0 23.5M12 M N Y 2 7 13 42.0 25.0M13a M N Y 2 7 13 42.0 9.5M50 M N Y 2 8 4 12.0 23.0M51 M N Y 2 8 4 12.0 24.5M39 M N Y 2 9 2 6.3 12.0M35 M N Y 2 10 4 11.8 7.0M36 M N Y 2 10 4 14.0 10.5M37 M N Y 2 10 4 11.8 10.5M38 M N Y 2 10 4 11.8 12.5M29 M N Y 2 11 5 18.0 4.0M30 M N Y 2 11 5 18.0 8.5M24 M N Y 2 11 6 21.0 2.5M52 M N Y 2 14 3 9.0 29.0M40 M N Y 2 15 1 4.0 13.5M41 M N Y 2 15 2 6.3 14.0M42 M N Y 2 15 1 4.3 15.5M43 M N Y 2 15 2 6.3 16.5M44 M N Y 2 15 1 4.0 17.5M45 M N Y 2 15 1 4.0 19.0M18 M N Y 2 16 8 29.0 12.5M13b M N Y 2 16 11 9.0 12.5M14 M N Y 2 16 11 35.5 9.5M15 M N Y 2 16 11 35.5 12.5M16 M N Y 2 16 11 35.5 14.0M17 M N Y 2 16 11 35.0 18.5M06 M N Y 2 17 12 40.0 9.5M08 M N Y 2 17 12 40.0 18.5M03 M N Y 2 17 13 43.0 8.0M05 M N Y 2 17 14 45.0 7.0M07 M N Y 2 17 14 45.0 9.0M02 M N Y 2 18 13 42.0 3.5M04 M N Y 2 18 12 41.0 5.5D01 MC N Y 2 4 3 11.0 2.0D02 MC N Y 2 4 3 11.0 9.0D03 MC N Y 2 5/3 3 11.0 16.0L01 K N Y 4 1 0.5 4.2 7.0L02 K N Y 4 1 0.5 4.2 8.5L03 K N Y 4 1 0.5 4.2 11.7L04 K N Y 4 1 0.5 4.1 13.0L05 K N Y 4 1 0.5 4.0 14.2L06 K N Y 4 1 0.5 4.0 15.2L07 K N Y 4 2 3.5 18.3 2.0L08 K N Y 4 2 3.5 18.7 3.5L09 K N Y 4 2 3.5 18.2 7.4L10 K N Y 4 2 3.5 17.7 9.4L11 K N Y 4 2 3.5 17.8 11.5L12 K N Y 4 2 4 21.5 5.0L13 K N Y 4 2 4 21.5 8.1L14 K N Y 4 2 4 21.5 11.5 56   Coded  CODED LOCATION DATA     GPR GRID COORDINATES: Report Coded Loc. GPR Data   Grid Grid From From ID# Site Mark? Sign? Grp     Refer Row E  N  L15 K N Y 4 2 4 21.5 12.5L16 K N Y 4 2 4.5 24.5 2.1L17 K N Y 4 6 0.5 4.1 17.1L18 K N Y 4 6 0.5 5.4 22.9L19 K N Y 4 6 0.5 4.6 24.2L20 K N Y 4 6 1 8.1 23.2L21 K N Y 4 6 1 8.1 24.2L22 K N Y 4 6 3 13.5 19.0L23 K N Y 4 6 3 13.2 22.2L24 K N Y 4 6 2 10.6 24.2L25 K N Y 4 6 0.5 4.6 18.5L26 K N Y 4 6 3.5 16.9 22.2L27 K N Y 4 7 4 23.0 20.0L28 K N Y 4 7 4.5 26.0 24.0L29 K N Y 4 8 --- 27.8 17.5L30 K N Y 4 8 --- 29.5 19.0L31 K N Y 4 8 --- 29.5 20.8L32 K N Y 4 8 --- 32.2 21.0L33 K N Y 4 8 --- 31.3 23.7L34 K N Y 4 8 --- 34.2 23.0L35 K N Y 4 8 --- 35.5 23.8L36 K N Y 4 9 --- 37.8 23.9L37 K N Y 4 9 --- 39.5 19.6L38 K N Y 4 9 --- 41.0 24.2L39 K N Y 4 12 3 13.1 25.1L40 K N Y 4 12 3 13.5 27.5L41 K N Y 4 11 0.5 7.0 27.2L42 K N Y 4 11 1 7.5 27.2L43 K N Y 4 14 --- 29.0 29.5L44 K N Y 4 12 2 11.0 26.0L45 K N Y 4 12 2 11.0 30.0L46 K N Y 4 12 2 11.5 21.3L47 K N Y 4 12 3 13.1 26.4L48 K N Y 4 12 3 12.9 29.5L49 K N Y 4 12 3 13.1 30.7L50 K N Y 4 12 3 13.0 32.9L51 K N Y 4 12 3.5 16.9 26.2L52 K N Y 4 12 3.5 16.9 27.2L53 K N Y 4 12 3.5 16.7 30.6L54 K N Y 4 14 --- 29.5 33.3L55 K N Y 4 13 4.5 25.0 30.4L56 K N Y 4 13 4 21.7 29.5L57 K N Y 4 13 4 21.8 32.5L58 K N Y 4 13 4.5 26.0 25.9L59 K N Y 4 15 6 33.0 28.6L60 K N Y 4 13 4.5 26.0 32.9L61 K N Y 4 1 2 10.5 14.2L62 K N Y 4 2 4.5 24.5 4.5L63 K N Y 4 2 4.5 24.5 6.4L64 K N Y 4 2 4.5 24.5 8.9L65 K N Y 4 6 0.5 4.5 19.8L66 K N Y 4 7 4.5 24.6 21.3L67 K N Y 4 8 --- 36.0 21.7L68 K N Y 4 8 --- 28.5 22.4L69 K N Y 4 8 --- 29.2 21.8L70 K N Y 4 8 --- 32.3 21.9L71 K N Y 4 15 6 34.0 31.2 57   Coded  CODED LOCATION DATA     GPR GRID COORDINATES: Report Coded Loc. GPR Data   Grid Grid From From ID# Site Mark? Sign? Grp     Refer Row E  N  LP01 K N Y 4 2 3.5 18.5 5.5LP02 K N Y 4 2 3.5 18.0 13.3LP03 K N Y 4 2 4 21.5 1.9LP04 K N Y 4 2 4 21.6 3.5LP05 K N Y 4 2 4 21.6 6.5LP06 K N Y 4 2 4 21.6 9.9LP07 K N Y 4 4 8 1.0 14.6LP08 K N Y 4 5 10 47.2 6.8LP09 K N Y 4 6 3 15.0 24.4LP10 K N Y 4 7 --- 19.1 24.5LP11 K N Y 4 7 4 21.5 24.5LP12 K N Y 4 7 --- 23.5 24.5LP13 K N Y 4 8 --- 31.7 24.2LP15 K N Y 4 8 --- 28.2 24.4LP14 K N Y 4 8 --- 30.3 24.5LP16 K N Y 4 9 --- 38.2 17.8LP17 K N