UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Remote sensing applications in an alfalfa capability assessment of saline soils Ross, Timothy J. 1988

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

Item Metadata

Download

Media
831-UBC_1988_A6_7 R66.pdf [ 16.9MB ]
Metadata
JSON: 831-1.0097838.json
JSON-LD: 831-1.0097838-ld.json
RDF/XML (Pretty): 831-1.0097838-rdf.xml
RDF/JSON: 831-1.0097838-rdf.json
Turtle: 831-1.0097838-turtle.txt
N-Triples: 831-1.0097838-rdf-ntriples.txt
Original Record: 831-1.0097838-source.json
Full Text
831-1.0097838-fulltext.txt
Citation
831-1.0097838.ris

Full Text

REMOTE SENSING APPLICATIONS IN AN ALFALFA CAPABILITY ASSESSMENT OF SALINE SOILS by TIMOTHY J. ROSS B.Sc.(Agr.), University of Guelph, 1979 A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS OF THE DEGREE OF MASTER OF SCIENCE in THE FACULTY OF GRADUATE STUDIES (DEPARTMENT OF PLANT SCIENCE) We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA 1988 (c) Timothy J. Ross In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Plant Science Department of The University of British Columbia Vancouver, Canada Date June 27, 1988 DE-6 (2/88) i i ABSTRACT In the Cariboo-Chilcotin region of British Columbia, two types of soil salinity have been recognized; natural salinity (Solonetzic soils) and secondary salinity. Soil salinity of either type can be a significant limiting factor in alfalfa (Medicago sp.) production. Stress-related changes in individual plant colour and plant morphology and invasion of the stand by salt-tolerant forbs and grasses affect the amount and direction of the radiation reflected and/or emitted by growing plants. Remote sensing techniques are capable of measuring that radiation and, therefore, offer the potential for quantitative assessment of plant stress caused by saline soils. A remote sensing study was initiated to develop a computer-assisted classification of categories of salt-affected soils. Site biophysical data and reflectance data from digitized infrared diapositives collected from a 50m x 50m grid of "training sites", were integrated to produce a series of digital numbers representative of different classes of salt-affected soils. The possible confounding effects of exposure, processing, geometric and atmospheric effects, and image window position on the interpretation of reflectance patterns were first evaluated. Alfalfa exposure fell on the linear portion of the Density-log Exposure curve indicating a satisfactory exposure. The IR balance of the film (22) was more appropriate to higher altitude use (12000 ft asl) than the altitude of the photographic flights (5500 ft asl) resulting in a slightly narrower range of near infrared reflectance values. Variation between four distinct images was not significantly different (P<0.05) so the complete data set was used for analysis. Sites positioned near photograph borders were avoided as decreased reflectance was noted at these locations. Parameters used in classification were selected on the strength of their correlation with other parameters in the system. The spectral-plant-soil model was developed using a correlation threshold (r> + 0.35) to select paranici-ers followed by a morphological systems i i i analysis to devise the component systems. The key system parameters were; Spectral: Yellow(Green), Magenta(Red), Cyan(NIR), Ratio R 3 ( M / C ) and Ratio R4(M/Y); Plant: ALF(alfalfa), SALT(salt-tolerant grasses), GRASS(domestic grasses) and PUNU (Puccinellia nuttalli); Soil: PHW(pH water), EC (electrical conductivity), ECa (exchangeable Ca), ENa (exchangeable Na + ), EK (exchangeable K + ) BD (Bulk density) and E (elevation). Spectral parameters C and R4 were most strongly correlated with A L F (r= +0.40 and r = -0.35, respectively) and were selected as grouping parameters for cluster analysis. The training sites were clustered into three groups. The spectral data for these groups was used to devise digital numbers in the Y,M and C dye-layers for use in the Meridian-PC image analysis system. The plant and soil parameter trends (x+ 1 sd and % CV) were used to determine continuous classifications representing LOW ALF (low alfalfa capability on saline and/or sodic soils), MED ALF (medium alfalfa capability, including areas of developing salinity with intermediate values for soil parameters) and HIGH ALF (high alfalfa capability which may include areas with high populations of domestic grass species in lieu of alfalfa which had declined for reasons unrelated to salinity). Parameter means were significantly different (P<0.05) for LOW ALF vs MED ALF, LOW ALF vs HIGH ALF and in some, but not all cases for MED ALF vs HIGH ALF. The BIOM (biomass) data set reflected higher production in the HIGH A L F class. Total nitrogen (TN) and organic matter (ORGC) showed negative correlations with saline soil indicators and positive trends with alfalfa. Total N levels were low to medium in comparison to common ratings for effective alfalfa production. More available phosphorus (PHTOS) was present in the LOW ALF than in the HIGH ALF, although all areas would require supplemental phosphorus to improve productivity. The spectral signatures supplied to the Meridian-PC image analysis system were used to produce capability maps of the study site depicting three alfalfa production classes and an unclassified category comprising values outside the spectral ranges established for the three classes. The training sites were grouped into LOW ALF (18%), MED ALF (43%) and HIGH ALF i v (37%). Computer-assisted supervised classification of the data improved the discrimination of the groups to produce a distribution of LOW ALF (31%), MED ALF (42%) and HIGH ALF (28%). This study successfully discriminated digital numbers which were indicative of classes of vegetation which are, in turn, reflective of gradations in saline soil conditions. Computer-based analysis of digitized CIR aerial photographs may, therefore, be a valuable tool in the identification and evaluation of the impact of saline soils on crop productivity. TABLE OF CONTENTS Page 1.0 INTRODUCTION 1 1.1 Soils 1 1.2 Plant Stress 3 2.0 DESCRIPTION OF STUDY AREA 5 2.1 Location: Site History 5 2.2 Geology 5 2.3 Soils 6 2.4 Climate 7 2.5 Vegetation 7 3.0 LITERATURE REVIEW 9 3.1 Soil Variability 9 3.2 Soil Salinity 10 3.3 Crop Responses to Salinity 15 3.4 Radiation Responses and Remote Sensing 18 3.41 Reflectance Characteristics of Vegetation 19 3.42 Principles of Colour Infrared Aerial Photography 21 3.43 Digital Analysis 23 3.5 Remote Sensing and Vegetation Stress 23 3.51 Spectral Changes in Aerial Photographs 2.5 3.52 Remote Sensing Applications 26 4.0 MATERIALS AND METHODS 30 4.1 Sampling Design . 30 4.2 Plane Surveying 30 4.3 Soil Sampling 35 v i Table of contents (cont'd) Page 4.4 Soil Physico-Chemical Analyses 36 4.41 Sample Preparation 36 4.42 Chemical Analysis 36 4.43 Physical Analysis 37 4.5 Plant Sampling 37 4.51 Canopy Cover 37 4.6 Aerial Photography 38 4.61 Photographic Mission 38 4.62 Densitometry 38 4.63 Digital Analysis 39 4.7 Statistical Methods 40 4.71 Correlation Analysis 40 4.72 Morphological Analysis 40 4.73 Cluster Analysis 41 4.74 Variability Assessment 41 4.8 Supervised Classification 41 5.0 RESULTS AND DISCUSSION 42 5.1 Introduction 42 5.2 Film Selection and Photographic Control 43 5.3 Comparison of Spectral Data by Photo 50 5.31 Plant Parameters 5 6 5.32 Soil Parameters 5 6 5.33 Spectral Parameters 5 6 vi i Table of Contents (cont'd) Page 5.4 Parameter Selection 58 5.41 Spectral Parameters 60 5.42 Soil Parameters - Chemical 60 5.43 Soil Parameters - Physical 61 5.44 Plant Parameters 61 5.5 Morphological Classification 70 5.51 Spectral System 70 5.52 Plant System 81 5.53 Soil System 85 5.54 Spectral-Plant System 87 5.55 Plant-Soil System 89 5.56 Spectral-Soil System 89 5.57 Total Spectral-Plant-Soil System 90 5.6 Cluster Analysis 94 5.61 Cluster Analysis of ALF - C and R 4 - 4 Classes 95 5.62 C and R 4 vs ALF - 3 Classes 100 5.621 Spectral Parameters 100 5.622 Plant Parameters 106 5.623 Soil Parameters 107 5.63 Cluster Analysis of ALF - C and R 4 - 3 Classes 109 5.7 Spectral Signatures 112 5.8 Continuous Soil and Vegetation Classes 114 5.81 Plant Parameters 115 5.82 Soil Parameters 115 Table of Contents (cont'd) Page 5.9 Cluster Analysis of SALT - C and R 3 - 3 Classes 124 5.91 Spectral Parameters 128 5.92 Plant Parameters 128 5.93 Soil Parameters 131 5.94 Mis-classified Units 131 5.95 Comparison of C and R 4 and C and R3 Classifications 132 5.96 Additional Soil and Plant Parameters 132 6.0 RESULTS OF COMPUTER-ASSISTED SUPERVISED CLASSIFICATION 136 6.1 Supervised Classification 136 6.2 Filtered Image 145 6.3 Supervised Classification - Other Applications 145 7.0 CONCLUSIONS 151 8.0 RECOMMENDATIONS 155 LITERATURE CITED 157 Appendix 1 164 Appendix 2 172 i x LIST OF TABLES Page Table 1 - Positional effects on dye-layer density measurements from CIR film 58 Table 2 -Comparison of dye-layer density and total film response for positive transparencies and paper prints from colour infrared aerial photographs of an alfalfa stand 49 Table 3 -Plant, soil and spectral parameters used in the analysis of aerial colour infrared photographs 55 Table 4 - Photo comparison summary 57 Table 5 - Correlation matrix of 25 selected spectral, plant and soil parameters 59 Table 6 - Particle size analysis of soils sampled on the five fields of the study site 62 Table 7 - Survey of vegetation on the five fields of the study site included in the remote sensing analysis 64 Table 8 - Cluster analysis for ALF-four classes 98 Table 9 - Comparison of four classes of ALF grouped using C and R 4 for spectral, plant and soil parameters 99 Table 10 - Cluster analysis for ALF-three classes 101 Table 11 - Comparison of three classes grouped using C and R 4 for spectral. plant and soil parameters 102 Table 12 - Computer reclassification of ALF 110 Table 13 - Digital numbers in the three film dye-layers(y,m,c) for the three ALF classes 113 Table 14 - Cluster analysis for SALT 125 Table 15 - Comparison of three classes of SALT using C and Rg for spectral, plant and soil parameters 129 Table 16 - Classification of A L F for associated soil parameters 134 Table 17 - Supervised computer classification of training sites 140 LIST OF FIGURES Figure 1 Reflectance spectra of green vegetation 19 Figure 2 Map - Alkali Lake IR 4,4a - Location of fields and field subunits 32 Figure 3 Map - Alkali Lake IR 4,4a - Location of flight lines and training sites 34 Figure 4 Density - log exposure curve 44 Figure 5 Altitude Aim Curve - Determination of IR balance 46 Figure 6 PH 8882 - Colour infrared image - Field 1 52 Figure 7 PH 8909 - Colour infrared image - Field 2 53 Figure 8 PH 8883 - Colour infrared image - Field 3; eastern half of Field 4 54 Figure 9 PH 8911 - Colour infrared image - western half of Field 4; Field 5 55 Figure 10 Morphological systems diagram for spectral parameters 71 Figure 11 Plot of digital numbers for Yellow (green) vs. Magenta (red) 72 Figure 12 Cumulative frequency plots - Cyan dye - layer (NIR) 74 Figure 13 Normal probability plot of Cyan dye - layer (NIR) 75 Figure 14 Cumulative frequency plots - Yellow dye - layer (green) 77 Figure 15 Cumulative frequency plots - Magenta dye - layer (red) 78 Figure 16 Cumulative frequency plots - R 4 (Magenta:Yellow) (red:green) 79 Figure 17 Cumulative frequency plots - R3 (Cyan:Yellow) (NIR:green) 80 Figure 18 Morphological systems diagram for plant parameters 82 Figure 19 Morphological systems diagram for soil parameters 86 Figure 20 Morphological systems diagram for spectral-plant-soil parameters 88 Figure 21 Morphological systems diagram for ALF 92 Figure 22 Morphological systems diagram for SALT 93 Figure 23 ALF classification in 2-D pixel value vector space - 4 classes. Cluster analysis grouping variables - C and R 4 (m/y) 97 Figure 24 ALF classification in 2-D vector value space - 3 classes. Cluster analysis grouping variables - C and R 4 (M/Y) 104 xi Page Figure 25 Continuous Classification - A L F Population trends of LOW, M E D and H I G H A L F classes 116 Figure 26 Continuous Classification - S A L T . Population trends of L O W , M E D and H I G H A L F classes 117 Figure 27 Continuous Classification - GRASS. Population trends of LOW, M E D and H I G H A L F classes 118 Figure 28 Continuous Classification - PHW. Distribution of values LOW, M E D and H I G H A L F classes 119 Figure 29 Continuous Classification - EC. Distribution of values of LOW, M E D and H I G H A L F classes 121 Figure 30 Continuous Classification - ECa. Distribution of values of LOW, M E D and H I G H A L F classes 122 Figure 31 Continuous Classification - ENa . Distribution of values of L O W , M E D and H I G H A L F classes 123 Figure 32 Continuous Classification - E K . Distribution of values of LOW, M E D and H I G H A L F classes 126 Figure 33 S A L T Classification of 2-D pixel value vector space - 3 classes. Cluster analysis grouping variables C and Rg (c/y) 127 Figure 34 Continuous Classification - BIOM. Distribution of values of LOW, M E D and H I G H A L F classes 135 Figure 35 Supervised Classification - P H 8882 - Field 1 138 Figure 36 Supervised Classification - P H 8909 - Field 2 141 Figure 37 Supervised Classification - P H 8883 - Field 3, eastern half of Field 4 142 Figure 38 Supervised Classification - P H 8911 - western half of Field 4; Field 5 144 Figure 39 Supervised Classification - 5 x 5 Pixel Filter - P H 8882 - Field 1 146 Figure 40 Supervised Classification - 5 x 5 Pixel Filter - P H 8909 - Field 2 147 Figure 41 Supervised Classification - 5 x 5 Pixel Filter - P H 8883 - Field 3, eastern half of Field 4 148 Figure 42 Supervised Classification - 5 x 5 Pixel Filter - P H 8911 - western half of Field 4; Field 5 149 ACKNOWLEDGEMENTS I would like to thank Dr. F.B. Holl, Department of Plant Science for his supervision and council throughout this project. I acknowledge with thanks the support and encouragement of the Alkali Lake Indian Band; in particular, Chief Andy Chelsea, Summer Agricultural Assistants and Economic Development Director, Mr. Steve Gibson. The financial contribution of the Western Indian Agricultural Corporation Ltd. to the project is also gratefully acknowledged. I appreciate the interest and contributions of my supervisory committee, Drs. V.C. Runeckles, M.D. Pitt, H. Schreier and particularly Dr. P.A. Murtha for his extensive advice in the remote sensing analysis. In addition, the technical support of remote sensing staff Raoul Wiart and Nadenia Krajci was invaluable. I would also like to thank the members of my family and Johnny, Mick and Keith for their continuing support. 1 1.0 INTRODUCTION 1.1 Soils Dryland soil salinity is considered to be the major soil degradation problem in Western Canada because of the extent of affected soils and the economic impact on agricultural production (Senate Committee on Agriculture, Fisheries and Forestry 1984). In the Canadian Prairies and U.S. Great Plains much of the land was saline prior to cultivation (natural salinity), whereas other areas have become saline because of man's activity (secondary salinity) (Sommerfeldt et al., 1984). Secondary salinity is primarily of two types: (i) salinization in irrigated areas due to saline irrigation water, canal seepage, inadequate drainage and poor irrigation management (Alberta Agriculture 1980), and (ii) saline seeps - areas of recently developed salinity in non-irrigated soils that are wet some or all of the time. The latter are often observed with white salt crusts in areas where crop or grass production is reduced or eliminated (Miller _et_al. 1981). The Cariboo-Chilcotin region of the Fraser plateau is the site of many saline and/or alkali soils. In British Columbia, all saline soils occurring in the Interior are geologic static types (Prairie Farm Rehabilitation Administration 1983). These soils, recognized as having a unique morphology and well-defined characteristics, are categorized within the Solonetzic Order. It is generally assumed that Solonetzic soils originated from parent material that was more or less uniformly salinized (Canadian System of Soil Classification 1978). The Solonetzic Order includes a relatively broad group of hardpan soils often associated with alkali. Aside from varying depths of topsoil these soils also vary in the relative degree of formation of the hardpan. Solonetzic soils are formed on parent material that is naturally rich in N a + or from materials that have been enriched with N a + salts through upward movement of ground water. Soils with an extensive hardpan layer ("tough" soils) are included in one of two Great Groups, Solonetz or Solodized solonetz, 2 whereas those soils with a weaker hard pan are classified in the Solod Great Group (Canadian System of Soil Classification 1978). These Great Groups have three horizons which contain large amounts of N a + and Mg salts and a strong columnar structure that breaks to sub-angular blocks. There are large accumulations of CaCOg at depth, and below this may be additional accumulations of CaS0 4 (Valentine et al. 1978). In British Columbia, Solonetzic soils are found in the sub-humid aspen parkland of the Peace River lowland and associated with Chernozems in the Dry Interior Plateau (Valentine et_al. 1978). Large, continuous areas similar to the Canadian Prairies and U.S. Great Plains do not exist. In British Columbia the majority of arable agriculture occurs in alluvial valleys. The nature of valley formation causes them to exhibit varied topography and stratigraphy, conducive to saline seep formation. Vander Pluym (1985) listed six stratigraphic factors which he proposed to evaluate saline seeps: i) the extent of the high water table immediately upslope of the seep. ii) the existence of coarse-textured, saturated conduits in draws leading to the seep, iii) upslope areas with thin and/or sandy overburden. iv) the direction and gradient of shallow ground water flows. v) shallow, sizable, pressurized aquifers in the bedrock. vi) firm, fine textured overburden directly downslope of the saline seep, creating a dam. Dominant recharge to seepage area ratios are estimated to be 2:1 to 5:1. While the immediate rooting zone of a plant stand may be the primary concern for agricultural production, drainage may be related to a much larger land area. Water flow from upper to lower levels, as well as laterally, is a function of topography, stratigraphy and hydraulic conductivity. In arid regions where the soils contain large amounts of soluble salts, water can transport dissolved salts downslope to low areas. Ponding or 3 shallow water tables result and evaporation leads to increased salt concentration in surface soils (Miller etal. 1981). The combination of circumstances leading to development of natural and secondary salinity is usually a localized problem. Therefore, a combination of terrain analysis and examination of plant responses is likely to be necessary, in addition to a conventional soil survey, in order to identify these salt-affected soils. 1.2 Plant Stress To understand the ecology of a plant community the autecology of the component species must be understood. Daubenmire (1965) defined autecology as "the study of interrelations between the individual and its environment". The alfalfa fields used in this study are not monocultures. They are complex vegetation communities encompassing remnant native grass and forage species, introduced alfalfa and domestic grasses and salt-tolerant invaders. Thus, the principle of synecology, "the study of the structure, development and causes of distribution of plant communities" (Daubenmire 1965), must also be introduced. Environmental factors may be grouped as climatic, biotic and edaphic. Edaphic factors are the primary focus of this study. Plants and soils strongly influence each other. Since edaphic factors can be both complex and dynamic, the adaptation of specific plants to particular sets of soil conditions is to be expected. The importance of a plant species relationship to a particular factor lies in the ability of that plant to tolerate extremes of the factor or to be excluded from a particular environment. In an agricultural situation the maximum yield of plants, as determined by their genetic potential, is seldom achieved within the adverse biotic and abiotic constraints which limit growth. Plants subjected to these limiting factors are said to be stressed. Historically the most widespread approach to stress detection has been by visual survey. Experienced producers and agrologists have developed the ability to discern subtle changes in plant colour or in plant morphology, often early indicators of stress. The 4 limitations of such a methodology are, that few people have the necessary experience or insight to detect these signs effectively and, that fields are generally too large to be accurately surveyed by eye. By the time visual and tactile symptoms are unambiguously evident, yield-limiting damage may already have occurred (Jackson 1986). Bresler et _al. (1982) reported that visual salt toxicity symptoms did not appear until significant yield depression had already occurred. Whether or not they are detectable by visual survey, stress-related changes affect the amount and direction of radiation reflected (and emitted) by plants. Remote-sensing techniques are capable of measuring that radiation and therefore offer the potential for quantitative assessment of plant stress caused by biotic and abiotic factors. The specific objectives of the research reported in this thesis include: 1. Quantification of plant population responses to edaphic conditions through sampling and analysis of plant populations, study site terrain and soil characteristics. 2. Evaluation of inferred spectral reflectance data obtained by low-level colour infrared photographs, using digital image analysis techniques. 3. Integration of site biophysical and remote sensing data collected from a series of "training sites", to produce a series of "signatures" representative of different categories of salt-affected soils. 4. Development of a computer-assisted classification of soils and vegetation on the study site. 5. Development of a predictive method for evaluation of reclamation or revegetation strategies using remote sensing techniques. 5 2.0 DESCRIPTION OF STUDY AREA 2.1 Location: Site History The study area is located in the Cariboo-Chilcotin region of the Fraser Plateau on reserves 4 and 4a of the Alkali Lake Indian Band. The reserves are located approximately 50 km south of Williams Lake, B.C. and east of the junction of the Fraser and Chilcotin rivers. Both reserves are located in the watershed of Alkali Creek which drains to the southwest. The site supported native range vegetation until 1981 when the land base was converted to alfalfa production; installation of an irrigation system was preceded by sod-breaking, rockpicking and alfalfa seeding. Following the first hay cutting of 1985 most study site soils received a surface application of 11-55-0 fertilizer. The exact area covered by the application is unknown. Rates were variable. Alfalfa winterkill was widespread throughout the B.C. Interior following the winter of 1985-86. In the Williams Lake area, the severity ranged from 20-90% mortality (B.C. Ministry of Agriculture and Fisheries 1986). On the study site, restricted areas were particularly affected. Dead alfalfa crowns and weakened, less vigourous plants were noted the following spring. 2.2 Geology The Fraser plateau-East is an undulating plateau of tertiary basalt flows ranging in elevation from 900-1750 metres (Holland 1976). Pleistocene and recent surficial materials include glacial moraine and glacio-fluvial outwash. The volcanic surface was over-ridden by the ice of the Fraser glaciation which flowed southwesterly to deposit a mantle of fine-textured glacial till. Ablation of the ice mass created a pitted outwash plain to the east. Alkali Creek, which drains southwesterly to the Fraser River, is a former meltwater channel; thus, associated coarse-textured glacio-fluvial deposits often overlie finer materials deposited from volcanic and morainal action (Tipper 1971). A further 6 complication arises from the presence of an eolian mantle of fine sand and silt. These windborne materials form thin layers over older landforms. The thickest and most extensive deposits are found on terraces, fans, floodplains and outwash surfaces (Valentine et al. 1978). On areas of the Fraser plateau east of the Fraser River, olivene basalts, andesites and related tuffs (fragmented volcanic rock the size of course sand or gravel) and breccias (angular fragments of rock within a matrix of the same, or another type) form the parent material (Tipper 1971). 2.3 Soils The 1:50,000 Soil Association map, on which the study area is classified as an Orthic Dark Brown Chernozem (Valentine 1984), lacks sufficient detail for specific management interpretation. However, some important inferences can be drawn from the available soil information. The Soil Associations in the study area, include the Chimney and Williams Lake series. Both series were derived from parent material described as gravelly and/or loamy skeletal morainal material. Such materials often are very fine-textured and compacted, especially when deposited as basal till (as the "blanket" deposits seem to indicate) rather than ablation till (resulting from in situ decay of the ice mass). Overlying glacio-fluvial deposits are much coarser in texture as a result of the sorting action of the glacial outwash. The Chimney Soil Association is described as undulating plateau grasslands with 10-15% saline soils in depressions. These associated saline soils suggest the B m or Btj horizon of the Orthic Brown Chernozem has been replaced by a B n j t horizon, formed as a result of N a + and clay accumulation. A description of basic soil classification terminology is included in Appendix 1. These "sodic" soils have very low water infiltration rates, a pH commonly greater than 9.0 and dispersed clay and organic fractions. Dispersed organic matter may accumulate at the surface of poorly drained areas and impart a black colour. 7 The Williams Lake Soil Association is an Orthic Gray Luvisol noted for its 10-15% imperfectly drained soils. Luvisolic soils commonly develop in forest-grassland transition zones from parent material which is neutral to alkaline in soil reaction (Canadian System of Soil Classification 1978). The B horizon of such soils is a site of clay accumulation providing the potential for imperfectly drained soils. Orthic Gray Luvisols often develop in depressional areas which may also have a high N a + concentration. The B horizon shows an accumulation of N a + as well as clays. These poorly drained microsites may be more correctly described as Solodic Orthic Gray Luvisols. 2.4 Climate Climate is a major determinant of the range of crops which may be grown in a region. The climate capability for agriculture described for the study site indicates a frost-free period of 75-89 days (B.C. Ministry of the Environment 1981). The range of growing degree days (GDD) above 5°C is 1170-1309. There is a climatic moisture deficit of 116-190 mm during the growing season (May - August). The frost-free period is suitable for the cultivation of cool-season vegetables, some berry crops, grains and forages. Irrigation is necessary for successful crop production as a consequence of the large climatic moisture deficit during the growing season. 2.5 Vegetation The study site is located at the northern extremity of the Upper Grassland Zone (McLean 1979). The Upper Grasslands are distinguished from the Lower and Middle Grassland Zones by a greater abundance of forbs, increased height of grasses and a denser plant cover. Increased vegetative productivity is a consequence of greater moisture effectiveness. Clumps of aspen (Populus tremuloides Michx.) occur frequently in draws and swales. Bluebunch wheatgrass [Agropyron spicatum (Pursh) Scribn. .& Smith] is the main decreaser on undisturbed sites. Kentucky bluegrass (Poa pratensis L.) generally dominates on deep, fine-textured soils and swales, along with Richardsons' needlegrass (Stipa richardsonii Link) but frequently increases at the expense of bluebunch wheatgrass. 8 Further deterioration is marked by the appearance of Sandberg's bluegrass (Poa sandbergii Vasey), compound fleabane (Erigeron compositus Pursh), pasture sage {Artemisia frigida Willd.), pussytoes (Antennaria spp.), western yarrow (Achillea millefolium L.), dandelion {Taraxacum officinale Weber), mullein (Verbascum thapsus L.) and goatsbeard (Tragopogon dubius Scop.) Saline areas are marked by various sedges, (Carex spp.), forbs and salt-tolerant grasses such as inland salt-grass (Distichlis stricta (Torr.) Rydb.), foxtail barley (Hordeum jubatum L.), alkali-blue grass (Poa juncifolia Scribn.) and NuttalPs alkaligrass (Puccinellia nuttallia (Schult.) Hitchc.) (Hitchcock and Cronquist 1973). 9 3.0 LITERATURE REVIEW 3.1 Soil Variability The combination of volcanic parent material modified by morainal and glacio-fluvial action, and overlain by a thin eolian mantle leads to complex soil associations. Kowalenko (1985) categorized factors affecting soil variability as: i) Natural a) deposition/genetic b) physical features such as slope and elevation c) spatial (3-Dimensional) d) nutrient (mobility/reactivity, response to redox conditions) ii) Management a) landscape (levelling, drainage) b) field (division and use arrangement of a parcel of land) c) soil (cultivation) d) amendment (manure, lime, fertilizer) e) cropping iii) Sampling procedures a) area specification (field size, number/field) b) distribution (spot, random, grid etc.) c) depth d) timing (spring vs. fall) e) method (shovel, core, pit) Spatial variability within landscapes is a continuum. The purpose of soil surveys is to partition this continuum into individual classes that have greater homogeneity for selected soil properties than the continuum as a whole (Wilding 1984). The scientific basis of a soil survey presumes that soils and their location on the landscape are predictable by an experienced soil scientist with knowledge of the geology, 10 vegetation, climate and landform patterns of the area. Only enough observations are made during mapping to establish predictive relationships and to confirm predictions made from these relationships. In addition to systematic variability, random variability among soil properties often causes them to exceed limits imposed by the definition of the class (Wilding 1984). An inherent danger in the application of soil surveys lies in their use for purposes requiring greater detail than the mapping scale warrants. Similarly, interpretation of soil data in relation to management practices must recognize that interpolation of mean values to a response curve can lead to inaccurate estimates of the degree of response (Moon & Schreier 1985). 3.2 Soil Salinity Soil salinity is usually confined to arid region soils. These soils occur where annual rainfall is less than 50 cm-yr" .^ Lack of intensive leaching leaves the base status of these soils high; a fully developed soil profile usually carries a CaCOg accumulation greater than that of the parent material at some point (usually in the C horizon). The lower the rainfall, the nearer to the surface this carbonate layer will be. When drainage of these soils is impeded and surface evaporation becomes excessive, soluble salts accumulate in the surface horizon (Brady 1984). Two types of salinity have been recognized. Soils which were saline prior to cultivation (natural salinity) have unique morphology and well-defined characteristics. These are categorized within the Solonetzic order. (See Appendix 1) Secondary salinity (due to hydrogeological, climatic or cultural factors) includes salinization in irrigated areas and saline seeps (Miller et_al. 1981). Conditions leading to seep formation require parent material to act as a source of salts. Basalts are rich in water-soluble alkaline earth cations (Ca , Mg , Na , K ); they are readily weathered and therefore provide a large source of soluble salts (Bohn et al. 1979). Few 11 areas become saline from weathering alone. Saline soils receive salts from other areas with water as the carrier. The load is dependent upon geological and soil conditions (United States Salinity Laboratory Staff 1954). The seepage mechanism involves the percolation of water, in excess of the soil water-holding capacity, beneath the rootzone in coarse-textured upper elevation soils. Dissolved soluble salts accumulate in shallow, less-permeable subsoil layers where a local groundwater flow system develops. Saline water moves laterally from the recharge area to the point of truncation between the subsoil layer and the adjoining landscape. Ponding or shallow water tables result and evaporation leads to an accumulation of salt in these seep or discharge areas (Miller etal. 1981). Upon weathering the primary minerals of the parent material can be easily altered to the secondary serpentine group and further to secondary clay minerals (Proctor and Woodell 1975). Such soils are often stony and shallow, with low Available Water Storage Capacity (AWSC) and reduced root penetration. Saline properties such as lumpiness, crustiness and vertical cracking when dry and increased viscosity, compactness and stickiness when wet are exhibited (Proctor and Woodell 1975). There are a number of chemical soil parameters which are commonly used to assess the status of the alkaline earth cations and soil moisture. The conductivity of the saturated extract (ECe) is recommended as a general method. Although slower than resistivity measurements of the saturated paste (ECg) the results are more applicable to soil/plant relationships. The lower limit (permanent wilting point) for plant growth is estimated to be 25% of the concentration of the saturated extract while the upper limit (field capacity) is estimated at 50% of the concentration of the saturated extract. The United States Salinity Laboratory Staff (USSLS) recommends chemical determinations of the saturated extract be linked to % field moisture (USSLS 1954). Soil reaction (pH) is another important chemical indicator which reflects the chemical constituents and moisture status of the soil. pH influences the composition of 12 exchangeable cations, composition and concentrations of soluble salts and the presence or absence of gypsum and alkaline-earth carbonates. Concentrations of cations, anions and salts in the soil solution are both pH and moisture dependent as are the processes influencing them such as CEC reactions, negative adsorption of ions, hydrolytic reactions and solubility. Sodium absorption ratio (SAR) is a measurement of which characterizes the relative sodium status of irrigation water and soil solutions: This relationship has been useful as a predictor of the effect of Na ions on soil structure. Soils with high exchangeable sodium levels frequently crust and swell or disperse, greatly decreasing hydraulic conductivity (Bohn et al. 1979). Salt-affected soils have been classified on the basis of Na + activity in the soil solution (Bridges 1970). The two major processes are: 1) alkalization (solonization) - accumulation of N a + ions on exchange sites. 2) de-alkalization (solodization) - leaching of N a + or Na salts from natric horizons. ECe and SAR have been used to describe these soils: (i) Saline [ECe > 4 and SAR < 15] These soils are called "white alkali" or "solonchak" by some workers due to the white salt crust they exhibit. They contain excess salts (from salt-rich geological substratum), but are usually low in Na + . They are well flocculated and thus have adequate drainage for the leaching of excess salts. Ca and 9 _ i _ o Mg are dominant cations, while Cf, SO4 and sometimes NOg" are the principal anions. The pH is usually < 8.5. (ii) Saline-sodic [EC > 4 and SAR > 15] These soils suffer from both excess soluble salts and a higher percentage of N a + . The pH often exceeds 8.5. Thus, structural problems may exist as SAR (meq l'1) ^ r [ C a z t 4 M g 2 + ] 13 N a + deflocculates fine particles. This dispersion leads to weakly stable aggregates and low permeability. Leaching prior to removal of excess Na + creates conditions in which the sodium ion reacts to form IS^COg. This reaction contributes to reduced penetration and movement of H^O in the soil profile, (iii) Sodic [EC < 4 and SAR > 15] Sodic soils are often called "black alkali" or Solonetz (exchangeable Ca:exchangeable Na < 10) (Canadian System of Soil Classification 1978).In semi-arid areas, solonetzic soils are often found as "slicks" or "blowouts". With time, clays are leached and the remaining surface soils, which are coarse and friable, are often susceptible to wind erosion. Sodium-saturated clay accumulates at lower levels in the profile exhibiting columnar or prismatic structure. The pH of sodic soils is usually 8.5 to 10.0. Dispersed and dissolved organic matter may accumulate in a puddled layer on the surface if the coarse soils have eroded, hence the name "black 9 alkali". Common anions are Cl", S O 4 , HCOg" and often small amounts of C 0 3 2 " . In the presence of COg 2" and high pH C a 2 + and M g 2 + will precipitate (USSLS 1954). The Cation Exchange Capacity (CEC) of a soil is a measure of the total amount of exchangeable cations a soil can retain. Primary binding sites for these minerals are clays and organic matter. The CEC reflects significant soil physical and chemical parameters. Exchange complex adsorbed cations can exchange freely with those in solution. I 9 4- 9 4- 4- 4-Na , Ca and Mg are readily exchangeable, whereas K and N H 4 may be fixed 9 4- 9 4-on clay minerals. At equivalent solution concentrations Ca and Mg adsorption strength is several times that of Na + . One-half or more of the soluble cations must be N a + before the exchange complex will absorb significant amounts of this ion (USSLS 1954). 14 Defining the strength of retention of exchangeable cations is important as these relationships, particularly between Na and Ca , influence the procedures required for successful management and reclamation of salt-affected soils. Cation strength of retention is governed by the nature of the soil colloid, concentration of the soil solution, size of CEC, % base saturation (%BS) and the size of the hydrated and charged particles. In general, retention is proportional to valency; trivalent > divalent > monovalent. Within valencies the hydrated radius is a more crucial factor. Sommerfeldt (1985) found divalent cations to o _i_ 9 4-be preferentially adsorbed at pH > 7.5. In a Ca -dominated system Ca ions are held tighter so the system is flocculated. In a Na +-dominated system the retention is looser as a result of the lower charge and larger hydrated radius. Clay particles in the latter system tend to deflocculate with attendant aeration and permeability problems. The physical consequences of salt-affected soils are most often manifested as problems in water relations. Deflocculation of clays and organic matter and the existence of indurate soil horizons restrict the permeability and aeration of the soil. Furthermore, combinations of slow infiltration rate and shallow water tables can cause ground water to seep upwards into the rootzone by capillary action thus aggravating the salinity problem (Hillel 1980). Soil water potential and hydraulic conductivity vary widely and non-linearly with water content for different soil textures. Moreover, these relationships are relatively difficult and expensive to measure or are not feasible for short-term or remote investigations. Soil texture predominantly determines the water-holding characteristics of most agricultural soils. Saxton et_al. (1986) used a large data base to develop predictive equations between selected soil parameters and soil texture. These authors found that their equations provided excellent estimates for model applications or as calibration parameters where field or laboratory data of soil water characteristics are available. High levels of secondary clay minerals and/or soil organic matter can considerably increase water retention ability. High N a + and/or low salt concentrations may cause soils 15 dominated by clays to swell considerably. Soils high in organic colloids are much more stable although they may disperse with high pH levels in the presence of high Na + /low salt (Bresler etal. 1982). Salt-affected soils of high CEC and large surface area possess considerable potential for the retention of nutrient cations (Bresler et al. 1982). Goh et_al. (1986) found that the clay fractions of southwestern Alberta Chernozems could adsorb one and one-half times more phosphate than the silt fraction. In comparison, the clay fraction of a central Alberta Solodized Solonetz was found to absorb two to ten times more phosphate. 