Y 4 10 9 44.0 17.4LP18 K N Y 4 10 9 44.1 18.9LP19 K N Y 4 10 9 44.3 23.8LP20 K N Y 4 11 1 8.1 25.6LP21 K N Y 4 11 0.5 4.8 28.7LP22 K N Y 4 11 0.5 5.4 30.7LP23 K N Y 4 12 3.5 17.0 28.5LP24 K N Y 4 12 3.5 14.4 28.5LP25 K N Y 4 12 3.5 17.0 29.5LP26 K N Y 4 12 3.5 14.0 31.9LP27 K N Y 4 12 3.5 16.8 32.4LP28 K N Y 4 12 2 11.5 32.5LP29 K N Y 4 15 6 34.0 30.0LP30 K N Y 4 16 --- 36.6 25.9LP31 K N Y 4 16 --- 40.2 26.4LP32 K N Y 4 18 --- 61.0 15.0  58  Appendix B: Coded Burial Location BCI Results Raw Data    BURIAL CONFIDENCE SURFACE CHARACTERISTICS GPR SIGNAL DATA EVALUATION Coded INDEX CALCULATIONS 1 2 3 4 5 6 7 8 9 10 11 12 Report TOT Surf GPR GPR GPR   Marker ? In Row Surf Hist Avg Sign Lgt Wid GPR Profile ID# BCI Char BCI Metr Eval Y/N Anch Row Spc Ind Knw Dep Area X Y Int Layer WEIGHT 1.00 0.40 0.60 0.20 0.40 0.20 0.02 0.10 0.02 0.03 0.03 0.05 0.05 0.05 0.05 0.20 0.20 R18 0.77 0.40 0.37 0.10 0.27 Y Y Y Y Y Y 1.10 0.40 0.80 0.50 Weak StrongR19 0.87 0.40 0.47 0.20 0.27 Y Y Y Y Y Y 1.40 2.00 2.50 0.80 Weak StrongR20 0.82 0.37 0.45 0.19 0.27 Y Y Y Y Y N 1.50 1.80 1.50 1.20 Weak StrongR21 0.66 0.40 0.26 0.13 0.13 Y Y Y Y Y Y 1.50 0.70 1.40 0.50 Weak WeakR22 0.84 0.40 0.44 0.18 0.27 Y Y Y Y Y Y 1.60 1.12 1.40 0.80 Weak StrongR31 0.84 0.40 0.44 0.18 0.27 Y Y Y Y Y Y 1.00 1.76 1.10 1.60 Weak StrongR53 0.76 0.40 0.36 0.16 0.20 Y Y Y Y Y Y 1.10 1.10 1.00 1.10 Weak ModerR54 0.91 0.40 0.51 0.18 0.33 Y Y Y Y Y Y 1.30 1.30 1.30 1.00 Moder StrongR55 0.80 0.40 0.40 0.20 0.20 Y Y Y Y Y Y 1.30 1.80 1.80 1.00 Weak ModerR56 1.00 0.40 0.60 0.20 0.40 Y Y Y Y Y Y 1.50 3.36 2.80 1.20 Strong StrongR52 0.91 0.37 0.54 0.14 0.40 Y Y Y Y Y N 1.10 0.72 0.90 0.80 Strong StrongR59 0.98 0.38 0.60 0.20 0.40 Y N Y Y Y Y 1.50 2.00 2.50 0.80 Strong StrongR60 1.00 0.40 0.60 0.20 0.40 Y Y Y Y Y Y 1.10 2.21 1.70 1.30 Strong StrongR63 1.00 0.40 0.60 0.20 0.40 Y Y Y Y Y Y 1.10 3.75 2.50 1.50 Strong StrongR129 0.90 0.37 0.53 0.20 0.33 Y Y Y Y Y N 1.00 1.92 2.40 0.80 Strong ModerR130 0.79 0.37 0.42 0.15 0.27 Y Y Y Y Y N 1.00 0.90 1.50 0.60 Strong WeakR131 0.97 0.37 0.60 0.20 0.40 Y Y Y Y Y N 1.00 3.12 2.60 1.20 Strong StrongR132 0.90 0.37 0.53 0.20 0.33 Y Y Y Y Y N 1.00 2.40 2.40 1.00 Strong ModerR133 0.88 0.37 0.51 0.18 0.33 Y Y Y Y Y N 1.00 1.19 1.70 0.70 Strong ModerR134 0.90 0.37 0.53 0.13 0.40 Y Y Y Y Y N 1.00 0.65 1.30 0.50 Strong StrongR135 0.97 0.37 0.60 0.20 0.40 Y Y Y Y Y N 1.30 2.73 2.10 1.30 Strong StrongR146 0.93 0.40 0.53 0.20 0.33 Y Y Y Y Y Y 1.10 3.57 2.10 1.70 Moder StrongR147 1.00 0.40 0.60 0.20 0.40 Y Y Y Y Y Y 1.10 2.20 2.00 1.10 Strong StrongR148 1.00 0.40 0.60 0.20 0.40 Y Y Y Y Y Y 1.10 2.80 2.00 1.40 Strong StrongR149 0.93 0.40 0.53 0.20 0.33 Y Y Y Y Y Y 1.10 3.15 2.10 1.50 Moder StrongR46 0.82 0.40 0.42 0.15 0.27 Y Y Y Y Y Y 1.10 0.80 1.60 0.50 Moder ModerR47 0.93 0.40 0.53 0.20 0.33 Y Y Y Y Y Y 1.10 1.71 1.90 0.90 Moder StrongR48 0.91 0.35 0.56 0.16 0.40 Y N Y Y Y N 1.10 1.08 1.80 0.60 Strong StrongR28 1.00 0.40 0.60 0.20 0.40 Y Y Y Y Y Y 1.10 3.20 3.20 1.00 Strong StrongR29 1.00 0.40 0.60 0.20 0.40 Y Y Y Y Y Y 1.50 3.25 2.50 1.30 Strong StrongR30 0.98 0.40 0.58 0.18 0.40 Y Y Y Y Y Y 1.50 1.20 1.50 0.80 Strong StrongR37 1.00 0.40 0.60 0.20 0.40 Y Y Y Y Y Y 1.50 3.30 2.20 1.50 Strong StrongR38 0.99 0.40 0.59 0.19 0.40 Y Y Y Y Y Y 0.90 4.50 2.50 1.80 Strong StrongR39 0.97 0.37 0.60 0.20 0.40 Y Y Y Y Y N 1.20 1.98 1.80 1.10 Strong StrongR40 0.83 0.40 0.43 0.16 0.27 Y Y Y Y Y Y 1.10 1.08 1.80 0.60 Weak StrongR41 0.73 0.40 0.33 0.13 0.20 Y Y Y Y Y Y 1.20 0.72 1.20 0.60 Weak