3.3 Crop Responses to Salinity Crop salt tolerance may be defined as the ability of the plants to survive and produce economic yields under the adverse conditions caused by soil salinity. Salt tolerance of agricultural crops is typically expressed in terms of the yield decreases associated with increases in soil salinity, or as relative crop yield on saline versus non-saline soils (Maas and Hoffman 1977). Crop survival and appearance, though usually used in association with ornamental species, is particularly important in applications relating remote sensing information to actual plant performance. Individual plants face two specific stresses in saline areas; i) general osmotic effects ii) specific ion effects on plant physiological processes. Most plants respond to soil salinity as a function of the total osmotic potential of the soil solution. The osmotic potential ( YTJ) is a component of the water potential (V) equation [Y= Yft+'MV] where ^ = pressure potential. The addition of solute particles to pure water at one atmospheric pressure ( Y = 0) decreases the osmotic potential of the solution. Decreasing water potential has the net effect of reducing the availability of water to plants, despite the ability of the plant to adjust its internal osmotic potential in relation to levels of moderate osmotic stress. This internal regulation mechanism occurs at a cost to individual plant growth and productivity (Salisbury and Ross 1978). Therefore, plants 16 growing on saline soil, often appear to be suffering from drought even in the presence of adequate soil moisture. Plants affected by salinity are generally stunted and bear smaller leaves which may be thicker than those of comparable normal plants. Some grasses may take on a bluish-green cast due to thickened layers of surface wax. Other species exhibit tip and margin burning (Bresler et_al. 1982). Salinity effects often vary across a salt-affected field. This characteristic variability is specifically apparent in crops grown on Solonetzic soils where normal crop growth may be associated with thin, stunted growth on solonetzic patches (Alberta Agriculture 1980). Such variable growth imparts the characteristic "wavy" pattern to the appearance of the stand. By the time visual salinity symptoms appear, extensive yield depression has generally occurred (Bresler et af 1982). Specific ion effects of salinity have also been observed. Excess quantities of a specific ion may be toxic to various plant physiological processes or may cause nutritional disorders. Ions contributing to soil salinity include Cl", S 0 4 , HCO3", Na , Ca , M g 2 + and occasionally N0 3 " and K + (Bernstein 1975). Shukla (1979) found the deleterious effects of anions in plants to be ranked; CO3 > HCO3" > S 0 4 > Cl". Excessive accumulations of Na"*" in tissues near the end of the transpiration stream lead to tip burn, necrosis and eventually death (Bohn et al. 1979). Na"1" toxicity has generally been shown to be greater in the presence of HCO3" or CO3 ". High bicarbonate concentrations can cause nutritional imbalances such as Fe deficiencies (Bohn et ail. I 9 4-1979). Shoot dry weight is decreased in soils with high concentrations of Na and Mg 4- 2 + (Shukla 1979). Sodium ions compete with K uptake if sufficient Ca is not present on 1 9 4-the exchange complex. Increased levels of Na and Mg in the soil lead to increased 9 _i_ levels in the crown (Salisbury and Ross 1978). Sodium and Mg may also interfere with 9 4- 9 4-Ca uptake when they are highly concentrated (Fowler 1981). Ca is an essential component in membrane selectivity and in stabilizing the mitotic spindle during cell 17 division (Salisbury and Ross 1978). Calcium deficiency leads to stunting and eventually death of the growing point. During cool spring weather, leaves may contain high N a + and Cl" concentrations without exhibiting toxicity symptoms. However, leaf burn often appears suddenly at the subsequent onset of hot, dry weather in early summer (Ehlig 1960). As symptoms are associated with conditions of high transpirational demand and because leaf tips and margins are affected first, there is support for the view that N a + and/or Cl" may interfere with transpiration mechanisms. 9 4 - 9 4 - -1- 4-In addition to interference in Ca , Mg and K uptake, excess Na levels also cause soil dispersion, limiting aeration, permeability and root penetration (Hillel 1980). Crops that may be tolerant of specific N a + toxicity may still fail as a consequence of more general adverse physical conditions. Plant sensitivity to salinity often varies with the phenological stage (Maas and Hoffman 1977). Barley, corn and wheat are more sensitive during emergence and early growth than either at germination or in later growth, including grain filling. The osmotic effects of salts on plant growth do not arise from a decrease in water uptake as the osmotic potential of the rooting mechanism decreases. Through a process called osmotic adjustment in the plant, the original osmotic gradient can be maintained. Thus, affected plants are able to extract water and maintain turgor pressure. However, such mechanisms for osmotic adjustment operate at the expense of plant growth (Bresler et al. 1982). Crop yields are generally not significantly decreased until the E C e exceeds a specific value for each crop known as the threshold E C g (Bresler et al. 1982). Maas and Hoffman (1977) presented relative crop yield information as a linear function of soil solution E C e for a variety of field, forage, vegetable and fruit crops, and ornamental species. Crop tolerances to salinity have been categorized by these authors into five 18 groups: sensitive, moderately sensitive, moderately tolerant, tolerant and unsuitable for crops. Relative yield for these groups was calculated from the relation: Y = 100 - B (EC e - A) Where A = salinity threshold in mmhocm"* B = % yield decrease/salinity increase Projected salinity thresholds and related productivity decreases for these different crop categories were tabulated by Carter (1981). Alfalfa yields for uniformly salinized rootzones, decrease approximately 7.3% per mmhocm"* when a threshold of 2.0 mmhocm" * is exceeded. Alfalfa is thus classified as a moderately tolerant species. Halvorson and Reule (1980) concluded alfalfa growth terminated at E C e = 38 mmhocm"-''. 3.4 Radiation Responses and Remote Sensing Radiation that reaches the earth's surface includes solar radiation within the wavelength region from. 0.25 ra to 3.0 m (this includes both the direct solar beam and the diffuse skylight) and radiation emitted from the atmosphere in the wavelength region from 3.0 m to greater than 20.0 m. The energy balance at the ground surface can be expressed as (Jackson 1986): R n = G + H + XE where R n = net (absorbed) radiation (W m ) G = heat flux, n to surface (W m" ) H = sensible heat flux into the air above the surface E = latent heat flux to air (W m"2) X = wavelength ( m) and the net absorbed radiation (Rn) is given by: R n = R s ' - V + R L * - R L * where R s * = incoming solar R s f = outgoing (reflected) solar R-^  i = incoming long wave Rj^ t = outgoing long wave (emitted thermal) These incoming components are a function of the solar intensity and the atmosphere, and depend only slightly on surface characteristics. Outgoing components, in contrast, strongly depend on surface characteristics. The magnitude and wavelength dependence of the reflected and emitted radiation are determined by both the reflective properties and the temperature of the surface features (Jackson 1986). 3.41 Reflectance Characteristics of Vegetation The diagram below shows the fraction of incident energy reflected from a typical leaf over the wavelength interval 0.4 - 2.5 /im. Little of the visible (0.4 - 0.7p.m) or near .2 .4 .6 .8 1 1.2 1.4 1.6 1.8 : 2 2.2 2.4 2.6 W A V E L E N G T H Cum) Figure 1 - Ref lectance spectra of green vegetation (Jackson 1986). 20 infrared (0.7 - 1.3jjxn) incident energy is reflected directly from the outer surface of the leaf because the cuticular wax layer is nearly transparent to radiation at these wavelengths (Knipling 1970). Leaf reflectance is low in the blue (0.45 - 0.52/jm), peaks in the green (0.52 - 0.55ju.m) and decreases to a minimum in the red (0.63 - 0.70>*m) regions of the spectrum, respectively. Leaves of most species absorb more than 90% of the violet and blue wavelengths that strike them as well as almost as high a percentage of the incident orange and red light. This absorption is essentially accomplished by the chloroplasts. The absorbing chlorophyll pigments transmit green wavelengths (Salisbury and Ross 1978). Radiation-absorbing yellow/orange pigments (carotenoids) are also found in the chloroplasts. All chloroplast pigments absorb radiation at 0.445>jjn but only chlorophyll is active in the red region (0.645jj.m). Thus, healthy, actively photo-synthesizing leaves exhibit low reflectance values in the blue and red portion of the spectrum. An increase in reflectance in these wavebands may signal a stress condition in the plants (Jackson 1986). High reflectance of leaves in the near infrared (NIR) (0.7 - l^ ^xm) band is a function of internal leaf structure (Gates 1968). Radiation is diffused and scattered by multiple reflections and refractions at the interface of hydrated cell walls and intercellular air spaces because of refractive index differences (1.4 for hydrated cells vs. 1.0 for air). Forty to sixty percent of this incident NIR is scattered upward from the leaf surface and is designated reflected radiation. The remainder is scattered downward and has been designated transmission radiation. Little NIR is absorbed by the leaves (Knipling 1970). Gausman (1974) detailed causes of variation in plant reflectance in terms of leaf physiology. The most important refractive index discontinuities he observed were between hydrated cell walls and intercellular air spaces (25 - 50%), with discontinuities among cellular constituents having secondary importance (8%). Explanation of spectral signature variations caused by stress or species differences can be found in-these observations. Salt-stressed leaves are often stunted and compressed compared to non-stressed, more porous 21 leaves; the former may be expected to exhibit less infra-red (IR) reflectance. Internal or external leaf discolorations from necrotic tissue result in IR absorption and, therefore, decreased reflectance. Leaf mesophyll cell arrangements may also vary considerably between species. Alfalfa leaf mesophyll consists of elongated palisade and spongy cells, both tissues containing air spaces, many of which are continuous with pores of the stomata (Grove and Carlson 1972). In contrast, leaf mesophyll of grasses is relatively compact. These structural differences are manifested in the relatively low NIR reflectance of grasses compared to alfalfa with a more porous mesophyll. The higher refractive index of alfalfa results in increased light scattering and subsequently increased IR reflectance. Remote measurements of the amount of reflected and emitted radiation at particular wavelengths can be used to infer properties of the measured surface (Jackson 1986). Such information forms the basis for remote sensing of soils and the evaluation of vegetation stress. 3.42 Principles of Colour Infrared Aerial Photography Photography has been a popular remote sensing technique since its inception over a century ago. Though practical use of aerial photography for military and photogrammetric purposes dates from the early days of World War I, for interpretive use, aerial photography was not extensively utilized by the earth and agricultural sciences until the 1930's (Fischer 1975). Early films were sensitive to nearly the same spectral range as the human eye (0.4 - 0.7/im). During World War II the films that were developed for camouflage detection extended film sensitivity into the near infrared wavelengths (0.7 - 0.9^an). The near infrared region of the spectrum has been most commonly used in vegetation analysis by remote sensing. Following its inceptional use in aerial reconnaissance, Kodak Ektachrome Infrared Aerial film was used for such diverse applications as the detection of disease in forests, 22 orchards and grain crops; the identification and inventory of tree species; the study of soil conditions; geological explorations and agricultural studies (Fritz 1967). The image formation process in black and white film occurs when, (i) a reflected photon from soil or vegetation strikes the film emulsion layer; (ii) silver halide is converted to metallic silver and a "latent image" is produced and (iii) upon reverse development, a dye-layer is created where there is no latent image and dye-layer thickness is inversely related to spectral reflectance (Lillesand and Kiefer 1979). A normal colour film has three principle dye-forming layers, one of which is sensitive to blue (B), one to green (G) and one to red (R) light respectively. Image formation is based on the principle of subtractive dyes. On processing, positive images of yellow (G + R - B), magenta (B + R - G) and cyan (B + G - R) are formed in the respective dye-layers (Fritz 1967). Ektachrome Infrared Film also consists of three layers, sensitive to the green, red and NIR regions of the spectrum. In addition, however, all three layers are sensitive to blue light. To limit exposure of each layer to only one spectral region, a yellow filter is always placed over the camera lens to absorb blue light before it reaches the film. Filtration of all wavelengths less than 0.525jum allows clear development of visual tones, densitometry of individual dye-layers and improved haze penetration due to Rayleigh scatter (refraction of light by gaseous water molecules) (Lillesand and Kiefer 1979). During processing the Yellow, Magenta and Cyan dyes are formed on the image, but the Yellow has resulted from green exposure, Magenta from red exposure and Cyan from infrared exposure (Fritz 1967). Thus, a normal-green leaf would appear as magenta with a higher proportion of red as a result of the higher IR reflectance. With these colour relationships, leaves under stress may appear as darker magenta with less red as a result of reduced IR reflection. Senescent, but still green vegetation appears as a lighter magenta with less blue and more red because of the higher IR reflection from senescent vegetation. 23 3.43 Digital Analysis The NIR photographic image provides an analog model of the scene in which brightness is modeled in shades of visual grey (Kelley 1983). With the development of digital computer analytical techniques, it is now possible to accomplish image analysis with the capacity to store, retrieve, analyze and interpret large quantities of data from multispectral images. Any image may be thought of as consisting of tiny equal-sized areas called picture elements or "pixels" arranged in regular lines and columns. The brightness of each pixel may be assigned a numerical value on a scale from zero (black) to a higher number, 255 (white). An image may be recorded (directly) in digital format as in the case of the Landsat Multispectral Scanner (MSS) with numerical terms on a three-coordinate system where x and y locate each pixel and z specifies the pixel brightness. An analog image recorded initially on photographic film may be converted to numerical format by a process known as digitization. The Optronics Colormation C-4500 (a High-speed scanning microdensitometer) is capable of digitizing images as small as 12 um x 12 um (Kelley 1983). A normal colour or colour IR aerial photograph can be converted into at least three pixel matrices according to the brightness values of the three dye layers - yellow, magenta and cyan. 3.5 Remote Sensing and Vegetation Stress The application of remote sensing data to interpretations of vegetation stress requires an understanding of the basis of stress, the effect of stress on vegetation, strain symptoms and the capabilities and limitations of the remote sensing technique (Murtha 1981). Levitt (1972) and Murtha (1982) have provided the following definition for evaluating relative degrees of plant damage: a) stress - any environmental factor capable of inducing a potentially injurious strain on a plant, b) strain - any physical or chemical change in a plant produced by stress, 24 c) injury - any stress on a plant that causes a detrimental strain and may be noted because of either temporary or permanent syndromes, d) damage - any biological or economic loss due to stress, e) damage type - any syndrome expressed by the plant of either temporary or permanent strain caused initially by stress. From the perspective of remote sensing applications, four associated subject areas must also be considered (Murtha 1982): (i) the environmental stresses capable of inducing injurious strain in the vegetation, (ii) the types of syndromes indicative of injurious strain, (iii) the effect of the strain on the normal reflectance patterns, (iv) the effects of spectral changes on the aerial photographs. The primary manifestations of injury are physiological or morphological. Physiological injury usually reflects disruption of plant functions including photosynthesis, membrane function or translocation. Normal chlorophyll absorbs red and blue light and transmits green. Damage to chloroplasts can produce detectable changes in the spectral reflectance from all three of these wavelength regions (Murtha 1981). Other forms of physiological disturbance may be manifested eventually as morphological change. For example, severe osmotic stress caused by saline soils interferes with the translocation of water. Such stress may not be noted until the cells lose turgidity and the plant wilts. Other physiological changes may result in morphological manifestations such as reduced growth, loss of foliage, top-killing and necrosis. Variation in spectral reflectance is noted for different species and also within species depending upon the phenological stage. Watson and van Ryswyk (1986) found range plants could be separated on the basis of different plant colour causing different spectral signatures. Part of the variation was due to the species being in different phenological states. Vegetative, heading (boot), flowering and senescence stages all contributed different reflectances. Thomson et_al. (1985) were able to detect encroaching 25 smooth brome (Bromus inermis Leyss.) and timothy (Phleum pratense L.) in rough fescue (Festuca scabrella Torr.) as the invading domestic species were a senescent yellow-brown two weeks before the fescue. Range plants showed a different plant colour for each of the phenological stages of initial growth, flowering and senescence. Watson and van Ryswyk (1986) also observed differences in the timing of development in response to elevation, moisture, temperature, soil type and site condition which affected spectral signatures. Within a given area, sampling of the vegetation can give a generalized spectral reflectance pattern that may be considered "normal". Strain symptoms affecting spectral reflectances represent the basis of the interpretation since they represent a deviation from the normal pattern (Murtha 1981). Similarly, the intrusion of undesirable plant species can be detected by the different spectral signatures they exhibit. Tueller et_al.(1982) has suggested that remote sensing data are best interpreted when supplemented by ground truthing using measured soil and vegetation parameters. Such ancillary knowledge is required to prevent the misidentification of vegetation or soil mapping units and to provide reliable correlations between the soil/plant system under study and the remote sensing technique. 3.51 Spectral Changes in Aerial Photographs There is a direct relationship between spectral reflectance and the final image seen in aerial photographs. The relationship can be expressed in terms of cause (exposure due to spectral reflectance of selected subject matter) and effect (individual dye layer densities) (Lillesand and Kiefer 1979). . When a normal green leaf is photographed with colour infrared (CIR) film, it is seen as a magenta hue (blue plus red). Since CIR film works on a subtractive dye-layer principle, relatively little red light is subtracted from the final image; the higher NIR reflectance produces a thin layer of cyan dye. The yellow dye-forming layer responds to green light and although there is a low level of green reflectance, the dye-layer density of yellow pigment is very thin and consequently blue light is transmitted through the yellow 26 dye-layer (Fritz 1967). In contrast, the magenta layer is relatively thick and absorbs most of the green light. Visual combination of transmitted blue and red wavelengths gives the final image a magenta hue (Murtha 1981). With any decrease in NIR reflectance, the density of the cyan dye-layer increases and green foliage will appear as a darker magenta. If the NIR reflectance is increased, the resulting image will be shown as a lighter magenta (Murtha 1981). The final image developed from aerial photographs may be affected by variations in spectral sensitivity of the film emulsion, film age, film filtration, exposure time, processing and atmospheric attenuation (Fritz 1967). Considerable variation in spectral reflectance occurs from leaf-to-leaf due to the high dependence on leaf orientation and the direction of the incident light source. By increasing the distance from the subject the variations in the individual leaves are integrated and the average effect can be detected (Fritz 1967). This phenomenon, known as image merging, can result in considerable change in the final photo image. Choosing the proper scale is important in order to discern significant variations. 3.52 Remote Sensing Applications Remote sensing technology has been widely applied to the study of soil/vegetation relationships. Lord and McLean (1969) studied black and white aerial photographs of Chernozemic soils in the Princeton, B.C. area. They found significant relationships between soil-vegetation physiographic units and photographic patterns. Colour infrared aerial photographs has been used in southern Alberta (Thompson 1979) for range evaluation and in the detection of saline soil units. Landsat information in all available wavelengths has been used in saline soil and mapping projects in southern Alberta (Thompson et al. 1984, Summerfeldt et al. 1984), North Dakota (Dalstead et al. 1979), Montana (Chaterverdi et al. 1983) and Texas (Everitt et al. 1977) in both arable agriculture and range management. 27 Colour infrared aerial photographs possess two major advantages over the black and white medium: i) additional tonal dimension is added for photointerpretation; ii) colour infrared is sensitive to changes in reflectance of infrared radiation from vegetation caused by species differences, phenological state or plant stress (Dalstead etal. 1979). Dalstead et al. (1979) used colour IR photos supplemented by thermal IR measurements to detect and categorize saline seeps. Seeps were observed to evolve through a series of development stages: (i) Incipient - the earliest stage, indicated by wetness, but not excess salinity, (ii) Intermediate - indicated by the appearance of salt-tolerant vegetation and excess wetness, (iii) Mature - characterized by a salt crust and excess wetness. Seep progression beyond the intermediate stage was dependent on the flux and continuity of flow of saline water (Dalstead et_al. 1979). Variability was reported to be a function of aspect, developmental stage, vegetative cove and composition, amount of salt crust and wetness. These factors produced a range of seep signatures which Dalstead et al. (1979) described according to the following features: i) irregular shape, ii) abrupt boundary in fallow areas; diffuse in small grain and rangeland, iii) mottled pattern, iv) magenta colour; particularly noticeable following harvest, v) salt crust, brilliant white with abrupt boundary, vi) small areas of standing water. These authors found that intermediate and mature seeps were best identified with CIR photographs. Ladouceur et_al. (1986) used CIR transparencies at a scale of 1:10,000 to categorize damaged crops into three mapping classes. Crown (1979) found crop discrimination among alfalfa, alfalfa-grass, canola and small grain spectral signatures which was aided by a priori knowledge of the phenological state of the crop. Crown (1979) developed spectral signatures based on a crop calendar. Thompson et_al. (1984) found 1:5,000 positive transparencies provided accurate biomass estimates to within 440 kg ha"-*- using interpreted texture in their studies of parkland seeded pastures and northern mixed grass prairie pastures. Tucker et_al. (1980) in attempting to relate crop radiance to alfalfa agronomic values, found significant inverse correlations between red reflectance and canopy cover and biomass, while NIR reflectance was directly related to the same parameters. These authors suggested that drought-induced stress in these studies reduced in vivo chlorophyll concentration by photo-oxidation at a rate dependent upon the leaf water potential when the forage water content becomes limiting. Consequently, no increased reflection from the red (chlorophyll absorption) band was observed. In an extension of this type of technology, Vickery et al. (1980) assessed the fertilizer requirement of improved pastures using radiometers. They gathered digital information which could be analyzed by a computer process similar to that used in satellite imagery. Smith et jd. (1987) used micro-computer based Geographic Information System (GIS) and image analysis techniques as a field management tool. Soil and forage characteristics are incorporated into a GIS system. Spatial information allows the producer to analyze soil and production variability in the field with the goal of devising variable fertilizer rates. Crop quality, as indicated by foliar K and P content was found to be related to red/infrared pixel values obtained from digitized CIR aerial photographs. There have been many applications of remote sensing techniques in vegetation classification. The Large Area Crop Inventory Experiment (LACIE) established the applicability of multispectral remote sensing to inventory and monitor global wheat production (Bauer et al. 1980). Driscoll et al. (1974) used image density differences in 29 colour infrared aerial photos to identify five general plant communities: ponderosa pine (Pinus ponderosa Dougl. ex Laws.), spruce-fir (Picea engelmannii Parry ex Engelm., - Abies lasiocarpa Hook.), aspen (Populus tremuloides), big sagebrush (Artemesia tridentata Nutt.) and native grassland. McGraw and Tueller (1983) used Landsat computer-aided analysis techniques to map a sagebrush grass community in northern Nevada. Supervised, and unsupervised classifications yielded 14 classes representing eight range plant communities. Brach and Mack (1977) found ratios of green, red and near-infrared reflectance collected from radiometer data to be useful indicators of aerial crop maturity. Lo et al. (1986) evaluated single date and multi-temporal application of Landsat MSS data. Their particular application merged the unsupervised classification technique with "temporal profiles" of the crop species they sought to inventory. They successfully discriminated 9 classes: woods, pasture/grass, corn (Zea mays L.), alfalfa (Medicago sativa L.), oats (Avena satiua L.), bare soil, water, mixed vegetation and impervious soils. 30 4.0 MATERIALS AND METHODS 4.1 Sampling Design The study site was partitioned into five separate "fields" on the basis of natural boundaries such as forest cover or water courses (Figure 2). The fields were then pre-typed into subunits on the basis of topography from 1972, 1:15,840 black and white aerial photographs. A systematic sampling design was incorporated covering all fields with a 50 m x 50 m grid. Surveying took place from June 10 - 27, 1986. 4.2 Plane Surveying The east/west (E/W) azimuth, which comprised the abscissa of the grid is an astronomic bearing referred to the meridian through the Southwest (SW) corner of the Northwest (NW) 1/4, Sec. 6, Tp. 76. This traverse originated at a B.C. Pipe Post (B.C.P.P.) which marked the corner of Section 1 and Section 6, of Tp. 78 located during a resurvej' of the Exterior Rectilinear boundaries of the Sandy Harry Indian Reserve No. 4 and Alkali Lake Indian Reserve No. 4a (Ringwood 1981). This B.C.P.P. serves as a reference point for the west end of the study area (Fields 4 and 5). A boundary traverse was completed enclosing two other B.C.P.P.'s at an azimuth of 90° to the NW corner of Sandy Harry Indian Reserve No. 4. The east end of this line serves as the reference point for Fields 1 and 2 and marks the corners of the NW and NE 1/4 of Tp. 76. A transit level and leveling rod were used to plot N/S transects at E/W intervals of 50 m. A bench mark of 912.0 m was located on site from the 1:50,000 Topographical map 920/NE Chilcotin River (BCMOE 1980). This bench mark served as the reference to the principal points on the main E/W traverse. Sample points were marked by wooden stakes and located at 50 m intervals on the N/S transects. Elevation was determined for each sample point by standard plane surveying and leveling techniques (Brinker 1969). The location of the sample points was recorded using numerical coordinates to facilitate sample location on maps and air photos (Figure 3). 31 Figure 2 - Alkali Lake Indian Reserve 4, 4a - Location of f ields and field subunits. Alkali Creek LEGEND Scale 1=5000 Reserve Boundaries Field Sub-unit Ponds _ E i r x ! r ^ 2 ^ L K A L ^ ^ 33 Figure 3 - A lkal i Lake Indian Reserve 4, 4a - Locat ion of f l ight l ines and tra in ing sites. LEGEND Scale 1-.5000 —(T)— Flight L ines 16 Training Sites Reserve Boundaries ~ Fences F I G U R E 3 - A L K A L I L A K E IR_J*___ 35 A 50 m x 50 m grid was chosen to complement the intended map and airphoto scale of 1:5000. 50 m = 1 cm at this scale. This relationship ensured that map units were no smaller than 1 cm , a scale desirable for map presentation and legibility (Valentine and Lidstone 1985). 4.3 Soil Sampling A systematic sampling strategy was used in which selected units were at regular distances from one another, in two dimensions. This procedure was used in an attempt to ensure better coverage of the population than a simple random or stratified random sample. Theoretical studies have assumed a correlation exists between units close to each other and that this correlation decreases exponentially with the distance between the units; these assumptions have been supported by empirical studies (Black 1965). Systematic designs cannot be used without consideration of the form of population variation. Saline seepage mechanisms involve the transfer of soluble salts from upper-elevation "recharge" sites to lower-elevation seepage sites. The systematic design ensures equal sampling density among the topographic subunits and meets constraints for mapping at 1:5000 (Sec 4.2). 145 sample points were distributed evenly among Fields and sub units on the 46 ha. site. This sampling density exceeds the criteria for Survey Intensity Level 1 (one inspection per 0.1-4 ha; inspection spacing less than 100 m) - appropriate for a mapping scale of 1:5,000 (Valentine and Lidstone 1985). In addition, it was critical that the sample points be systematically placed to facilitate location of the points for later densitometric studies. Composite samples were collected from the 0-30 cm at each training site. 36 4.4 Soil Physico-Chemical Analyses 4.41 Sample Preparation Soil samples were air dried in the laboratory, ground with a hammer mill and passed through a 10 mesh sieve to remove all particles larger than 2 mm in diameter. The sieved soil samples were stored in air-tight containers prior to chemical and physical analysis. 4.42 Chemical Analysis All chemical analyses were carried out on air-dried, sieved, soil samples following standard analj'tical techniques. Soil pH was measured electrometrically with a Fisher Accumet Model 420 Digital pH/ion meter. A 2:1 water: soil ratio was used to provide sufficient clear supernatent for measurement. A 2:1 CaC^soil ratio was also evaluated for all samples. A 0.01 M CaCl£ solution is approximately equivalent to the total electrolyte concentration of the soil solution at optimum field capacity (Black 1965). Therefore, pH (CaC^) should provide a more accurate reflection of field conditions. Such pH values are commonly 0.5 pH units lower than those estimated using water dilution. Electrical conductivity was evaluated with a Radiometer Conductivity Meter Type 2E from extracts of 2:1 H^O to soil ratios (Black 1965) Sodium hexametaphosphate was not added to prevent CaCOg precipitation, as the samples were not allowed to stand before measurement. Available phosphorus was determined by Olsen's method of colorimetric determination using a Pye Unicam Spectrophotometer (Black 1965). This method is more accurate for soils exhibiting pH values above neutralitjr ( > pH 7.0). Total nitrogen was analyzed by the Semi-micro Kjeldahl method, followed by colorimetric analyses. The method utilized a Technicon block Digester and Technicon Auto Analyzer (Black 1965). Organic matter content was determined by the Walkley-Black method and reported as % organic carbon (Black 1965). Available cations and exchangeable cations were 37 estimated on ammonium-acetate extracts followed by atomic absorption spectrophotometry analysis using a Perkin Elmer Atomic Absorption Spectrophotometer Model 373. Both exchangeable (cations held by charge on the exchange complex) and available (water-soluble) cations were calculated; a correction for available cations may be important for dry soils (Black 1965). 4.43 Phj'sical Analyses Bulk density was estimated by the excavation method. A quantity of soil was removed, dried and weighed, while water was used to determine the volume of the excavation. This method was developed for use in gravelly soils and is preferred to methods such as the core method for such soils (Black 1965). Particle-size analysis was completed using the pipette method which is a sedimentation procedure involving repeated sampling at controlled depths and times (Black 1965). 4.5 Plant Sampling 4.51 Canopy Cover Canopy cover estimates were performed using a stratified systematic sampling design from Ju\y 8 - 28, 1986. Cases were selected on the basis of their location within the pre-typed management units and sub-units. This approach ensured that the variety of plant micro-communities encountered could be sampled while a minimum of four cases were evaluated per sub-unit. The point sampling method (100 points/training site) was used for each individual case and species present were recorded as a percentage of the total canopy cover (See Appendix 2). Canopy cover was tabulated on a per species basis as well as for the general groups; alfalfa (ALF), salt-tolerant grasses (SALT), domestic grasses (GRASS), forbs (FORB), and sedges (SED). Five 10 m parallel transects were centered on the survey point at 2.5 m intervals. Twenty point estimates of species frequency were performed per transect (5 x 20 = 100 total). Species were recorded on a percentage basis. Replicated biomass clippings were obtained on July 6, 1986 from a 38 subset of the canopy cover plots. Stratification was carried out on the basis of four samples/ management unit, one sample for each of three elevation/ soil texture classes (low elevation/fine textured - middle elevation/loam - upper elevation/stony coarse) and one sample chosen from outside these classes. A survey of the visual characteristics of the plant communities in the pre-typed sub-units was completed from June 12 - 15, 1986. Vegetation in the sub-units was evaluated in terms of species composition, phenological state, height, colour, and observations including visible soil characteristics, the appearance of SALT species, and general vegetation patterns. 4.6 Aerial Photography 4.61 Photographic Mission Positive transparencies and paper prints were obtained in the 23-cm format from a photographic mission flown by Selkirk Remote Sensing on June 26, 1986 between 11:19 AM and 11:35 AM PDST. Stereo photographs were taken with a Wild-Herbrugge-RClO wide-angle aerial reconnaissance camera (focal length 152.4 mm) with a Universal Aviogon lens set to f/5.6 and using a Wild 525 nm, Anti-vignetting (2X) filter and Kodak Aerochrome Infrared 2443 film. The resultant image was produced at a scale of approximately 1:5,000. Altitude varied during the flight from 5450 - 5500 feet ASL on a magnetic heading of 242°. 4.62 Densitometry A Macbeth transmission/reflectance densitometer (Model TR521) was used to record density readings of the images. Density measurements of the yellow, magenta and cyan dye-layers were made using blue, green and red filters respectively (Hall et_al. 1983). Data points used for densitometry corresponded to the topographic survey points. An examination of the variation induced by the positioning of the image window was carried out using a MacBeth densitometer. Alfalfa variation between photographs was measured for the same cases on three successive photos. The cases were located on the 39 left, center and right sides of the photo, respectively. Variation among the different photographs was examined by grouping the cases in the data bank by photo and then comparing mean values for plant, soil and spectral parameters for each photograph (Dixon 1985). Positive transparencies and paper prints were compared by dye-layer density in the green, red and infrared region of the spectrum by recording transmission and reflectance data of four vegetation types. 4.63 Digital Analysis The five previously designated Management Units were identified on the aerial photographs and images were selected which would place these study areas in the center of the image. This approach reduces the vignetting effect of the camera system which can introduce unnecessary variability into the image (Moore 1980). The selected areas were then "windowed" out of the complete image and digitized in the U.B.C. Laboratorj' for Computational Vision using an Optronics Colornation C-4500 colour film scanner with an optical aperture of 100 um. For a scale of 1:5,000, 100 um = 0.5 m on the ground area. This relationship corresponds to the sampling interval on the plant transects. Scanning was completed in the yellow, magenta and cyan dye-layers and the pixel matrices were stored on computer tape. The numerical scale of pixel values ranged from O for black to 255 for white (numerical pixel value = digital number (DN)). The information was then transferred via UBC.Net to the Meridian PC image analysis system in the Faculty of Forestry Remote Sensing Laboratory. Ground reference (training sites) points were located on the image and a sampling grid was constructed to locate the sample points. Each point was sampled using a 5 X 5 pixel matrix corresponding to a ground area of 2.5 m x 2.5 m. The recorded value for the DN is the mean of the 25 pixels. 