ModerR42 0.75 0.40 0.35 0.15 0.20 Y Y Y Y Y Y 1.20 0.85 1.70 0.50 Weak ModerR34 0.93 0.40 0.53 0.13 0.40 Y Y Y Y Y Y 1.00 0.64 0.80 0.80 Strong StrongR35 0.99 0.40 0.59 0.19 0.40 Y Y Y Y Y Y 1.10 1.95 1.30 1.50 Strong StrongR36 0.91 0.40 0.51 0.18 0.33 Y Y Y Y Y Y 1.10 1.20 1.50 0.80 Moder StrongR49 0.97 0.37 0.60 0.20 0.40 Y Y Y Y N Y 1.00 1.87 1.70 1.10 Strong StrongR50 0.87 0.40 0.47 0.20 0.27 Y Y Y Y Y Y 1.20 2.04 1.70 1.20 Weak StrongR51 0.87 0.40 0.47 0.20 0.27 Y Y Y Y Y Y 1.40 2.40 2.00 1.20 Weak StrongR27 0.79 0.40 0.39 0.13 0.27 Y Y Y Y Y Y 0.70 0.91 1.30 0.70 Moder ModerR23 0.99 0.40 0.59 0.19 0.40 Y Y Y Y Y Y 0.90 2.20 2.00 1.10 Strong StrongR24 1.00 0.40 0.60 0.20 0.40 Y Y Y Y Y Y 1.20 2.60 2.00 1.30 Strong StrongR66 0.82 0.35 0.47 0.20 0.27 Y N Y Y Y N 1.10 1.68 2.10 0.80 Moder ModerR67 0.90 0.37 0.53 0.20 0.33 Y Y Y Y Y N 1.60 2.70 1.80 1.50 Strong ModerR68 0.93 0.40 0.53 0.20 0.33 Y Y Y Y Y Y 1.30 1.84 2.30 0.80 Strong ModerR69 0.90 0.40 0.50 0.16 0.33 Y Y Y Y Y Y 0.50 1.87 1.70 1.10 Moder StrongR70 0.87 0.40 0.47 0.20 0.27 Y Y Y Y Y Y 1.30 3.22 2.30 1.40 Strong WeakR71 0.93 0.40 0.53 0.20 0.33 Y Y Y Y Y Y 1.00 2.40 3.00 0.80 Strong ModerR72 0.91 0.40 0.51 0.18 0.33 Y Y Y Y Y Y 0.70 1.76 2.20 0.80 Strong ModerR76 0.83 0.37 0.46 0.19 0.27 Y Y Y Y Y N 1.30 1.50 1.50 1.00 Moder ModerR77 0.80 0.40 0.40 0.20 0.20 Y Y Y Y Y Y 1.20 2.52 2.10 1.20 Weak ModerR78 0.78 0.40 0.38 0.18 0.20 Y Y Y Y Y Y 1.30 1.30 1.30 1.00 Weak ModerR79 0.79 0.40 0.39 0.19 0.20 Y Y Y Y Y Y 1.10 1.56 1.30 1.20 Weak ModerR80 0.74 0.28 0.46 0.19 0.27 Y Y N N Y Y 1.30 3.64 1.30 2.80 Moder ModerR81 0.72 0.28 0.44 0.18 0.27 Y Y N N Y Y 1.30 2.50 1.00 2.50 Moder ModerR100 1.00 0.40 0.60 0.20 0.40 Y Y Y Y Y Y 1.10 3.50 2.50 1.40 Strong StrongR108 0.90 0.37 0.53 0.20 0.33 Y Y Y N Y N 1.50 2.00 2.00 1.00 Strong StrongR109 0.78 0.37 0.41 0.14 0.27 Y Y Y Y Y N 0.70 1.04 1.30 0.80 Moder ModerR110 0.76 0.37 0.39 0.19 0.20 Y Y Y Y Y N 0.80 1.80 2.00 0.90 Moder WeakR111 1.00 0.40 0.60 0.20 0.40 Y Y Y Y Y Y 1.10 2.34 1.80 1.30 Strong StrongR112 0.93 0.40 0.53 0.20 0.33 Y Y Y Y Y Y 1.10 2.16 1.80 1.20 Moder StrongR113 1.00 0.40 0.60 0.20 0.40 Y Y Y Y Y Y 1.10 2.43 2.70 0.90 Strong StrongR118 0.87 0.37 0.50 0.10 0.40 Y Y Y Y Y N 1.10 0.38 0.25 1.50 Strong StrongR119 0.87 0.40 0.47 0.20 0.27 Y Y Y Y Y Y 1.30 3.64 2.80 1.30 Moder ModerR120 0.97 0.37 0.60 0.20 0.40 Y Y Y Y Y N 1.40 3.57 2.10 1.70 Strong StrongR124 0.97 0.37 0.60 0.20 0.40 Y Y Y Y Y N 1.30 2.40 2.00 1.20 Strong StrongR125 0.97 0.37 0.60 0.20 0.40 Y Y Y Y Y N 1.10 4.00 2.50 1.60 Strong StrongR126 0.96 0.37 0.59 0.19 0.40 Y Y Y Y Y N 0.90 2.75 2.50 1.10 Strong StrongR127 1.00 0.40 0.60 0.20 0.40 Y Y Y Y Y Y 1.20 2.70 1.80 1.50 Strong StrongR128 0.99 0.40 0.59 0.19 0.40 Y Y Y Y Y Y 0.90 2.10 2.10 1.00 Strong StrongR99 1.00 0.40 0.60 0.20 0.40 Y Y Y Y Y Y 1.00 3.60 3.00 1.20 Strong StrongR123 0.90 0.37 0.53 0.20 0.33 Y Y Y Y Y N 1.00 4.00 2.00 2.00 Moder Strong 59     BURIAL CONFIDENCE SURFACE CHARACTERISTICS GPR SIGNAL DATA EVALUATION Coded INDEX CALCULATIONS 1 2 3 4 5 6 7 8 9 10 11 12 Report TOT Surf GPR GPR GPR   Marker ? In Row Surf Hist Avg Sign Lgt Wid GPR Profile ID# BCI Char BCI Metr Eval Y/N Anch Row Spc Ind Knw Dep Area X Y Int Layer WEIGHT 1.00 0.40 0.60 0.20 0.40 0.20 0.02 0.10 0.02 0.03 0.03 0.05 0.05 0.05 0.05 0.20 0.20 C01 0.94 0.35 0.59 0.19 0.40 Y N Y Y Y N 0.80 2.16 1.80 1.20 Strong StrongC02 0.97 0.37 0.60 0.20 0.40 Y Y Y Y Y N 1.00 1.92 1.60 1.20 Strong StrongC03 0.97 0.37 0.60 0.20 0.40 Y Y Y Y Y N 1.00 2.16 1.80 1.20 Strong StrongC04 1.00 0.40 0.60 0.20 0.40 Y Y Y Y Y Y 1.00 2.52 2.10 1.20 Strong StrongC05 0.65 0.37 0.28 0.15 0.13 Y Y Y Y Y N 1.50 0.80 1.00 0.80 Weak WeakC06 0.59 0.35 0.24 0.11 0.13 Y N Y Y Y N 1.80 0.56 0.80 0.70 Weak WeakC07 0.79 0.40 0.39 0.13 0.27 Y Y Y Y Y Y 1.30 0.72 0.80 0.90 Weak StrongC08 0.73 0.40 0.33 0.13 0.20 Y Y Y Y Y Y 1.60 0.54 0.60 0.90 Weak ModerC09 0.58 0.37 0.21 0.08 0.13 Y Y Y Y Y N 1.30 0.20 0.40 0.50 Weak WeakC10 0.56 0.37 0.19 0.06 0.13 Y Y Y Y Y N 1.70 0.06 0.20 0.30 Weak WeakC11 1.00 0.40 0.60 0.20 0.40 Y Y Y Y Y Y 1.30 3.12 2.40 1.30 Strong StrongC12 1.00 0.40 0.60 0.20 0.40 Y Y Y Y Y Y 1.30 3.12 2.40 1.30 Strong StrongC13 0.98 0.40 0.58 0.18 0.40 Y Y Y Y Y Y 1.30 1.17 1.30 0.90 Strong StrongC14 1.00 0.40 0.60 0.20 0.40 Y Y Y Y Y Y 1.30 2.40 2.00 1.20 Strong StrongC15 1.00 0.40 0.60 0.20 0.40 Y Y Y Y Y Y 1.00 2.64 2.40 1.10 Strong StrongC17 1.00 0.40 0.60 0.20 0.40 Y Y Y Y Y Y 1.20 2.40 2.00 1.20 Strong StrongC18 1.00 0.40 0.60 0.20 0.40 Y Y Y Y Y Y 1.20 3.00 2.50 1.20 Strong StrongC16 0.59 0.40 0.19 0.06 0.13 Y Y Y Y Y Y 1.80 0.12 0.60 0.20 Weak WeakK01 0.59 0.40 0.19 0.06 0.13 Y Y Y Y Y Y 0.70 0.30 1.00 0.30 Weak WeakK02 0.87 0.40 0.47 0.14 0.33 Y Y Y Y Y Y 0.60 1.10 1.00 1.10 Strong ModerK03 0.93 0.40 0.53 0.13 0.40 Y Y Y Y Y Y 0.60 0.90 1.50 0.60 Strong StrongK04 0.90 0.40 0.50 0.16 0.33 Y Y Y Y Y Y 0.60 1.50 1.50 1.00 Strong ModerK05 0.76 0.40 0.36 0.16 0.20 Y Y Y Y Y Y 0.50 1.44 1.60 0.90 Weak ModerK06 0.99 0.40 0.59 0.19 0.40 Y Y Y Y Y Y 0.90 2.00 2.00 1.00 Strong StrongK07 0.96 0.40 0.56 0.16 0.40 Y Y Y Y Y Y 0.70 1.43 1.30 1.10 Strong StrongK08 0.65 0.40 0.25 0.05 0.20 Y Y Y Y Y Y 0.50 0.27 0.90 0.30 Weak ModerK09 0.59 0.40 0.19 0.06 0.13 Y Y Y Y Y Y 0.80 0.24 0.80 0.30 Weak WeakK10 0.87 0.40 0.47 0.20 0.27 Y Y Y Y Y Y 1.00 1.76 2.20 0.80 Moder ModerK11 0.79 0.40 0.39 0.19 0.20 Y Y Y Y Y Y 0.90 2.00 2.00 1.00 Weak ModerK12 0.77 0.40 0.37 0.10 0.27 Y Y Y Y Y Y 0.90 0.50 1.00 0.50 Moder ModerK13 0.92 0.40 0.52 0.19 0.33 Y Y Y Y Y Y 0.80 1.60 2.00 0.80 Moder StrongK14 0.90 0.40 0.50 0.10 0.40 Y Y Y Y Y Y 0.70 0.77 1.10 0.70 Strong StrongK15 0.98 0.40 0.58 0.18 0.40 Y Y Y Y Y Y 0.80 1.54 2.20 0.70 Strong StrongK16 0.96 0.40 0.56 0.16 0.40 Y Y Y Y Y Y 0.80 1.30 1.30 1.00 Strong StrongK17 0.94 0.40 0.54 0.14 0.40 Y Y Y Y Y Y 0.70 1.21 1.10 1.10 Strong StrongK18 0.82 0.40 0.42 0.15 0.27 Y Y Y Y Y Y 1.00 1.00 1.00 1.00 Weak StrongK19 1.00 0.40 0.60 0.20 0.40 Y Y Y Y Y Y 1.00 2.40 2.00 1.20 Strong StrongK20 0.84 0.40 0.44 0.18 0.27 Y Y Y Y Y Y 0.60 1.80 1.80 1.00 Moder ModerK22 0.94 0.40 0.54 0.14 0.40 Y Y Y Y Y Y 0.80 1.05 1.50 0.70 Strong StrongK25 1.00 0.40 0.60 0.20 0.40 Y Y Y Y Y Y 1.10 3.25 2.50 1.30 Strong StrongK28 0.89 0.40 0.49 0.09 0.40 Y Y Y Y Y Y 1.30 0.14 0.20 0.70 Strong StrongK31 0.93 0.40 0.53 0.20 0.33 Y Y Y Y Y Y 1.10 1.70 1.70 1.00 Moder StrongK32 0.93 0.40 0.53 0.20 0.33 Y Y Y Y Y Y 1.10 1.60 1.60 1.00 Moder StrongK35 0.54 0.40 0.14 0.01 0.13 Y Y Y Y Y Y 0.50 0.06 0.30 0.20 Weak WeakK36 0.54 0.40 0.14 0.01 0.13 Y Y Y Y Y Y 0.50 0.06 0.30 0.20 Weak WeakK37 0.54 0.40 0.14 0.01 0.13 Y Y Y Y Y Y 0.50 0.06 0.30 0.20 Weak WeakK40 0.84 0.40 0.44 0.04 0.40 Y Y Y Y Y Y 0.60 0.06 0.20 0.30 Strong StrongK43 0.66 0.40 0.26 0.06 0.20 Y Y Y Y Y Y 0.80 0.20 0.40 0.50 Moder WeakK47 0.91 0.40 0.51 0.18 0.33 Y Y Y Y Y Y 1.10 1.12 1.40 0.80 Strong ModerK48 1.00 0.40 0.60 0.20 0.40 Y Y Y Y Y Y 1.00 2.21 1.70 1.30 Strong StrongK49 0.88 0.40 0.48 0.15 0.33 Y Y Y Y Y Y 1.10 0.90 1.00 0.90 Strong ModerK50 