40 The sampling cursor was located at each topographic sample point and (DNs) in the Yellow, Magenta and Cyan dye-layers was recorded. Ratios and products for the dye-layers were generated and also included. This data bank of 84 training sites was then used in the derivation of relationships between inferred spectral, plant and soil parameters. 4.7 Statistical Methods Data analysis was performed using procedures in the BioMedical Data Processing Package (BMDP) unless otherwise specified. 4.71 Correlation Analysis A correlation matrix was run on the data bank using BMDP3D (Dixon 1985). Correlation analysis provided the basis for the initial parameter selection for use in predicting spectral-plant-soil relationships. An arbitrary selection threshold of r > 0.35 (p < 0.05) was used in this initial selection. Parameters which were not significantly correlated with others (p <C 0.05), which exhibited correlations of < 0.35 or which expressed redundant information were removed from the analysis. 4.72 Morphological Analysis A morphological analysis was performed on the spectral-plant-soil system and on species component systems (Chorley and Kennedy 1971). This sort of analysis is used to develop the relationship between parameters in a system in terms of their positive and negative correlations. The scatter of points is presented, not in x and y, but in r, (correlation coefficient) a dimensionless unit and presented in box-diagram form. The sj'stem parameters are considered to be non-additive in that some factors may replicate or reinforce the effect of others. Correlations between parameters are significant at p _< 0.05 with no direction of causation implied. The sign of the correlation coefficient indicates the nature of the relationship (Chorley and Kennedy 1971). These systems aid in selection of the parameters which best identify the plant/soil relationships on the site and the spectral parameters useful in delineating them. 41 4.73 Cluster Analysis Cluster analysis was used to group training sites. The spectral parameters are utilized as grouping parameters. UBC C Group / C GMEM (Patterson and Whitaker 1982) is a multivariate classification method which groups individual cases such that within group variation is minimized. Optimum groupings are determined by defining each case as a group and reducing these N groups by a series of step decisions until all cases have been classified into one of the desired number of groups. Raw data are used to calculate the variance for each pair of items. Each grouping is made by combining the two items with the minimum variance. 4.74 Variability Assessment Differences (p < 0.05) among groups were determined using parametric and non-parametric t-tests. BMDP7D (Dixon 1985) calculated the range, mean, standard deviation and coefficient of variation. Variability among parameters can be directly compared using the CV statistic. For mean comparisons with unequal variances (determined by Levene's test) the Welsh separate variance t-test was used. Bonferroni probabilities were calculated by dividing the level of significance (* = 0.05) by the number of groups (g) - ("Bonferroni probability" = o C /g) (Dixon 1985). 4.8 Supervised Classification Following the development of spectral signatures for the desired classes, this information is integrated into the Meridian PC-image analysis system. The system uses the means provided and standard deviations of 30 units, or at a point midway between two means for three dye-layers. The classifier assigns pixels to the closest class (by the sum of squares of distance) in multi-dimensional intensity space. This is known as a "nearest neighbour" or "minimum distance" classification (McDonald, Dettwiler and Associates 1986). Image smoothing is accomplished by means of a 5 X 5 pixel filter. This approach re-assigns isolated pixels to larger, adjacent groups to improve visual presentation. 42 5.0 RESULTS AND DISCUSSION 5.1 Introduction The fundamental objective of this study was to detect variation in reflectance between different plant communities indicative of a range of saline soil properties. Development of a remote sensing system necessitates a systematic analysis of the components of a system to discover their relationships; it ma}' also describe the effects changes in these relationships produce in spectral reflectance. Thus, interpretation of reflectance patterns should not proceed without first knowing the capabilities and limitations of the remote sensing technique. Analysis of the potential confounding influences of processing, positional, geometric and atmospheric effects must precede the development of spectral-plant-soil relationships. Specifications of the photo mission and image quality factors are examined along with densitometry studies of the stepwedge, paper prints and positive transparencies to determine the quality of the image for subsequent plant-soil interpretations. The training site digital reflectance data are then examined by image position to locate anj' variability between photo images. The first stage in developing the system model is the determination of the key parameters. Correlation and morphological systems analysis are used to select these parameters and to describe relationships between them. Morphological systems analysis identifies, describes and interprets the associated physical parameters of this system. Spectral parameters most strongly associated with plant species of interest (ALF, SALT, GRASS) are used, as grouping parameters in cluster analysis of the training sites. This procedure groups the parameters such that total within-group variation is minimized. Different spectral signatures (in Y, M and C reflectance) exhibited by these groups are the result of unique plant communities indicative of severity classes of saline soils. The validity of the groups is tested by mean comparisons. Derived classes were used to provide digital numbers (DNs) for supervised computer-assisted classification of the 43 entire study site and the development of continuous classifications for plant and soil parameters. The resultant model will portray an alfalfa capability assessment of study site soils. 5.2 Film Selection and Photographic Control Sensitometric control is necessary for all quantitative densitometric film anatysis. The step wedge is a form of exposure control consisting of twenty-one "steps" of known density which attenuate the light source into nominal steps of 0.15 logE where E is exposure measured in lux<sec? (Lillesand and Kiefer 1979). After processing, the individual steps were measured with a densitometer and plotted against the step number and a scale of relative logE. The gamma (v) values for the step wedge are 2.89 (Yellow), 3.22 (Magenta) and 2.50 (Cyan). The plot of the "step wedge density (D) versus logE is shown in Figure 4 for Kodak 2443 CIR film. The exposure latitude of the film is the range of logE values which will yield an acceptable image. For interpretive purposes density values should be on the linear portion of the curve or a fraction of the toe (Lillesand and Kiefer 1979). Subtle variations in spectral reflectance cannot be detected if values register on the toe or shoulder of the curve as there is little change in density for large changes in exposure. Gamma (D/logE), the slope of the linear portion, is a determinant of contrast. The higher the gamma value, the higher is the contrast. A high contrast film has its scene-exposure range distributed over a large density range (Lillesand and Kiefer 1979). Density values for alfalfa fall within the linear portion of the spectral curve. Densitj' values for alfalfa are, therefore, in the most appropriate range for interpretation. The speed (sensitivity) of the cyan dye-layer has been made deliberately slower than the yellow and magenta layers. If the cyan layer was as sensitive the IR would be over-exposed and consequently register on the toe of the curve (Fritz 1967). The IR balance of a film relates the speed of response to infra-red and visible light. An IR balance of 35 is designated as "normal" (Fleming 1978). Values below this are called enhanced INTEGRATED RESOURCES PHOTOGRAPHY LTD. A e r i a l f i l m S a m i t o m a t r l c • l a t l i n g S h « « t •1 4 < * •* •» « | Log E (lux u c l Density - log exposure curve. Resu l ts of the f i lm step wedge (21 steps of known dens i ty wh i ch attenuate the^ight source into nominal s teps of 0.15E, when E is the exposure in lux sec ). Image densi ty in the Ye l low, Magenta and Cyan dye- layers is plotted against the step number (relative log E). 45 and above are degraded. Enhancement of film is advocated when attempting high altitude photography. The longer path length and increased water vapour concentration result in IR light absorption and upward scattering of visible light. Thus the goal is to enhance IR sensitivity relative to visible light. Enhancement of the relative IR sensitivity is accomplished by shifting the IR curve to the left (IR enhancement) or by shifting the visible curves to the right (visible attenuation). Kodak CC Blue filters attenuate visible radiation while allowing almost unimpeded transmission of IR radiation (Fleming 1978). In practice, Kodak CC filters in the optical path can cause image loss (P. Williams, Selkirk Remote Sensing; personal communication, March 1987). Real IR enhancement occurs when the IR curve is moved to the left by means of a genuine increase in speed of the IR-sensitive layer. The IR-balance of the film used in this study is 22, significantly enhanced for an altitude of 5450-5500 feet ASL (Figure 5). This enhanced sensitivity (more appropriate for an altitude of 12,000-CCRS) produces an image which is more red or magenta due to the earlier response of the cyan layer (IR balance = 28 for 5500 ASL). In some cases the bright magenta tones produced may overpower the scene and slight variations in IR reflectance are indistinguishable. For example, the senescent vegetation in the rangelands adjacent to this studj' is highlj' reflective of IR radiation; the image scene for rangelands is overexposed by the equivalent of 1/2 stop (P. Murtha, Dept. of Soil Science, U.B.C; personal communication, February 1987). In spite of this disadvantage, the enhanced sensitivity of the film was preferred because of the focus of the study on alfalfa-growing units. Exposure fall-off, the variation in focal plane exposure associated with the distance an image point is from the image center was avoided by incorporation of a 2X-anti-vignetting filter in the Wild 525 nm (minus-blue) filter. This combination is well within the constraints described by Moore (1980) for maintaining uniform illumination over the area of interest. Figure 5 - Altitude Aim Curve. Determination of the optimal IR Balance (difference in sensitivity of the film to infrared and visible light) for the anticipated flight altitude for aerial IR photography (Moore 1980). 47 Since natural materials are not truly diffuse reflectors, their reflectance depends on the angle from which they are illuminated and viewed (Lillesand and Kiefer 1979). Fritz (1967) noted the high dependence of reflectance on leaf orientation. The angular sun-object-image relationship is defined in terms of three angles; the solar elevation, the azimuth angle and the viewing angle. On June 26, 1986 at 11:30 am the solar altitude and solar azimuth were 61° and 165.8° respectively for the photographic mission flown at 52°N latitude. The viewing angles for the three photographs were 76.1° (PH8882), 76.2° (PH8883) and 76.4° (PH8884). As this study is based on one photo mission only, no variability was introduced through solar elevation and sun azimuth angle. The time and date of this photo mission yields the best photometric information due to maximum brightness (Egan 1985). Photographic quality and scene variation may also be affected by viewing angle (the angle between the camera and the sun) and by differential atmospheric scattering. The aerial photographs were taken at.nine second intervals with a ground speed of approximately 200 km hr"*. Viewing angles calculated for a three-photo series were found to vary between 76.1° and 76.4° ; these differences are negligible. The quantity of scattered light entering the lens will be influenced by prevailing atmospheric conditions. As the flight was made on a clear, relatively calm day with little haze, differential atmospheric scattering is likely to have no significant impact on scene variation. Variation among photos was detected by recording values for the same cases on three successive photos (Table 1). 48 Table 1. Positional effects on dye-layer density measurements from CIR film. Cases were selected from areas on the left, center and right portion of three successive photographs for density measurement (n = 25). Values followed by different letters in the same row are significantly different at P<0.05. Dye-layer Density (x) Position Left Center Right Yellow 0.96 (2.3)1 1.17 (2.0) 1.61 (2.8) Magenta 1.23 (2.6) 1.41 (2.8) 1.83 (3.0) Cyan 0.62 (2.4)b 0.70 (1.9)b 0.88 (3.8)a Figures in brackets represent the coefficient of variation (%) The cases were located on the left, center and right respectively. Although this procedure violates evidence from Moore (1980) concerning exposure fall-off it is useful in that exposure was found to have left to right variation as well as radial variation. Differences (p _< 0.05) in density were found only between the right and left photos. Viewing angle did not change appreciably, but there may be some overexposure due to specular reflection on the left side in the IR. No differences (p _< 0.05) were found between the left and center in the yellow and magenta dj'e-layers. These data suggest that variation may have resulted during processing with the right side of the scene being underexposed. The results of a comparison between positive transparencies and paper prints by dye-layer densit}7 in the green, red and infrared region of the spectrum are shown in Table 2. Transparenc3' values were significantly different (p < 0.001) from print values. The print values were consistently lower in all comparisons and the variation was also consistently less. These lower values would be associated with less exposure latitude and, therefore, reduced ability to discriminate subtle differences. The density values obtained from paper prints lie on the toe of the transparency sensitometric curves for all three dye layer sensitivities. There is no step wedge available for the paper prints therefore no sensitometric plot can be made. Quantitative radiometric analysis of paper prints is Table 2 - Comparison of dye-layer density and total film response for positive transparencies and paper prints from colour infrared aerial photographs of an alfalfa stand. Values shown are the means and standard deviation for each parameter measured. Transparencies Paper Prints Dye-layer densities1 Y 1.20 + 0.001 1.03 + 0.021 M 1.46 + 0.029 1.15 + 0.024 C 0.67 + 0.011 0.59 + 0.006 Total film response^ G 1.45 + 0.025 1.24 + 0.026 R 1.36 + 0.024 1.08 + 0.021 NIR 0.35 + 0.004 0.31 + 0.002 Density = Iog10 (1/Transmittance) Density values are obtained by calibrating the MacBeth TR densitometer with a Step Wedge having 21 steps (1 step = 0.15 logE) of known transmittance. Total film responses in the green, red and near infrared regions of the spectrum were determined by using the weighted combination of yellow (Y), magenta (M) and cyan (C) dye layers given by Moore (1980): green = Y + 0.12M + 0.12C; red = 0.88M + 0.36C; NIR = 0.52C. 50 not possible. Thompson etal. (1984) in their technology transfer development paper, found biomass could be analysed by 1:5,000 35 mm CIR slides and 1:20,000 23 cm positive transparencies. Paper prints were found to be unsuitable for parametric vegetation analysis. Moore-transformed data (Hall etjal. 1983) were calculated using the spectral response curve for the Wild Pan 525 nm filter. The total film response was evaluated as each dye-layer has some response in wavebands other than the band in which it peaks. For example, the IR layer peaks in the near-infrared, but 48% of its response area is in the red and green spectral regions (Moore 1980). Classification of vegetation types without considering the total response of the film could lead to biased results. The significantly different relationship between the prints and transparencies was maintained using Moore-transformed data. Hall etal. (1983) also found their classifications were unchanged when Moore-transformed data were used. 5.3 Comparison of Spectral Data by Photograph The size of the study area (46 ha) precluded capturing the entire site on one photo at the selected scale of 1:5,000. Indian Reserves 4 and 4a were covered by three flight lines. Four image windows were selected from the most appropriate photographs located on two of the flight lines (See Figure 2, page 32 ). Selected photographs were located as close as possible to the photograph principal point. Image windows from E to W are PH 8882 (Field 1), PH 8909 (Field 2), PH 8883 (Fields 3" and 4) and PH 8911 (Fields 4 and 5). Figures 6 - 9 present these CIR images. The training sites were grouped by photograph to detect any variations in spectral reflectance between photos. Parametric and non-parametric t-tests compared digital number means for all spectral parameters among photos. Plant, soil and spectral parameters were analyzed by photo (Table 3). Should substantial variation be detected which is not explainable in terms of plant or soil parameters the combined data bank should not be used in derivation of spectral signatures Figure 6 - PH 8882 - Colour infrared image of Field 1, Alkali Lake Reserve #4, 4a. Figure 7 - PH 8909 - Colour infrared image of Field 2, Alkali Lake Reserve #4, 4a. Figure 8 - PH 8883 - Colour infrared image of Field 3 (right side and the eastern half of Field 4, Alkali Lake Reserve #4, 4a. Figure 9 - PH 8911 - Colour infrared image of the western half of Field 4 (right side) and Field 5, Alkali Lake Reserve #4, 4a. Table 3 - Plant, soil and spectral parameters used in the analysis of aerial colour infrared photographs. Type Abbreviation Units SPECTRAL yellow magenta cyan Product 1 (y*m) Product 2 (y*c) Product 3 (m*c) Ratio 1 (y/m) Ratio 2 (y/c) Ratio 3 (m/c) Ratio 4 (m/y) Ratio 5 (c/y) Ratio 6 (c/m) y m c PR1 PR2 PR3 R1 R2 R3 R4 R5 R6 0-255 (Black-White) PLANT Alfalfa (Medicaqo spp.) Salt-tolerant grass species) Domestic grasses Poa pratense  Bromus inermis  Distichlis stricta  Hordeum jubatum  Puccinellia nuttallia  Poa juncifolia  Artemesia friqida Biomass ALF SALT GRASS POPR BRIN DIST HOJU PUNU POJU ARFR BIOM % canopy cover kg ha-1 SOIL-CHEMICAL pH - water pH - calcium chloride Electrical conductivity of saturated extract Available phosphorus Total nitrogen Organic carbon Available Ca Mg K Exchangeable Ca Mg Na K pHW pHC EC PHOS TN ORGC ACA AMG AK ECA EMG ENA EK -1 mmho cm ppm % cmoles + kg -1 SOIL-PHYSICAL Elevation Bulk Density Particle size - Sand Silt Clay E BD SAND SILT CLAY metres ASL 3 kg m % % % 56 since such spectral differences may be accounted for by film/flight/processing variation. The results of the photograph comparisons are presented in Table 4. 5.31 Plant Parameters Plant parameters ALF (alfalfa) and GRASS (grass) showed significant differences (p _< 0.05) between PH 8909 and PH 8883 and between PH 8909 and PH 8911 respectively. Field 2 (PH 8909) had a vigorous ALF/GRASS stand over most of the field while Fields 4 and 5 (PH 8883 and PH 8911) exhibited variable stands of ALF and denser populations of GRASS. No other plant parameters were significantly different between photographs. 