0.99 0.40 0.59 0.19 0.40 Y Y Y Y Y Y 1.20 2.25 1.50 1.50 Strong StrongK51 0.98 0.40 0.58 0.18 0.40 Y Y Y Y Y Y 0.90 1.50 1.50 1.00 Strong StrongM31 0.70 0.12 0.58 0.18 0.40 N N Y Y N N 0.70 3.45 2.30 1.50 Strong StrongM32 0.60 0.12 0.48 0.15 0.33 N N Y Y N N 1.30 0.88 1.10 0.80 Moder StrongM33 0.56 0.12 0.44 0.18 0.27 N N Y Y N N 1.50 1.20 1.50 0.80 Weak StrongM34 0.40 0.00 0.40 0.20 0.20 N N N N N N 1.00 2.70 1.80 1.50 Weak ModerM25 0.38 0.12 0.26 0.06 0.20 N N Y Y N N 0.90 0.21 0.70 0.30 Weak ModerM26 0.65 0.12 0.53 0.20 0.33 N N Y Y N N 1.40 2.00 2.00 1.00 Moder StrongM27 0.36 0.12 0.24 0.11 0.13 N N Y Y N N 1.20 0.50 1.00 0.50 Weak WeakM28 0.35 0.12 0.23 0.10 0.13 N N Y Y N N 1.20 0.48 1.20 0.40 Weak WeakM19 0.56 0.03 0.53 0.20 0.33 N N N N Y N 1.40 2.34 1.80 1.30 Moder StrongM20 0.38 0.12 0.26 0.13 0.13 N N Y Y N N 1.30 0.64 0.80 0.80 Weak WeakM21 0.59 0.12 0.47 0.14 0.33 N N Y Y N N 1.30 0.84 1.20 0.70 Moder StrongR58 0.78 0.18 0.60 0.20 0.40 N N Y Y Y Y 1.40 2.50 2.50 1.00 Strong StrongM22 0.68 0.12 0.56 0.16 0.40 N N Y Y N N 1.30 1.25 2.50 0.50 Strong StrongM23 0.35 0.12 0.23 0.10 0.13 N N Y Y N N 0.80 0.60 1.20 0.50 Weak WeakM46 0.67 0.12 0.55 0.15 0.40 N N Y Y N N 1.00 0.95 1.90 0.50 Strong StrongM47 0.68 0.12 0.56 0.16 0.40 N N Y Y N N 1.00 1.04 1.30 0.80 Strong StrongM48 0.53 0.12 0.41 0.14 0.27 N N Y Y N N 1.20 0.80 2.00 0.40 Moder ModerM49 0.71 0.12 0.59 0.19 0.40 N N Y Y N N 1.00 1.82 1.30 1.40 Strong StrongM09 0.72 0.12 0.60 0.20 0.40 N N Y Y N N 1.10 2.60 2.00 1.30 Strong StrongM10 0.65 0.12 0.53 0.20 0.33 N N Y Y N N 1.30 2.38 1.70 1.40 Moder StrongM11 0.65 0.12 0.53 0.20 0.33 N N Y Y N N 1.00 2.00 2.00 1.00 Moder StrongM12 0.72 0.12 0.60 0.20 0.40 N N Y Y N N 1.20 2.16 2.40 0.90 Strong StrongM13a 0.70 0.12 0.58 0.18 0.40 N N Y Y N N 1.20 1.19 1.70 0.70 Strong StrongM50 0.53 0.12 0.41 0.14 0.27 N N Y Y N N 1.40 0.96 0.80 1.20 Moder ModerM51 0.53 0.12 0.41 0.14 0.27 N N Y Y N N 1.40 1.04 0.80 1.30 Moder ModerM39 0.71 0.12 0.59 0.19 0.40 N N Y Y __N_ N 1.00 1.40 1.40 1.00 Strong Strong 60     BURIAL CONFIDENCE SURFACE CHARACTERISTICS GPR SIGNAL DATA EVALUATION Coded INDEX CALCULATIONS 1 2 3 4 5 6 7 8 9 10 11 12 Report TOT Surf GPR GPR GPR   Marker ? In Row Surf Hist Avg Sign Lgt Wid GPR Profile ID# BCI Char BCI Metr Eval Y/N Anch Row Spc Ind Knw Dep Area X Y Int Layer WEIGHT 1.00 0.40 0.60 0.20 0.40 0.20 0.02 0.10 0.02 0.03 0.03 0.05 0.05 0.05 0.05 0.20 0.20 K48 1.00 0.40 0.60 0.20 0.40 Y Y Y Y Y Y 1.00 2.21 1.70 1.30 Strong StrongM35 0.35 0.12 0.23 0.10 0.13 N N Y Y N N 1.50 0.40 0.80 0.50 Weak WeakM36 0.67 0.12 0.55 0.15 0.40 N N Y Y N N 1.40 0.91 1.30 0.70 Strong StrongM37 0.46 0.12 0.34 0.08 0.27 N N Y Y N N 1.30 0.20 0.50 0.40 Moder ModerM38 0.59 0.12 0.47 0.20 0.27 N N Y Y N N 1.30 2.30 2.30 1.00 Moder ModerM29 0.68 0.12 0.56 0.16 0.40 N N Y Y N N 0.80 1.17 1.30 0.90 Strong StrongM30 0.50 0.12 0.38 0.11 0.27 N N Y Y N N 1.10 0.56 1.40 0.40 Weak StrongM24 0.68 0.12 0.56 0.16 0.40 N N Y Y N N 0.90 1.17 1.30 0.90 Strong StrongM52 0.38 0.12 0.26 0.13 0.13 N N Y Y N N 0.70 1.00 1.00 1.00 Weak WeakM40 0.52 0.12 0.40 0.20 0.20 N N Y Y N N 1.40 2.34 1.80 1.30 Weak ModerM41 0.52 0.12 0.40 0.20 0.20 N N Y Y N N 1.00 2.00 2.00 1.00 Weak ModerM42 0.65 0.12 0.53 0.20 0.33 N N Y Y N N 1.40 3.60 1.80 2.00 Moder StrongM43 0.71 0.12 0.59 0.19 0.40 N N Y Y N N 1.30 3.00 1.50 2.00 Strong StrongM44 0.71 0.12 0.59 0.19 0.40 N N Y Y N N 1.30 2.08 1.30 1.60 Strong StrongM45 0.50 0.12 0.38 0.18 0.20 N N Y Y N N 1.50 1.20 1.50 0.80 Weak ModerM18 0.72 0.12 0.60 0.20 0.40 N N Y Y N N 1.20 2.00 2.00 1.00 Strong StrongM13b 0.62 0.12 0.50 0.10 0.40 N N Y Y N N 0.30 1.00 2.50 0.40 Strong StrongM14 0.72 0.12 0.60 0.20 0.40 N N Y Y N N 1.50 2.00 2.50 0.80 Strong StrongM15 0.62 0.12 0.50 0.16 0.33 N N Y Y N N 0.30 2.60 2.00 1.30 Strong ModerM16 0.58 0.12 0.46 0.19 0.27 N N Y Y N N 0.90 2.31 2.10 1.10 Moder ModerM17 0.62 0.12 0.50 0.10 0.40 N N Y Y N N 0.80 0.80 1.00 0.80 Weak ModerM06 0.72 0.12 0.60 0.20 0.40 N N Y Y N N 1.00 1.80 1.80 1.00 Strong StrongM08 0.57 0.12 0.45 0.19 0.27 N N Y Y N N 0.90 2.86 2.20 1.30 Weak StrongM03 0.60 0.00 0.60 0.20 0.40 N N N N N N 1.30 4.05 2.70 1.50 Strong StrongM05 0.72 0.12 0.60 0.20 0.40 N N Y Y N N 1.10 0.30 0.60 0.50 Weak WeakM07 0.59 0.12 0.47 0.20 0.27 N N Y Y N N 1.30 2.60 2.00 1.30 Moder ModerM02 0.40 0.00 0.40 0.20 0.20 N N N N N N 1.30 2.00 2.00 1.00 Moder WeakM04 0.47 0.00 0.47 0.20 0.27 N N N N N N 1.30 2.50 2.50 1.00 Moder ModerD01 0.57 0.12 0.45 0.05 0.40 N N Y Y N N 0.20 0.40 0.80 0.50 Strong StrongD02 0.71 0.12 0.59 0.19 0.40 N N Y Y N N 1.80 1.56 1.30 1.20 Strong StrongD03 0.41 0.12 0.29 0.09 0.20 N N Y Y N N 1.80 0.30 0.60 0.50 Weak ModerL01 0.70 0.12 0.58 0.18 0.40 N N Y Y N N 0.70 1.80 2.00 0.90 Strong StrongL02 0.58 0.12 0.46 0.19 0.27 N N Y Y N N 1.00 1.68 2.80 0.60 Moder ModerL03 0.71 0.12 0.59 0.19 0.40 N N Y Y N N 0.90 2.50 2.50 1.00 Strong StrongL04 0.72 0.12 0.60 0.20 0.40 N N Y Y N N 1.00 1.80 1.80 1.00 Strong StrongL05 0.72 0.12 0.60 0.20 0.40 N N Y Y N N 1.00 2.64 2.20 1.20 Strong StrongL06 0.72 0.12 0.60 0.20 0.40 N N Y Y N N 1.30 2.38 1.70 1.40 Strong StrongL07 0.62 0.12 0.50 0.16 0.33 N N Y Y N N 1.00 1.32 1.10 1.20 Strong ModerL08 0.66 0.12 0.54 0.14 0.40 N N Y Y N N 1.00 0.80 0.80 1.00 Strong StrongL09 0.71 0.12 0.59 0.19 0.40 N N Y Y N N 1.10 2.25 1.50 1.50 Strong StrongL10 0.72 0.12 0.60 0.20 0.40 N N Y Y N N 1.00 2.60 2.00 1.30 Strong StrongL11 0.70 0.12 0.58 0.18 0.40 N N Y Y N N 0.80 1.50 1.50 1.00 Strong StrongL12 0.68 0.12 0.56 0.16 0.40 N N Y Y N N 0.90 1.35 1.50 0.90 Strong StrongL13 0.65 0.12 0.53 0.13 0.40 N N Y Y N N 0.70 1.00 1.00 1.00 Strong StrongL14 0.67 0.12 0.55 0.15 0.40 N N Y Y N N 0.70 1.35 1.50 0.90 Strong StrongL15 0.67 0.12 0.55 0.15 0.40 N N Y Y N N 0.70 1.12 1.40 0.80 Strong StrongL16 0.42 0.12 0.30 0.10 0.20 N N Y Y N N 1.00 0.30 0.30 1.00 Weak ModerL17 0.71 0.12 0.59 0.19 0.40 N N Y Y N N 0.80 1.44 1.80 0.80 Strong StrongL18 0.71 0.12 0.59 0.19 0.40 N N Y Y N N 0.80 2.00 2.00 1.00 Strong StrongL19 0.71 0.12 0.59 0.19 0.40 N N Y Y N N 0.90 1.60 1.60 1.00 Strong StrongL20 0.71 0.12 0.59 0.19 0.40 N N Y Y N N 0.90 1.80 2.00 0.90 Strong StrongL21 0.71 0.12 0.59 0.19 0.40 N N Y Y N N 0.80 2.50 2.50 1.00 Strong StrongL22 0.70 0.12 0.58 0.18 0.40 N N Y Y N N 0.90 1.65 1.50 1.10 Strong StrongL23 0.71 0.12 0.59 0.19 0.40 N N Y Y N N 0.90 2.52 2.10 1.20 Strong StrongL24 0.72 0.12 0.60 0.20 0.40 N N Y Y N N 1.20 2.40 2.00 1.20 Strong StrongL25 0.66 0.12 0.54 0.14 0.40 N N Y Y N N 0.80 0.80 1.00 0.80 Strong StrongL26 0.59 0.00 0.59 0.19 0.40 N N N N N N 0.90 2.21 1.70 1.30 Strong StrongL27 0.58 0.12 0.46 0.06 0.40 N N Y Y N N 0.90 0.15 0.30 0.50 Strong StrongL28 0.66 0.12 0.54 0.14 0.40 N N Y Y N N 0.80 0.91 1.30 0.70 Strong StrongL29 0.56 0.00 0.56 0.16 0.40 N N N N N N 0.80 1.20 1.50 0.80 Strong StrongL30 0.60 0.00 0.60 0.20 0.40 N N N N N N 1.00 1.44 1.80 0.80 Strong StrongL31 0.55 0.00 0.55 0.15 0.40 N N N N N N 0.90 1.04 1.30 0.80 Strong StrongL32 0.53 0.00 0.53 0.13 0.40 N N N N N N 0.60 0.80 1.00 0.80 Strong