5.32 Soil Parameters With the exception of exchangeable K + (EK) no soil parameters were significantly different. This result is a reflection of the relatively even distribution of saline, intermediate and non-saline soils among the five fields. Exchangeable K + was inversely correlated with ALF and significantly different (P < 0.05) between PH 8909 and PH 8911. This difference could, in part, account for the reduced alfalfa population in Field 5. Severe alfalfa winterkill was experienced regionally in the fall of 1985 (See Site History). 5.33 Spectral Parameters Spectral parameters M (Red) and R^ (Red/Green) were not different (P _< 0.05) between photos, whereas yellow (Green) was different (P _< 0.05) between PH 8909 and PH8883. The vegetation in PH8909 had a higher population of vigorous ALF/GRASS, while PH8883 had fewer such areas, as well as two large areas of senescent SALT vegetation. Furthermore, much of the GRASS in Field 4 was also senescent; thus, PH8909 would be expected to show a higher Y (Green) reflectance. The cyan (NIR) response was different between PH8883 and PH8909 and PH8883 and PH8911 respectively. PH8883 (Fields 3 and 4) exhibited lower NIR digital numbers due to the smaller population of vigorous ALF/GRASS and higher populations of SALT compared to PH8911 (Fields 4 and 5) and PH8909 (Field 2). 57 Table 4. Photo comparison summary. Dye-layer values are given as d i g i t a l numbers. Parameter PH8882 PH8909 PH8883 PH8911 (n=8) (n=23) (n=27) (n=26) Spectral Yellow x 123.0 131.1a 119.1b 124.0 s 14.4 10.5 17.7 11.0 %CV 11.7 8.1 14.9 8.8 Magenta x 91.8 93.4 90.2 88.3 s 23.7 17.7 31.2 20.6 %CV 25.4 18.9 35.6 23.2 Cyan x 164.1 174.9b 157.7a 166.9b s 10.1 20.1 12.8 7.4 %CV 6.2 11.5 8.1 4.4 Plan t ALF x 15.5 28.6a 9.5b 12.8b s 14.9 21.4 11.5 15.5 %CV 96.1 74.8 121.0 121.1 GRASS x 56.5 43.9a 58.3 63.4b s 19.9 19.9 25.0 28.6 %CV 35.2 45.3 42.9 45.1 S o i l EK x 1.01 1.01a 1.00 0.70b s 0.71 0.42 0.64 0.21 %CV 70.0 40.0 64.0 30.0 Means w i t h i n rows followed by the same l e t t e r are not s i g n i f i c a n t l y d i f f e r e n t (p < 0.05) 58 Examination of the data base revealed a number of cases with very low NIR (C) digital numbers. Four edge cases were excluded from PH8883 (54, 58, 59, 65), and one from PH8882 (27), as they were judged to be affected by shadowing or vignetting. A further three cases (56, 57, 60) were excluded due to low Y values (also a result of possible vignetting). These cases were located by the forested E border of PH8883. As a result of these exclusions Y (Green) and C(NIR) no longer significantly differed (P < 0.05) between PH8909 and PH8883. However, ALF remained significantly different (P < 0.05) between PH8909; 8883 and 8911; GRASS remained different (P < 0.05) between 8909 and 8883, and EK remained significantly different (P < 0.05) between PH8909 and PH8883. No change in the relationships of these parameters between photos suggested that the exclusion of these eight cases may be desirable to avoid biasing relationships between spectral, plant and soil parameters. Vignetting can introduce error into the statistical assessment of parameter relationships. These data are consistent with the evidence used by Moore (1980) to caution against the inclusion of edge measurements in digital spectral analysis of aerial photographs. 5.4 Parameter Selection At the onset of this analysis each case was described by 47 variables (Table 3). Chorley and Kenned.y (1971) described the effects of variables as being additive or non-additive; most are non-additive. Thus, when they are combined some parameters replicate the effects of others and some reinforce them. The goals of this phase of the project were to i) identify the key spectral, plant and soil parameters ii) remove those parameters which contributed no significant information or replicated key parameters iii) identify parameters which reinforce the effects of the key parameters. These selected parameters will be used in later stages of the analysis. To evaluate the relationships between spectral, plant and soil parameters a correlation matrix was produced using data for 47 variables and 84 cases (Dixon 1985). As described in Section 4.71, variables which exhibited r > 0.35 were retained for further analysis. A subset of 25 variables is presented in Table 5. 59 ALF SALT GRASS POPR BRIN HOJU OIST PUNU PHW EC 4 5 6 32 33 35 34 37 12 14 ALF 4 1 . 0000 SALT 5 -0. 3535 1 . 0000 GRASS 6 -0. 1823 -0. 65 IS 1. OOOO POPR 32 -0. 1610 -0. 4686 0 7815 1 . OOOO BRIN 33 0. 0811 -0. 3853 0. 3523 -0. 1339 1. OOOO HOJU 35 -0. 2E89 0. 6104 -0. 3158 -0. 3132 -0. 3933 1 . OOOO oisr 34 -0. 1797 0. 3S18 -0. 2172 -0. 3307 -0. 262 1 0. 4312 1 . OOOO PUNU 37 -0. 3177 0. 8595 -0. 4903 -0. 4247 -0. 3387 0. 5598 0. 3788 1 . OOOO PHW 12 -0. 3359 0. 6412 -0. 4069 -0. 3888 -0. 3217 0. 4643 0. 4458 0. 6670 1 . OOOO EC 14 -0. 2833 0. 5137 -0. 3038 -0. 3734 -0 2902 0. 4503 0. 6592 0 607 3 0. 8 167 1 . O O O O E CA 24 0. 3822 -0. 4521 0. 1667 0. 2049 0. 1577 -0. 3929 -0. 3807 -o 4402 -0. 5824 -0. 4866 E N A 26 -o. 2730 0. 5824 -0. 3803 -0. 4432 -0. 2626 0. 4943 0. 6215 0. 6486 0. 8578 0. 9383 EK 27 -0. 18 18 0. 4876 -0. 3558 -0. 3435 -O. 3230 0. 4666 0. 3725 0. S706 0 5744 0. 4652 PHOS 18 -0. 1208 0. 2772 -0. 123B -0. 1478 0. 004 1 0. 1285 0. 074 1 0 3257 0. 3869 0 2810 TN 19 0. 1864 -0. 2043 0. 1226 0. 10O0 0. 1111 -0. 0026 -0. 2105 -0. 2612 -0. 3336 -0. 4151 ORGC 20 0. 2710 -0. 2138 0. 0603 0. 0278 0. 1429 -0. 0058 -0. 2684 -0. 3171 -0. 3202 -0 4246 BD 15 0. 0447 0. 2951 -0. 1892 -0. 093G 0. 0786 0. 0073 -0. 1869 0. 1948 0 0856 -0 0585 £ 3 0. 5635 -0. 0623 -0. 2492 -0. 2168 0. 0006 -0. 0327 -0. 0099 -o. 104 1 -0 0007 -o. 0356 ¥ 40 0 0363 0. 1076 -0. 2014 -0. 2390 -0. 194 1 0. 1779 0. 3624 0. 1385 0 2864 0 47 13 M 41 -0. 2762 0. 3527 -0. 2485 -0. 2588 -0. 2402 0. 3302 0. 4376 0. 3578 0, 4486 0 .5773 C 42 0. 4520 -0. 3603 0. 0850 0. 07 18 0. 0516 -0. 2084 -o . 1739 -0 3568 -0 2915 -0 . 2228 02 44 -0. 2962 0. 4 152 -0. 3032 -0. 3275 -0. 2S06 0. 354S 0. 5174 0 4 4 4 4 0 .5490 0 6824 R3 45 -o. 3760 0. 4516 -0. 2735 -0. 2744 -0. 2413 0. 3716 0. 4577 0. 4590 0 .5163 0 .6063 B4 46 -0. 387 1 o. 4 4 4 S -0. 2467 -o. 2244 -0. 2201 0. 3547 0 3705 0. 4274 0 4435 0 . 4B45 PR t 47 -0 2149 0. 2860 -0. 2269 -0. 2532 -0 2472 0. 3061 0. 4701 0. 3053 0 4 304 0 .6095 ECA ENA EK PHOS TN ORGC BO E V M 24 26 27 18 19 20 15 3 40 4 1 ECA 24 1. OOOO E N A 26 -0. 5989 1. 0000 E K 27 -0. 3163 0. 52 16 1. OOOO PHOS 18 -o. 0973 0 3506 0. 3935 1. OOOO TN 19 0. 3135 -0. 3633 0 0774 0. 0403 1 . OOOO • RGC 20 0. 2984 -0. 3484 0 0652 0. 0631 0. 9151 1 OOOO 50 15 -0 OTOO -0 014 1 0 1 166 0. 0538 -0. 0383 -0 0235 1. OOOO E 3 0. 0529 0 0050 0. 14 11 -0. 1377 0. 148 1 0 2752 0. 0685 1. OOOO Y 40 -0. . 1977 0 3958 0 .2802 0. 0703 -0. 2420 -0 .2037 -0 0298 0. 3131 1 OOOO M 41 -o 3255 0 .5158 0 .4248 0. 1 143 -0. 2997 -0 2998 0 .0330 0 0954 0 .8058 1 .0000 C 42 0. . 1975 -0. .2511 -0. . 1929 -0. 0890 0 1 137 0 .1921 -0 0144 0. 4160 0 . 4379 -0 . 1321 R2 44 -0 .3813 0 6338 0 . 4609 0. 1616 -0 3533 -0 .3718 -0 .0105 0. 0307 0 74 18 0 9524 R3 45 -0 .3782 0 5655 0 .4638 0. 1405 -0. 3190 -0 .3420 0 .0430 -0 0161 0 6250 0 9580 R4 46 -0 3297 0 455 1 0 4211 0 . 1265 -0. 2783 -0 3037 0 .0834 -0 04 1 1 0 531 1 0 9225 PR t 47 -o. 3029 0 5306 0 .4039 0 0924 -0 2920 -0 2818 -0. .0090 0. 1526 0 .8819 0 9800 c R2 R3 R4 PR 1 42 44 45 46 47 C 42 1 .0000 R2 4 4 -0 . 2734 1 OOOO R3 45 -0 .4013 0 .9644 1 .0000 R4 46 -0 4458 0 .8943 0 .9763 1 .0000 PP.1 47 0 .0025 0 .9305 0 . 8992 0 .8320 1 .0000 Table 5. Correlation matrix for 25 selected spectral, plant and soil parameters Variables were selected on the basis of correlation analysis using those parameters which had correlation coefficients of r > 0.35. 60 5.41 Spectral parameters Twelve spectral parameters were included in the initial analysis. The following parameters were eliminated from further analysis; Y (green), PR2 (Y=|:C), PR3 (M:|:C), Ratio 1 (Y/M), Ratio 5 (C/Y) and Ratio 6 (C/M). These parameters showed little or no relationship to plant or soil parameters, or in the case of the ratios, higher R values were shown by their inverses (Ratios 4 (MAT), 2 (Y/C) and 3 (M/C) respectively. Y (green) was strongly correlated with EC (r = 0.47) as well as with DIST (r = 0.36) and ENA (r = 0.40). Y will be used in later stages of the analysis. 5.42 Soil Parameters - Chemical The available cations ACA, AMG and AK were removed from the analysis as the relationships between the exchangeable cations and spectral/plant parameters were stronger. Phosphorus (PHOS) was excluded at this stage in the analysis as it exhibited no strong correlations (r > 0.35; P < 0.05) with spectral or plant parameters. The relationship of PHOS with pHW and EK may be important indicators of effects on plant growth. Phosphorus management is often important in alfalfa production. However, its inclusion in the analj'sis at this stage would contribute no useful information toward an understanding of the spectral-plant-soil system. Total nitrogen, also an important element in plant growth, was highly correlated to organic carbon (ORGC) and the latter has been positively correlated with spectral parameters in studies by Smith et_al. (1987). Nevertheless, organic carbon was also / excluded from subsequent analysis of the spectral-plant-soil system at this stage as no correlations exceeding 0.35 were noted with spectral or plant parameters. Its relationship with E C g will be considered in the interaction of soil factors and the development of continuous soil classes. Of the soil reaction parameters measured, pHW exhibited strongly correlations with 14 different parameters. pHC (calcium chloride) showed a lower correlations with soil 61 and plant, parameters and was excluded from further analysis. 5.43 Soil Parameters - Physical Elevation showed very low correlations with the other soil parameters and was, therefore, excluded. Although an inverse relationship between elevation and salinity indicators may be expected for areas with saline seeps the different saline soil units of this study were located at different elevations. The complex topography of the study site confounded this relationship, although examination of individual seeps supported such an inverse trend. A subset of 20 cases collected from pre-typed areas representing low, intermediate and high saline conditions was analyzed to ascertain textural characteristics (% SAND, % SILT, % CLAY) (Table 6). All but two of the samples proved to be silt loams. Soils of the Fraser plateau are overlain by an eolian (wind-deposited) mantle (Valentine et_al. 1978). This la37er consists primarily of silts. The sampling depth of 30 cm proved insufficient to discern textural differences although varying bulk densities and coarse-fractions were observed. Percentage SAND, % SILT and % CLAY were eliminated from further analysis. Bulk density was retained in the analysis because of the high correlation (r = 0.55) observed with SALT. A stronger relationship was expected between bulk density and texture, although discarding the coarse fragments prior to textural analysis may have contributed to these observed low correlations. 5.44 Plant Parameters Of the initial 12 plant parameters, four were discarded from subsequent analysis; FORB, SEDGE, p O J U a n d ARFR. Although GRASS showed little correlation with any of the spectral parameters, the density of grass species varied widely across the study site and, therefore, must have contributed to the spectral signature. The nature of this contribution was not clear from the analysis of all domestic grass species collectively. GRASS was retained for use in later stages of the analysis. 62 Table 6 - Particle size analysis of soils sampled on the five fields of the study site. Location of the samples is indicated by the case numbers listed for each f i e l d . Particle Size (%) Textural Case Sand S i l t Clay Class (Field 1) 11 46. .4 43. .2 10.4 loam 12 28. .9 58. 5 12.6 s i l t loam 15 26. .3 62. 0 11.7 s i l t loam 26 27. ,6 59. 0 13.4 s i l t loam (Field 2) s i l t loam 36 29. ,4 61. 7 8.9 42 47. ,1 50. .6 2.3 sandy loam 46 26. .2 64. .7 9.1 s i l t loam 49 33. .8 59. .9 6.3 s i l t loam (Field 3) s i l t loam 65 31. .5 61. .2 7.3 79 32, .4 59. .4 8.2 s i l t loam 80 26, .0 64, .0 10.0 s i l t loam (Field 4) 82 37 .0 55. .4 7.6 s i l t loam 85 33 .7 53, .5 12.8 s i l t loam 86 18. .4 69 .8 11.8 s i l t loam 87 26 .1 65 .7 8.2 s i l t loam 109 37 .0 53 .8 9.2 s i l t loam 114 36 .1 57 .3 6.6 s i l t loam (Field 5) 125 26 .9 66 .2 6.9 s i l t loam 127 36 .2 55 .0 8.8 s i l t loam 133 31 .6 60 .9 7.5 s i l t loam 6 3 Removing BRIN and POPR will clarify the relationship between spectral and plant parameters as they exhibited low correlations. However, since the field was initially seeded with BRIN, and POPR was present prior to seeding, the field is an ALF/GRASS mixture, though predominantly ALF (See Site History, Section 2.1). Areas of high soil capability for alfalfa in which stands have declined for reasons other than salinity could support actively growing domestic grasses. Healthy stands dominated by POPR or BRIN could have a similar spectral signature to ALF as senescence is delayed. Additionally strong negative correlations between these species and PHW (r = -0.44) and ENa (r = -0.40) signify an important relationship between these domestic grasses and saline soils. The GRASS population was predominantly composed of Kentucky bluegrass and smooth bromegrass. Neither of these components was strongly correlated with spectral or soil parameters. An examination of the phenological state of the study site vegetation at the time of the photo imagery noted varying developmental status in the POPR and BRIN among Fields (Table 7). Vegetation consisted of greater proportions of BRIN and POPR as compared to ALF but the grasses were less dense and approaching senescence, possiblj' due to moisture stress or salinity stress. Watson and van Ryswyk (1986) noted range and domestic grass species exhibited different spectral signatures as they progressed through the phenological states. This variation results in the low correlations exhibited. Poa juncifolia was not strongly correlated with spectral or soil parameters. As it was rarely present except in areas of medium pH, ECe, ENa and low ECa, it is noted as an excellent indicator of a intermediate saline condition - possibly warranting further investigation. Artemisia frigida was distributed throughout much of the study area but was not strongly correlated with either spectral or soil parameters. As the alfalfa fields on the study site were formerly native grassland in which Artemesia frigida is strongly represented this population can be considered a remnant. Table 7 - Survey of vegetation on the five fields of the study site included in the remote sensing analysis Field 1 (PH8882) Field % Phenological Height Colour and Composition Stage (cm) Appearance Observations 1A ALF (30-70) POPR(15-25) BRIN(25-40) 1- 2 2- 3 2-3 30-55 30-55 20-40 green ALF more dense with elevation BRIN more mature with elevation in the SE corner 1B 1C 1D 1E ALF(20-50) POPR(10-60) BRIN(0-20) ALF (40-60) POPR(10-40) BRIN(20-40) ALF(20-40) POPR(10-20) BR IN (40-70) ALF-patchy (10-50) POPR -variable BRIN -variable 1- 2 2- 3 2-3 1- 2 2- 3 2-3 1 2-4 2-4 1- 4 2- 4 2-4 30-35 ALF pale green in some areas 40 - dark patches 40-60 patches 35 ALF - variable colour 40 BRIN-some pale green to 45 yellow 25-40 ALF -variable - light 20-40 and dark patches 15-40 tall patches -green Some Aqropyron riparium sodic patches HOJU, POJU noted HOJU, PUNU, POJU noted DIST, HOJU, PUNU, POJU -patches of SED & salt-FORB -standing water (<15 cm) -shorter Phenological stages 1- Vegetative 2- Boot or Bud 3- Flowering 4- Senescent Table 7 - cont'd Field 2 (PHB909) Field % Composition Phenological _ a Stage Height (cm) Colour and Appearance Observations < 2A 2B 2C ALF - 0 POPR -variable BRIN -variable ALF (20-60) POPR (10) BRIN (30) ALF (10-20) POPR (20-40) BRIN (10-30) 2-4 2-4 1- 2 2- 3 3 1-3 3- 4 3-4 15-30 15-30 30-60 20-30 30-50 10-40 < 20 20-30 GRASSES -pale green to green dark green medium to dark green light to dark green pale brown tips green to yellow HOJU, SED some A. elongatum  Trifolium sp. dense vegetation sedge up to 60% scattered Trifolium HOJU 2D ALF (20-60) 2-3 POPR (30-40) 3-4 BRIN (40-80) 3-4 2E ALF (10-50) 2-3 POPR (10-30) 2-4 BRIN (10-30) 2-4 2F ALF (30-50) 2-3 POPR (10-15) 2-4 BRIN (40-50) 2-4 25-35 20-40 20-50 30-60 10-50 10-50 30-50 20-40 30-50 dark green -some pale dark green to yellow generally dark green yellow-green to dark green -green to dark green -green green; heads brown to reddish brown small, pale leaves on alfalfa on west side some ALF-pale, brown crinkled at tips in NW portion -vegetation denser at bottom of slope-patchy at top vegetation denser at bottom of slope Phenological stages 1- Vegetation 2- Boot or Bud 3- Flowering 4- Senescent Table 7 - cont'd Field 3 (PH8883) Field % Phenological Height Colour and Composition Stage (cm) Appearance Observations 3A ALF (20-60) 2-3 POPR (10-40) 2-3 BRIN (10-40) 2-3 3B ALF (20-45) 2-3 POPR (10-50) 2-3 BRIN (10-35) 2-3 3C ALF (0-20) 2-4 POPR (30-60) 2-3 BRIN (10-35) 2-3 3D ALF (0-20) 2-4 POPR (15-80) 2-4 BRIN (10-30) 2-4 3E ALF (0-40) 2-4 POPR (15-90) 2-4 BRIN (15-30) 2-4 3F ALF (0-20) 2-4 POPR (10-40) 2-4 BRIN (10-30) 2-4 35 10-35 10-40 30-40 10-35 10-40 15-30 10-40 10-25 20-30 10-40 10-25 10-25 10-40 10-25 10-25 10-40 10-25 green good ALF survival green pale green to green green to dark green yellow/green to green pale green to green yellow brown to green yellow brown to green green pale green to green pale green to green yellow/green to green brown to green brown to green ALF less dense than 3A, POPR generally in head, less BRIN than ALF sodic patches ALF patchy green water-course vegetation bands of POPR with ALF mainly SALT vegetation-senescent yellow/brown green dark patches of ALF-BRIN associated with ALF dominated by bare soil; SALT vegetative DIST near pond PUNU on edge of bare soil road through center Phenological stages 1- Vegetation 2- Boot or Bud 3- Flowering 4- Senescent Table 7 - cont'd Field 4 (PH8883, PH8911) Field % Phenological Height Colour and Observations Composition Stage (cm) Appearance 4A ALF (5-50) 2-3 POPR (5-30) 2-4 BRIN (5-70) 2-4 4B ALF (5-80) 1-3 POPR (10-50) 2-4 BRIN (5-70) .2-4 4C ALF (0-50) 1-3 POPR (10-40) 2-4 BRIN (5-20) 2-4 4D ALF (10-50) 1-3 POPR (10-80) 2-4 BRIN (10-70) 2-4 4E ALF (0-40) 1-3 POPR (10-60) 2-4 BRIN (10-70) 2-4 4F ALF (0-40) 1-2 POPR 1-3 BRIN (20-60) 1-2 10-40 15-35 15-35 10-45 10-40 15-40 10-35 10-35 15-40 10-45 10-40 15-40 10-35 10-30 15-40 yellow/brown to green 10-35 10-40 10-30 yellow/green to green yellow/brown to green pale green to green yellow/brown to green green yellow/green to green green yellow/green to green yellow/brown to green pale green to green green scattered dense patches of ALF & BRIN some patches of dense BRIN some > 80% ALF vegetation sparse in some areas ALF very patchy sodic patches some dense patterns of ALF - areas of senescent POPR very saline (white) areas bordering pond -areas of SED near pond scattered ALF and DIST near pond on E side patches of ALF > 80%, some sodic areas -dense band of brome (> 80%) -areas of dense POPR < 10 cm Phenological stages 1- Vegetation 2- Boot or Bud 3- Flowering 4- Senescent Table 7 - cont'd Field 5 (PH8911) Field % Phenological Height Colour and a Composition Stage (cm) Appearance Observations 5A ALF (10-50) 1-3 POPR (10-60) 1-3 BRIN (10-60) 1-3 5B ALF (20-60) 1-3 POPR (10-40) 1-3 BRIN (10-45) 1-3 5C ALF (15-40) 1-3 POPR (30-70) 2-3 BRIN (5-20) 1-3 5D ALF (15-35) 1-3 POPR (10-40) 1-3 BRIN (20-50) 1-3 5E ALF (5-20) 1-4 POPR (10-40) 1-4 BRIN (15-30) 1-4 5F ALF (5-40) 1-3 POPR (10-40) 1-3 BRIN (10-40) 1-3 5G ALF (5-40) 1-3 POPR (10-70) 1-3 BRIN (10-60) 1-3 10-35 10-35 10-40 5-30 10-30 10-40 10-45 10-35 10-60 10-45 10-35 10-50 10-35 10-35 10-40 10-50 10-35 10-60 10-50 10-35 10-60 green to dark green red/brown to green yellow/green to green pale green to green Area of ALF > 75% FORB (goatsbeard) dense in a r e a s pale green to green green to dark green green green brown-green green pale green -green pale -dark green yellow green -green ALF patchy in m a n y a r e a s - areas of stunted plants patch of PUNU at S e n d scattered patches of dense ALF -dead ALF crowns - weak plants.with l a rge a r e a s of sodic soil large areas of PUNU dense bank of ALF-N s i d e of mainline areas of dense BRIN SALT in E end -dead ALF crowns weak plants Phenological stages 1 -Vegetation 2- Boot or Bud 3- Flowering 4- Senescent 69 The FORB and SEDGE parameters were both removed from the analysis at this stage as they exhibited no strong correlations with either spectral or soil parameters. Correlations are low (r < 0.10) as a result of the many zero or low value cases. However, in some instances FORB or SEDGE are present in sufficient population densities to influence the spectral signatures of ALF, ALF/GRASS or SALT. These potential conflicts in spectral signatures were considered in the subsequent analj'sis. 5.5 Morphological Classification The morphological analysis of the parameters details the nature of the relationship between parameters of the different categories (spectral-plant-soil). In addition the most appropriate spectral parameters for cluster analysis can be selected. These parameters will be related to the maximum number of plant and soil parameters. Parameters which are highly correlated provide essentially the same information. Maximum separation between groups can be expected when parameters with low correlations are used. In this way each clustering parameter contributes additional information for class separation. Functionally, morphological systems may be defined as isolated, closed or open (Chorley and Kennedy 1971). The system under investigation is an open system. Chorley and Kennedy (1971) stated, "in an open system many of the parameters can be viewed as responses or adjustments to flows of energy or mass through cascading (interconnected) systems". The total system under study is interconnected with other systems and can also be dis-assembled into component systems. This analysis will proceed by investigating the spectral, plant and soil systems separately and in combination. 5.51 Spectral System The morphological systems diagram for spectral parameters is presented in Figure 10. Y (Green) is highly correlated with M (Red) (r = 0.81) and less strongly with C(NIR) (r = 0.44). The strength of the relationship between Y and M (Green vs. Red) is maintained with green, actively growing vegetation as the % reflectance remains relatively constant in o the green and red wavelengths. As the r for this relationship is 0.81, there is still a significant source of unexplained variation. When vegetation is senescent or stressed, green reflectance decreases and the red increases. Murtha (1981), Watson and van Ryswyk (1986) and Tucker etal. (1980) reported similar inverse trends among these spectral parameters. The plot of Yellow vs. Magenta is presented in Figure 11. Points grouped about the means (Y = 124, M = 92) are representative of green, actively growing vegetation. Points outside this area may represent stressed or senescent vegetation. ( Y * M ) Morpho log ica l sys tems diagram for se lec ted spectra l parameters. Parameters were se lec ted on the basis of corre lat ions (> + 0.35). Values shown on in terconnect ing l ines are the r va lues relating parameters in the boxes. 72 +....+ ...+....+....+....+....+. 160 + +....+....+....+. .+X. . . + . 1 1 1 150 + 140 + 130 + 120 + 1 1 1 1 1 1 1 1 1 11 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 2 2 11 1 11111 11 2 111 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 . 1 1 10 + 1 1 1 2 1 1 Y 1 1 100 90. X + ...+ ...+....+....+. . + . . . . + . 45 . N = 84 R = .8058 P(R) 0.000 63. 81. 99. 117 135 153 54. 72. 90. 108 '126 144 162 Figure 11 - Plot'of digital numbers (DN's) for Y (yellow) vs. M (magenta) (n = 84). Yellow and magenta represent reflectance in the green and red regions of the spectrum respectively. The numbers 1 and 2 on the plots represent frequency counts for the respective DN's. 73 Thomson et_al. (1985) were able to detect invasion of rough fescue (Festuca scabrella) grasslands by Kentucky bluegrass and smooth brome because these species senesced 2 - 3 weeks earlier than rough fescue. Dalstead et_al. (1979) noted the senescence of salt-tolerant species to be an important discriminating factor in spectral differences. 9 The correlation coefficient r = 0.44 (r = 0.19) indicated considerable variation between Y (Green) and C (NIR). It has been previously reported (Fritz 1967) that considerable NIR variation can occur between different plant species and in stressed vs. non-stressed vegetation; this relationship (Y and C) was not, therefore, anticipated to show a high correlation. The relationship between M (RED) and C (NIR) showed an even lower correlation coefficient (r = -0.13). The negative sign of the coefficient suggested that decreased NIR reflectance was associated with increased R values. However, the wide variety of spectral signatures produced highly variable reflectance, particularly in C (NIR). This relationship makes these potentially useful clustering parameters in analyzing co-correlated plants or soil parameters. Figure 12 depicts a cumulative frequency distribution of C. As the data are skewed to the right of the graph, indicating a large proportion of the training sites exhibited high NIR digital numbers, an analysis was performed (Dixon 1985) to determine if data were normally distributed. Digital numbers from a normally distributed population will be on or near the straight line except for random fluctuations (Figure 13). Non-random deviations from the normal plot reflect lack of homoscedasticity and confirm the need for data transformations (Sokal and Rolf 1987). C (NIR) exhibited the narrowest distribution of the spectral parameters with DN's ranging from 127 - 191 (x = 165.9) within the overall scale of 0 - 255. C had the lowest % CV (7.9% vs 11.7% for Y, 27.6% for M). The C distribution indicated that the healthy, green vegetation at most of the case locations was exhibiting similar NIR spectral reflectance. However, examination of the cumulative frequency plot (Figure 12) reveals 74 COUNT MEAN ST.DEV. 84 165.881 13.15G 90 * 81 72 + 63 27 18 * * * * * * * * * _**.*•* O. + 132- 140 148 156 164 172 180 188 128 136 144 152 160 168 176 184 Figure 12 - Cumulative frequency plot for C (cyan) (n = 84). This plot represents the range of digital numbers in the NIR region of the spectrum. E a c h asterisk represents one observation. 75 SYMBOL COUNT MEAN ST.DEV. • 84 165.881 13.15G 2 .4 II-1.6 + ' / / // // . 80 + 0 . 0 + . 80 // // •// •// // /* II" II II' I I'-ll ' II' II*' I I" I* * * /* // // // - 2 . 4 +« 128 1 3 2 140 148 156 164 172 180 136 144 152 160 168 176 184 188 Figure 13 - Normal probability plot for C (cyan). Data from a normal distribution will be on or near the straight line except for .random fluctuations. Each asterisk represents one observation. 76 three separate areas of concentration. These data suggested that there were three areas of concentration, possible DNs within the group. Y (green) was the next most variable parameter (% CV = 11.5). The data was evenly distributed about the mean (124) with groupings in the upper and lower end of the plot indicating a variety of spectral signatures (Figure 14). This distribution represents the variety of phenological states and plant species encountered. The data for M (Red) (% CV = 27.3) showed the widest range of values indicating a diversity of spectral responses in the red region. The cumulative frequency plot of this spectral data revealed two areas of concentration; the majority of data points were grouped about the mean R = 91. A second group at R = 128 represented approximately 25% of the cases (Figure 15). The latter group of cases represents a distinct distribution of DNs -indicative of a separate plant/soil system. Any physiological change in vegetation such as senescence, disease, moisture or nutrient stress, which reduces photosynthetic activity, will result in increased red reflectance (Jackson 1986; Murtha 1981). Fritz (1967) noted that actual values of reflectance were often affected by the age of the leaf, season of year, water and mineral content of the soil or the type of soil. Such factors may, therefore, confound the interpretation of single waveband data. Products, ratios and indices have been used in soil and vegetation studies using Landsat and MSS data to interpret spectral signatures more clearly than possible using single wavelength data. Of the products evaluated, PR1 (Y*M) was retained in the analysis as it showed higher correlations with more plant and soil parameters than either PR2 or PR3. PR1 was highly correlated with Y and M (r = 0.88, r = 0.98) but was not correlated with C (r = 0.02). The two parameters (PR1 and C) may have potential as clustering parameters. In a further attempt to limit the number of parameters used in the analysis R^ (Y/M) was excluded in favour of its reciprocal R 4 (M/Y) (Figure 16). R^ was more highly correlated with spectral and soil parameters than R-^ . The reciprocal ratios R2 (Y/C) and Ro (C/Y) (Figure 17) were both selected on the strength of their correlations with salt 77 90 81 + 72 + 63 F 54 R E 0 U E 45 N C Y 36 27 + 18 + COUNT MEAN ST.DEV. 84 124.274 14.247 * * * * * * * * * * * _ * * * * * * * O. + 9 5 . 105 115 125 135 145 155 165 9 0 . 100 110 120 130 140 150 160 Figure 14 - Cumulative frequency plot for Y (yellow) (n = 84). This plots represents the range of digital numbers in the green region of the spectrum. Each asterisk represents one observation. 78 COUNT MEAN S T . D E V . 84 9 0 . 6 4 3 2 4 . 7 4 9 90 81 + 72 + 63 F 54 + R E 0 U E 45 + N C Y 36 18 * * * * * * * * * * * . * * * * * * * _ ***** + * 56 7 2 . 8 8 . 104 120 136 152 168 4 8 . 6 4 . 8 0 . 9 6 . 112 128 144 160 Figure 15 - Cumulative frequency plot for M (magenta) (n = 84). This plots represents the range of digital numbers in the red region of the spectrum. Each asterisk represents one observation. 79 . + . . . + . COUNT MEAN ST.DEV. 84 0.72) O.135 90 8 1 72 + 63 + F 54 + R E O U E 45 + N C Y 36 9 . + * * * * * * • + . • • • ' f - - . . + . . . + . . . . + . . . . + . . . . • • • . . . . + . . . . + . . . . + . . + . . . . + . . . . + . . . . + . . . , + . . . 4 4 0 . 5 2 0 . 6 0 0 . 6 8 0 .760 .840 ^920 1.00 . 4 8 0 .560 . 6 4 0 .720 . 8 0 0 .880 .960 1.04 F igure 16 - Cumulat ive f requency plot for ratio 4 ( R 4 ) (m/y) (n = 84). Th is plot shows the range of digital numbers for the red.green ratio. Each aster isk represents one observat ion. 80 COUNT MEAN S T . D E V . 84 0 . 5 5 2 0 . 1 6 8 . . . . + ....+ ....+ . . . . + . . . . + ....-* 90 72 + 63 F 54 + R E 0 U E 45 + N C Y 36 27 18 + .********** * * * * * * * . + + .. + ... . + . . . . + . ... + . . . . + . . . . + . . . . + , . . . + .. ..-*•.... + . . . . + . . . . + . ... + . . . . + .. . 2 5 0 . 3 5 0 .450 .550 .650 .750 .850 .950 1 .05 ' . 3 0 0 .400 .500 .600 .700 . 8 0 0 .900 1.00 Figure 17 - Cumulative frequency plot for ratio 3 (R 3) (c/y) (n = 84). This plot shows the range of digital numbers for the NIR.green ratio. Each asterisk represents one observation. 81 tolerant species and saline soil indicators. Both parameters were highly correlated with M and moderately with Y while showing little correlation with C. Combinations of C and R2 or C and R3 may prove to be useful clustering parameters. 5.52 Plant System The morphological system of plant parameters is shown in Figure 18. Positive correlations between species indicate these species occupy similar edaphic conditions. The strength of the correlation reflects the strength of the relationship. Negative correlations between species are indicative of an adaptation to different soil conditions. This procedure will identify the composition of the distinct vegetation communities on the study site. Correlations between plant species and the soil parameters will produce a preferred set of soil conditions for the various plant communities. ALF showed negative correlations with SALT (r = -0.38) and PUNU (r = -0.35). Correlations were low (r < 0.35) for the other plant parameters although the relationships were negative for all species but BRIN. Negative correlations with salt-tolerant indicator species (Hitchcock & Cronquist 1973) (HOJU, DIST, POJU, PUNU) implied that ALF does not prefer the saline conditions to which these species are adapted. The negative relationship of ALF and POPR may reflect a competitive interaction between these species. Kentucky bluegrass (POPR), as well as BRIN have been noted for their competitiveness with alfalfa in a mixed stand (Chamblee 1972). Alfalfa was generally in the bud or flowering stage at the time of analysis, but vigour and green colour varied, particularly in Field 4 (PH8883, PH8911) and adjacent to established saline areas where plants were often shorter and had chlorotic tissue. Tucker et al. (1980) found alfalfa exhibited different spectral signatures when moisture stress reduced leaf water potential and thus chlorophyll activity. They noted this effect would be more apparent spectrally in the red region than the NIR region. Bresler etjal. (1982) noted plants suffering from salinity stress initially appear moisture-stressed. A L F 0 .35 -0.35 S A L T -0.69 G R A S S Figure 18 - Morphological systems diagram for selected plant parameters. Parameters were selected on the basis of correlations (r > + 0.35). Values shown on inter-connecting lines are the r values relating parameters in the boxes. D I S T 0.43 0.36 SALT showed strong negative correlations with POPR (r = -0.49) and BRIN (r = -0.40) and very strong positive correlations with PUNU (r = 0.86), HOJU (r = 0.61), and DIST (r = 0.35). The spectral reflectance of the salt-tolerant species is clearly linked to their phenological state. PUNU and HOJU were senescent or nearly so. Senescent yellow vegetation was found to exhibit increased green and red reflectance over that of green vegetations by Watson and van Ryswyk (1986). Increased red reflectance results in a thinner M dye-layer; thus more green tones are admitted to the image. Decreased NIR reflectance was also noted with senescent yellow vegetation. This results in a thicker C dye-layer and more red will be subtracted from the final image. The resulting image produces a darker tone. Murtha (1981), Thomson et_al. (1985), Lo et al. (1986) and Brach and Mack (1977) were able to segregate senescent vegetation based on reflection differences. Murtha (1981) has also noted that senescent vegetation can exhibit increased NIR reflectance. Gausman (1974) proposed that such a relationship occurred when foliage was dehydrated. As the leaf dries, the number of air spaces increase. The most important refractive index discontinuity is the cell-wall/ air-space interface, with those between cellular constituents having less importance. Increasing the number of cell-wall/air-space discontinuities produces increased IR reflectance. DIST, a rhizomatous grass, exhibited a low, dense mat of vegetative growth, usually green with mid-brown flowerheads. POJU was a light-bluish green. These species will exhibit reflectances typical of green vegetation; moderately high Green reflectance, low Red reflectance and high NIR reflectance. These species will contribute slightly lighter magenta tones than PUNU and HOJU. As PUNU and HOJU exhibit stronger correlations to the SALT parameter than DIST and POJU the spectral signature of SALT will be dominated by PUNU and HOJU. Dense populations of DIST were only found near saline ponds, whereas HOJU and PUNU were more widely distributed. 84 GRASS is very strongly correlated with POPR and shows a low correlation with BRIN. An r = 0.62 indicates that POPR contributed more than 62% of the variation in GRASS. GRASS was not strongly correlated with A L F and the ALF/GRASS ratio thus, and varies widely over the study site. These data indicated that there was no inherent equilibrium between ALF and GRASS populations. Therefore, it might be anticipated that plant populations will vary subject to other soil and site factors. GRASS exhibited a strong negative correlation with PUNU (r = -0.52), similar to ALF's relationship to PUNU. This implies PUNU has a preference for different soil conditions than those preferred by ALF and GRASS. The phenological state of POPR varied considerably throughout the study site, while BRIN was not as variable. Chlorotic tissue m POPR -was noted at several locations. In Fields 1, 2, 3 and the western portion of 5, POPR was predominantly green and actively growing, whereas in Field 4, the eastern portion of Field 5 and adjacent to saline areas the POPR populations were a senescent yellow-brown; BRIN followed a similar trend. Both species also exhibited considerable height variation. Populations were noticeably shorter in Field 4, the eastern portion of Field 5 and bordering seepage areas compared to the remainder of the site. These areas may be stressed due to salinity, low fertility or soil moisture conditions. Thomson et _al. (1985) found soil moisture conditions resulted in earlier senescence and changing spectral signatures of timothy and smooth brome relative to rough fescue. ALF and SALT are the most important species in defining ALF capability classes of saline soil. PUNU and DIST have potential as indicator species of highly saline conditions. Spectral parameters which are strongly correlated with these plant parameters are the most suitable clustering parameters. The remaining plant species will be analyzed in context of the clustered groups to devise continuous classifications in response to varying degrees of soil salinity. 85 5.53 Soil System The soil morphological system is presented in Figure 19. The soil parameters selected for further analysis are primarily saline soil indicators. This indicates soil fertility and soil physical parameters are less important in determining ALF populations in this study site. Soil fertility parameters PHOS, TN and ORGC and physical parameters BD and Elevation appear to be of secondary importance as they exhibited fre correlations of r > 0.35 with other soil parameters. pHW, EC, ENA, ECA, EK and ENa appear to be the most important indicating parameters. pHW was strongly positively correlated with EC (r = 0.82), ENa (r = 0.94) and EK (r = 0.57) and showed a strong negative correlation with ECa (r = -0.58) (Figure 19). EC was also positively correlated with ENa (r = 0.94) and EK (r = 0.43). A negative trend similar to that of pHW was noted for comparisons involving ECa (r = -0.49); negative correlations were observed between ECa and ENa (r = -0.60), EK and EMg (r = 0.19) (Figure 19). In contrast, EK showed a strong positive correlation with ENa (r = 0.52). Bulk density was not correlated significantly with any parameter except EK (r = 0.43). Elevation showed no significant relationship with any of the soil parameters (r < 0.35). With the exception of EK, none of the soil fertility indicators were retained in the analysis based on their correlations with plant or spectral parameters. However, the parameters PHOS, TN and ORGC were significantly correlated with the saline soil indicators retained in the analysis. TN and ORGC showed moderately strong negative correlations with EC. ORGC and TN levels may serve as a measure of the organic matter content, of the soil. The low levels associated with high EC likely reflected historically lower contributions of detritus as a consequence of lower biomass production. Low EC was correlated with higher levels of ORGC and TN (r = -0.42). O R G C F i g u r e 19 - M o r p h o l o g i c a l s y s t e m s d i a g r a m f o r s e l e c t e d s o i l p a r a m e t e r s . P a r a m e t e r s w e r e s e l e c t e d o n t h e b a s i s of c o r r e l a t i o n s (r > + 0.35). V a l u e s s h o w n o n i n t e r - c o n n e c t i n g l i n e s a r e t h e r v a l u e s r e l a t i n g p a r a m e t e r s in t h e b o x e s . 