StrongL33 0.59 0.00 0.59 0.19 0.40 N N N N N N 0.80 1.87 1.70 1.10 Strong StrongL34 0.55 0.00 0.55 0.15 0.40 N N N N N N 0.70 1.12 1.40 0.80 Strong StrongL35 0.59 0.00 0.59 0.19 0.40 N N N N N N 0.80 1.76 1.60 1.10 Strong StrongL36 0.42 0.00 0.42 0.09 0.33 N N N N N N 0.60 0.48 0.80 0.60 Strong ModerL37 0.49 0.00 0.49 0.09 0.40 N N N N N N 0.60 0.48 0.80 0.60 Strong StrongL38 0.50 0.00 0.50 0.10 0.40 N N N N N N 0.70 0.60 1.00 0.60 Strong StrongL39 0.66 0.12 0.54 0.14 0.40 N N Y Y N N 0.90 0.99 1.10 0.90 Strong StrongL40 0.57 0.12 0.45 0.11 0.33 N N Y Y N N 0.80 0.56 0.70 0.80 Moder StrongL41 0.70 0.12 0.58 0.18 0.40 N N Y Y N N 0.70 2.85 1.90 1.50 Strong StrongL42 0.71 0.12 0.59 0.19 0.40 N N Y Y N N 0.90 2.55 1.70 1.50 Strong StrongL43 0.59 0.00 0.59 0.19 0.40 N N N N N N 1.30 2.10 1.50 1.40 Strong StrongL44 0.71 0.12 0.59 0.19 0.40 N N Y Y N N 0.80 2.00 2.00 1.00 Strong StrongL45 0.70 0.12 0.58 0.18 0.40 N N Y Y N N 0.90 1.40 1.40 1.00 Strong StrongL46 0.70 0.12 0.58 0.18 0.40 N N Y Y N N 1.20 1.20 1.50 0.80 Strong StrongL47 0.65 0.12 0.53 0.13 0.40 N N Y Y N N 1.10 0.60 1.00 0.60 Strong StrongL48 0.63 0.12 0.51 0.11 0.40 N N Y Y N N 0.80 0.60 1.00 0.60 Strong Strong 61     BURIAL CONFIDENCE SURFACE CHARACTERISTICS GPR SIGNAL DATA EVALUATION Coded INDEX CALCULATIONS 1 2 3 4 5 6 7 8 9 10 11 12 Report TOT Surf GPR GPR GPR   Marker ? In Row Surf Hist Avg Sign Lgt Wid GPR Profile ID# BCI Char BCI Metr Eval Y/N Anch Row Spc Ind Knw Dep Area X Y Int Layer WEIGHT 1.00 0.40 0.60 0.20 0.40 0.20 0.02 0.10 0.02 0.03 0.03 0.05 0.05 0.05 0.05 0.20 0.20 L49 0.68 0.12 0.56 0.16 0.40 N N Y Y N N 1.00 1.04 1.30 0.80 Strong StrongL50 0.67 0.12 0.55 0.15 0.40 N N Y Y N N 1.20 0.90 0.90 1.00 Strong StrongL51 0.72 0.12 0.60 0.20 0.40 N N Y Y N N 1.00 2.30 2.30 1.00 Strong StrongL52 0.72 0.12 0.60 0.20 0.40 N N Y Y N N 1.00 2.30 2.30 1.00 Strong StrongL53 0.70 0.12 0.58 0.18 0.40 N N Y Y N N 1.00 1.35 1.50 0.90 Strong StrongL54 0.46 0.00 0.46 0.06 0.40 N N N N N N 0.20 0.50 1.00 0.50 Strong StrongL55 0.59 0.00 0.59 0.19 0.40 N N N N N N 1.10 1.40 2.00 0.70 Strong StrongL56 0.60 0.00 0.60 0.20 0.40 N N N N N N 1.10 2.64 2.20 1.20 Strong StrongL57 0.60 0.00 0.60 0.20 0.40 N N N N N N 1.40 3.30 2.20 1.50 Strong StrongL58 0.68 0.12 0.56 0.16 0.40 N N Y Y N N 0.60 1.50 1.50 1.00 Strong StrongL59 0.71 0.12 0.59 0.19 0.40 N N Y Y N N 1.10 1.80 1.50 1.20 Strong StrongL60 0.58 0.00 0.58 0.18 0.40 N N N N N N 1.50 1.12 1.40 0.80 Strong StrongL61 0.64 0.12 0.52 0.19 0.33 N N Y Y N N 0.80 3.20 2.00 1.60 Strong ModerL62 0.43 0.12 0.31 0.11 0.20 N N Y Y N N 1.10 0.45 1.50 0.30 Weak ModerL63 0.60 0.12 0.48 0.15 0.33 N N Y Y N N 1.00 0.80 1.00 0.80 Moder StrongL64 0.70 0.12 0.58 0.18 0.40 N N Y Y N N 0.70 2.60 2.00 1.30 Strong StrongL65 0.62 0.12 0.50 0.10 0.40 N N Y Y N N 0.80 0.50 1.00 0.50 Strong StrongL66 0.48 0.12 0.36 0.09 0.27 N N Y Y N N 0.70 0.56 0.80 0.70 Moder ModerL67 0.43 0.00 0.43 0.10 0.33 N N N N N N 0.80 0.48 0.80 0.60 Moder StrongL68 0.41 0.00 0.41 0.14 0.27 N N N N N N 0.80 0.80 1.00 0.80 Moder ModerL69 0.46 0.00 0.46 0.06 0.40 N N N N N N 0.90 0.24 0.80 0.30 Strong StrongL70 0.54 0.00 0.54 0.14 0.40 N N N N N N 0.70 1.10 1.10 1.00 Strong StrongL71 0.67 0.12 0.55 0.15 0.40 N N Y Y N N 0.80 1.04 1.30 0.80 Strong StrongLP01 0.43 0.12 0.31 0.11 0.20 N N Y Y N N 0.80 0.70 1.00 0.70 Weak ModerLP02 0.33 0.12 0.21 0.08 0.13 N N Y Y N N 0.70 0.36 0.40 0.90 Weak WeakLP03 0.42 0.12 0.30 0.10 0.20 N N Y Y N N 1.00 0.36 0.30 1.20 Weak ModerLP04 0.37 0.12 0.25 0.05 0.20 N N Y Y N N 0.50 0.18 0.30 0.60 Weak ModerLP05 0.45 0.12 0.33 0.13 0.20 N N Y Y N N 0.70 0.80 1.00 0.80 Weak ModerLP06 0.93 0.40 0.53 0.20 0.33 N N Y Y N N 0.70 1.26 1.40 0.90 Weak WeakLP07 0.29 0.12 0.17 0.04 0.13 N N Y Y N N 0.90 0.06 0.30 0.20 Weak WeakLP08 0.23 0.00 0.23 0.10 0.13 N N N N N N 1.00 0.36 0.60 0.60 Weak WeakLP09 0.26 0.00 0.26 0.13 0.13 N N N N N N 0.70 1.00 1.00 1.00 Weak WeakLP10 0.43 0.00 0.43 0.10 0.33 N N N N N N 1.00 0.40 0.80 0.50 Moder StrongLP11 0.29 0.00 0.29 0.09 0.20 N N N N N N 1.00 0.33 1.10 0.30 Moder WeakLP12 0.28 0.00 0.28 0.08 0.20 N N N N N N 1.00 0.20 0.40 0.50 Moder WeakLP13 0.32 0.00 0.32 0.05 0.27 N N N N N N 0.80 0.10 0.50 0.20 Weak StrongLP15 0.32 0.00 0.32 0.05 0.27 N N N N N N 0.80 0.09 0.30 0.30 Weak StrongLP14 0.35 0.00 0.35 0.09 0.27 N N N N N N 0.90 0.35 0.50 0.70 Weak StrongLP16 0.19 0.00 0.19 0.06 0.13 N N N N N N 0.70 0.25 0.50 0.50 Weak WeakLP17 0.36 0.12 0.24 0.04 0.20 N N Y Y N N 0.60 0.12 0.40 0.30 Weak ModerLP18 0.42 0.12 0.30 0.10 0.20 N N Y Y N N 0.60 0.45 0.50 0.90 Weak ModerLP19 0.40 0.12 0.28 0.08 0.20 N N Y Y N N 0.60 0.27 0.30 0.90 Weak ModerLP20 0.41 0.12 0.29 0.09 0.20 N N Y Y N N 0.60 0.48 0.80 0.60 Weak ModerLP21 0.51 0.12 0.39 0.19 0.20 N N Y Y N N 1.40 1.54 1.40 1.10 Weak ModerLP22 0.50 0.12 0.38 0.11 0.27 N N Y Y N N 0.80 0.63 0.90 0.70 Moder ModerLP23 0.44 0.12 0.32 0.05 0.27 N N Y Y N N 0.70 0.32 0.80 0.40 Weak StrongLP24 0.36 0.12 0.24 0.04 0.20 N N Y Y N N 0.60 0.12 0.40 0.30 Weak ModerLP25 0.47 0.12 0.35 0.09 0.27 N N Y Y N N 0.80 0.39 1.30 0.30 Weak StrongLP26 0.34 0.12 0.22 0.09 0.13 N N Y Y N N 1.00 0.30 1.00 0.30 Weak WeakLP27 0.31 0.12 0.19 0.06 0.13 N N Y Y N N 0.80 0.28 0.70 0.40 Weak WeakLP28 0.38 0.12 0.26 0.06 0.20 N N Y Y N N 1.20 0.12 0.40 0.30 Weak ModerLP29 0.28 0.12 0.16 0.03 0.13 N N Y Y N N 0.30 0.16 0.40 0.40 Weak WeakLP30 0.26 0.00 0.26 0.06 0.20 N N N N N N 0.90 0.20 0.50 0.40 Moder WeakLP31 0.31 0.00 0.31 0.11 0.20 N N N N N N 1.10 0.50 1.00 0.50 Weak ModerLP32 0.24 0.00 0.24 0.11 0.13 N N N N N N 0.50 0.84 1.40 0.60 Weak Weak 62  Appendix C: Sample GPR Trace Signals  Plan View (Top Down)    Key: In this sample plan view GPR trace diagram (from one of the 288 locations surveyed but unidentified for privacy purposes), several different signal intensities may be observed ranging from strong to weak. The analysis scale is guided by a colour range which in this configuration varies from light yellow against the blue soil matrix background (weak to non-existent) to bright red (“Strong”).  The scales are shown in meters with the East-West and North-South dimensions for each axis. 63  Appendix C: Sample GPR Trace Signals  Profile View (Side)    Key: In this sample profile GPR trace diagram (from one of the 288 locations surveyed but unidentified for privacy purposes), several different signal intensities may be observed ranging from strong to weak. The substantial hyperbolic traces toward the left side of the diagram at a depth of 0.8 to 1.5 meters are typical of “Strong” indications of a sub-surface anomaly. Toward the middle and right side of the profile, there are a number of “Moderate” and “Weak” GPR signals evident.  The scale is shown in meters with the East-West distance across the top and the DBS (depth below surface) on the side (X) axis.64  Appendix D: GPR Project Management – Sample Grid Information   65   

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:
https://iiif.library.ubc.ca/presentation/dsp.24.1-0166116/manifest

Comment

Related Items