87 PHOS was found to be positively correlated with PHW (r = 0.39). This relationship 9 i reflected the Ca domination of the study site soils; Olson's available phosphorus test was used in recognition of this relationship (Black 1965). In calcareous soils phosphorus is more available at pH > 7.5 (Bohn et al. 1979). 5.54 Spectral-Plant System In order to select the most appropriate spectral parameters for use in distinguishing the plant communities on this site, the spectral and plant systems were analyzed as a unit (Figure 20). Seven spectral parameters were selected for their high correlation with indicator plant species. ALF was strongly correlated with C(NIR) and R4 (M/Y). The positive correlation with C indicated increasing NIR reflectance with increasing ALF populations. Tucker et _al. (1980) found alfalfa plant height, canopy cover and biomass to be highly correlated with R and NIR. The negative correlation of ALF with R 4 indicated that an increase in the value of R 4 corresponds to decreasing ALF populations. An increasing R^ value resulted from decreased green and increased red reflectance, depicting altered spectral signatures indicative of stress, senescence or species differences. Crown (1979) found alfalfa could be visually distinguished on the basis of image tonal and textural differences from CIR air photos. SALT was inversely related to the C(NIR) parameter (r = -0.36) and positively related to R4 (r = 0.44) - exactly the opposite of the ALF response. In addition, SALT was also strongly correlated with R 2 (Y/C) (r = 0.42) and R3 (r = 0.45) (C/Y). These relationships supported the hypothesis that salt-tolerant vegetation can be discriminated from alfalfa on the basis of differences in spectral reflection. The SALT parameter was composed of four component species PUNU, DIST, HOJU AND PUNU. Although HOJU and POJU were eliminated as individuals from the analysis they make a collective contribution to the overall SALT parameter and serve as important indicator species. Figure 20 - Morphological systems diagram for the spectral-plant-soil system. The plant parameters are the key indicator species. The spectral and soil parameters chosen are those with the strongest correlations (r > + 0.35) to the plant parameters. Values shown on inter-connecting lines are the r values relating parameters in the boxes. 89 HOJU displayed digital numbers similar to that of senescent PUNU but the former species was not sufficiently represented to be strongly correlated with spectral and soil parameters. SALT was inversely correlated with ALF (r = -0.38) and showed a strong inverse relationship with GRASS (r = 0.64). DIST and PUNU did not exhibit a strong relationship with M (r = 0.44), r = 0.36) but showed a stronger correlation with the ratios R 2 (r = 6.52, r = 0.44), Rg (r = 0.46, r = 0.47) and R 4 (r = 0.37, r = 0.43). Rg was determined to be the most promising predictor of spectral properties for PUNU. R 2 , with a correlation of r = 0.52, was the best predictive parameter for DIST. 5.55 Plant - Soil System The plant and soil relationships pertinent to this study are those between the principal plant parameters (ALF, SALT, PUNU, DIST) and these parameters which describe saline soils (PHW, EC, ECa, ENa and EK). Although ALF was positively correlated with ECa (r = 0.38) it exhibited no other strong correlations with soil parameters. SALT shows very strong positive correlations with PHW (r = 0.64), EC (r = 0.51), ENa (r = 0.58) and EK (r = 0.49). Strong negative correlations were observed for SALT with ECa (r = -0.45) and the Ca:Na ratio. The salt-tolerant grass species DIST, HOJU and PUNU showed similar correlations with PHW, EC, ENa, EK and ECa, while PUNU was also negatively correlated with Ca:Na and positively correlated with PHOS (r = 0.32). Not unexpectedly, ALF and GRASS showed an inverse relationship with saline soil parameters, in contrast to the positive correlation of SALT. These data reflect a fundamental response in terms of adaptability to edaphic conditions. This response is the foundation for indicator species selection and use in remote sensing and the basis of the ability to discriminate ALF capability classes of saline-affected soils. 5.56 Spectral - Soil System The spectral-plant/plant-soil relationship trends observed were supported by subsequent evaluation of the spectral-soil relationships. Except in cases of extreme salinity bordering seepage ponds with visible white crusts or on some of the seriously 90 eroded "blowout" areas, the soil component made little direct contribution to the spectral signature. These exceptional areas were generally confined to field edges. PHW, EC, ENa and EK showed strong positive correlations with M (Red), PR-p R 2 , R3 and R 4 (r = 0.42 to 0.57). ECa was negatively correlated with the same spectral parameters (r = -0.35 to 0.38). Y is positively correlated with EC (r = 0.47) and ENa (r = 0.40). This apparent digression from expected trends is due in part to DIST. DIST vegetation is predominantly green, and exhibits very strong positive correlations with PHW, ENa and Y. C(NIR) reflection did not correlate as strongly with the saline soil indicators (r = -0.25 to 0.35), but discriminating trends did exist. Negative correlations with PHW, EC, ENa and EK and positive correlations with ECa followed an opposing trend of CIR reflectance to that observed for red wavelengths. Soils, high in PHW, EC, ENa and EK and low in ECa, will support vegetation which exhibited low CIR and high R reflectance. Conversely, soils low in PHW, EC, ENa and EK and high in ECa may be expected to support vegetation which exhibited high CIR reflectance and low R reflectance. The importance of the Y dye-layer is evident from its inclusion as a constituent of two of the relevant ratios. 5.57 Total Spectral - Plant - Soil System Following analysis of the complete morphological system a total of six spectral parameters, four plant parameters and seven soil parameters (n = 17) were selected for subsequent analysis based on their correlations with other parameters (Figure 20). Many of the parameters excluded from further analysis made contributions to the variability of other (retained) parameters, directly or indirectly. For example, the GRASS parameter was not included in the total system. Although it was not strongly correlated with any spectral parameters, GRASS exhibited strong negative correlations with PHW and ENa. Furthermore, it was strongly negatively correlated (r = -0.69; r = 0.48) with SALT 91 which suggested that high SALT populations will be associated with low GRASS populations. Conversely, parameters included in the analysis may have additional relationships with other parameters which would be useful in describing related groups. Y (Green) was not included in the analysis although its high correlations with all other spectral parameters will aid in distinguishing differences in green reflectance. As this study is attempting to define soil conditions based on the preferred edaphic conditions of indicator plant species the logical division of the total system was by species. Although alfalfa was the dominant species over much of the study site and the principal species of interest, it was strongly correlated with the fewest number of spectral and soil parameters (Figure 21). These observations would seem to limit the effectiveness of the analysis in partitioning study site soils into distinct alfalfa capability classes. However, as SALT was clearly shown to be inversely related to ALF, soil conditions which arefavourable for growth of SALT vegetation should be unfavourable for ALF. PUNU and DIST, the main component species of SALT in the analysis, may also help to separate classes of SALT. SALT was correlated positively with the spectral parameters M, R 2 , Rg, and R^ and negatively with C (See Figure 22). SALT was also positively correlated with PHW, EC, ENa, EK and BD and negatively correlated with ECa. ECa was the only parameter strongly correlated to both ALF and SALT in the analysis but ALF and SALT have inverse relationships with it. The important parameters retained in the analysis in describing categories of saline and non-saline soils and their capability for ALF growth and survival were PHW, EC, ECa, ENa and EK;secondary parameters identified were E and BD. Soil units exhibiting high capability for alfalfa production are best distinguished as having high Ca levels and being positioned at higher elevations relative to low capability soils (saline). These upper elevation soils were lighter and stonier, therefore + 0.39 +0.45 Morpho log ica l sys tems diagram for A LF (Medicago sativa). Spectra l and soi l parameters were se lec ted on the basis of corre lat ion to ALF (r > + 0.35). Va lues shown on inter-connect ing l ines are the r values relat ing parameters in the boxes. E N A Figure 22 - Morphological systems diagram for SALT (salt-tolerant grasses). Spectral and soil parameters were selected on the basis of correlation SALT (r > + 0.35). Values shown on inter-connecting lines are the r values relating parameters in the boxes. 94 well-drained (Saxton et_al. 1986). Soil units which exhibit low capability for alfalfa production are distinguished on the basis of the correlation between SALT and ALF. The LOW ALF soil units corresponded with high SALT. High SALT is characterized by high PHW, EC, ENa and EK and by low ECa. Medium capability soils correspond to soil conditions encompassing a mid-range between these extremes. This intermediate group would represent soil units affected by saline seepage activity and other soil problems. PUNU and DIST were used to distinguish between highly saline soils and the seepage areas of the MED ALF GROUP. DIST had a negative correlation with ECa while PUNU was not significantly related. Both were, however, strongly correlated with ENa indicating a preference of DIST for established saline areas with high ENa and low ECa and no significant relationship to EK, while PUNU favours high ENa and EK but has no significant relationship to ECa. PUNU was a better indicator of intermediate salinity composed of sodic soils and seepage areas. The large negative correlation with SALT indicates GRASS does not favour the high SALT (LOW ALF) soil conditions. POPR and BRIN favour different soil conditions. POPR has higher negative correlations with SALT, PUNU, DIST, PHW, EC, ENa and EK than BRIN. Therefore, higher populations of POPR may help distinguish between Low and Med groups. The low correlations between ECa and POPR indicated POPR does not have a strong preference for soil units high in ECa as does ALF. However, in the mixed stand studied here, these species are competitive with ALF (Chamblee 1972). Stresses which remove ALF favoured increased densities of POPR and BRIN. High populations of GRASS, therefore can also be expected in the HIGH ALF Class. 5.6 Cluster Analysis The Spectral-Plant-Soil Morphological System is composed of the 17 associated parameters which most strongly represent the system under study. This system, in conjunction with a number of related parameters, can explain the trends between different 95 vegetation composition on the site and the various soil microsites. Trends in the Yellow. Magenta and Cyan DN's were strongly correlated with plant communities indicative of varying degrees of soil salinity. Chorley and Kennedy (1971) state "subsequent analysis proceeds intuitively with limitations set by the investigator's insight into the system". In the case of this study, direction in the analysis is provided by the goal of the study. Detection and evaluation of saline seepage areas is important in the context of alfalfa capability of the study site soils. Field observations (Table 7, page 64) of vegetation and soil characteristics and densitometry studies (Section 5.2) provide supplementary information for interpreting of relationships between the vegetation and soil training site data and the photo image results. The analysis of the system performed to this point was extended by a cluster analysis of the correlation matrix. Chorley and Kennedy (1971) used cluster analysis to identify, describe and interpret their data. Spectral parameters, most strongly related to the indicator species, are used to cluster the training sites in such a way as to minimize the overall variation within groups (Patterson and Whitaker 1982). As each training site has spectral, plant and soil information the groups can be evaluated to determine if they are significantly different between their respective spectral, plant and soil parameters. Spectral signatures (DN's in the Y, M and C dye-forming layers) exhibited by the different vegetation classes can then be used in a supervised computer classification of the study site soils. Plant and soil data are used to define soil capability groups. Ground-truthing data are required in interpretation of spectral signatures (Tueller et _al. 1982). Plant survey data (Table 7, page 64; and training site data (Appendix 2) supplement interpretation of spectral data in this study. 5.61 Cluster Analysis of ALF - C and R 4 - Four Classes Defining the alfalfa relationship to saline soils is the primary aim of this study. It has been established that alfalfa is inversely related to saline soil indicator parameters and to 96 plant parameters SALT, DIST and PUNU. Remote sensing applications are possible as ALF, SALT, DIST and PUNU are strongly correlated with M, C, PR^ R 2 , R 3 , R 4 . ALF exhibits its strongest correlations with C and R 4 . As C and R 4 are not strongly correlated with one another they would both impart a significant amount of information in clustering the training sites. The analysis was first performed requesting four groups LOW ALF, MED ALF, HIGH ALF, VERY HIGH ALF. The result of the plot of R 4 vs. C is presented in Figure 23. Cluster separation for the four groups is within the two dimensional pixel value vector space. The center of each group represents mean pixel values and the diameter represents plus and minus one standard deviation (Schreier and Zheng 1986). Group membership is presented in Table 8. Table 9 summarizes results of mean separation tests. Of the spectral parameters only C was significantly different (P < 0.05) for all groups. Y, M and R 4 failed to discriminate between Medium (M) and Very High (VH); High (H) and Very High (VH). Although the distribution of the VHIGH ALF Class (Cases 12, 54, 58, 59, 65) was consistent with the trend of C, its scatter plot group is widely distributed by R 4 (Figure 13, page75). Four of these points (Cases 54, 58, 59, 65) were on the edge of photograph PH8883 and likely showed abnormally low values due to vignetting even though an anti-vignetting filter was used. The VHIGH class was not significantly different from any other class in terms of either plant or soil parameters. Plant parameters SALT, GRASS, POPR, BRIN, DIST, POJU and PUNU were not significantly different between any of the groups. ALF showed separation between LOW (L) and HIGH (H). HOJU showed separation between LOW and MED and LOW and HIGH. PHW and ECa showed separation between the LOW and MED, and LOW and HIGH groups. ENa was significantly different between LOW and HIGH groups. Clearly insufficient range of spectral tones existed for the discrimination of four separate groups. The LOW, MED and HIGH groups showed poor separation in most parameters. This analysis was useful as it identified a number of cases possibly affected H V V + . . . . + . . . . + . . . . + . . . . + + . . . . + . . . . + . . . . + . . . . .+ . . . . + . X . . + . . . . + . . . . + . . . . . 4 5 0 . 5 5 0 . 6 5 0 . 7 5 0 . 8 5 0 . 9 5 0 1 . 0 5 4 0 0 . 5 0 0 . 6 0 0 . 7 0 0 . 8 0 0 . 9 0 0 1 . 0 0 1. R4 ALF (Medicago sativa) classification in 2-D pixel value vector space - 4 classes. L (LOW ALF), M (MEDIUM ALF), H (HIGH ALF) and V (Very High ALF) classes were determined by cluster analysis with C and R 4 (m/y) as grouping variables. The center of each class represents the mean pixel value and the diameter represents plus and minus one standard deviation. 98 Table 8 - Cluster analysis for ALF. C and R4 were used as grouping variables giving four classes. Numbers within each group represent the training site numbers for each category. LOW ALF MED ALF HIGH ALF VHIGH ALF (n=12) (n=47) (n=20) (n=5) 1, 26, 31, 3, 8, 9, 12, 15, 6, 11, 16, 27, 54, 58, 49, 70, 79, 17, 22, 29, 33, 18, 21, 63, 59, 65 85, 92, 93, 34, 35, 36, 38, 72, 73, 76, 98, 99, 123 40, 41, 42, 43, 89, 100, 103, 44, 45, 46, 53, 114, 118, 127, 56, 57, 60, 62, 129, 133, 134, 67, 68, 80, 82, 137, 140 86, 87, 90, 95, 101, 102, 106, 109, 111, 112, 115, 121, 122, 125, 130, 135, 142, 145 99 Table 9 - Comparison o f f o u r c l a s s e s o f ALF grouped u s i n g C and R 4 f o r s p e c t r a l , p l a n t and s o i l p a r ameters. Mean s e p a r a t i o n was a c c o m p l i s h e d u s i n g a no n - p a r a m e t r i c t - t e s t (* = 0.05). Means w i t h i n rows f o l l o w e d by the same l e t t e r a re not s i g n i f i c a n t l y d i f f e r e n t . CLASS Parameters LOW ALF MED ALF HIGH ALF VHIGH ALF (n = 12) (n = 20) (n = 47) (n = 5) S p e c t r a l Y e l l o w x s %CV 140.3ace 17.4 12.9 123.0b 11.1 9.0 123.3d 7.2 5.8 100.2f 14.7 14.7 Magenta x s %CV 133.7ac 24.4 18.2 95.2be 10.6 11.1 78.7d 14.1 17.9 81.6f 28.8 35.3 Cyan x s %CV 155.5ac 8.2 5.3 162.5gi 4.6 2.8 173.6beh 8.0 4.6 131.6dfj 4.0 3.0 Ratio 4 (M/Y) x s %CV 0.95ac 0.07 7.4 0.77be 0.05 6.5 0.63df 0.07 11.1 0.80 0.17 21.3 P l a n t ALF x s %CV 2.4a 4.7 195.8 9.7c 12.2 125.8 23.0bd 18.9 82.1 13.8 17.8 128.9 BRIN x s %CV 6.6a 12.3 196.9 21.2b 17.2 81.1 16.7 12.3 73.6 12.6 11.4 90.5 S o i l PHW x s %CV 8.97ac 1.06 11.8 7.73b 0.70 9.1 7.70d 0.78 10.1 8.24 1.22 14.8 ECa x s %CV 5.43ac 1.65 30.4 8.21b 1.64 20.0 8.14d 2.09 25.7 8.13 3.46 42.6 100 by vignetting and grouped them as VHIGH ALF analysis. The analysis was re-run requesting three groups. 5.62 Cluster Analysis of ALF - C and R 4 - 3 Classes Group membership, as determined by the results of cluster analysis of C and R 4 of ALF, is presented in Table 10. Tests for significance of spectral, plant and soil parameters are presented in Table 11. The scatter plot (Figure 24) of C vs. R 4 indicates the 5 cases formerly of VHIGH ALF are distributed throughout the 3 groups; 59 and 65 in L (LOW), 27 in M (MED) and 54 and 58 in H (HIGH). The results in Table 11 were computed with these cases remaining in the analysis. 5.621 Spectral Parameters Green reflectance (Y) tended to decrease from LOW ALF to HIGH ALF. LOW and HIGH; MED and HIGH were significantly different (P < 0.05) however, indicating good separation by the clustering technique. All groups had low %CV (12.5%, 7.6%, 11.0%). However, the narrow distribution of the Y parameters data points prevented all groups from being significantly different. The LOW and MED groups were not significantly different (p < 0.05). Within group variation is low (% CV _< 12.0). The LOW group is composed primarily of saline soil units. These groups exhibited the highest Y DN's. Watson and van Ryswyk (1986) found senescent, yellow and white vegetation to be highly reflective of radiation in the green band. The vegetation composition in these units is dominated by salt-tolerant vegetation such as HOJU, PUNU and DIST which are largely senescent (with the exception of DIST bordering saline ponds). Hall et_al. (1983) found a trend of increased dye-layer densities in beetle-attacked Douglas-fir (Pseudotsuga menziesii) Mirbl. Franco); the spectral data for stressed and senescent vegetation in this study was consistent with these observations. Thomson et al. (1985) found senescent vegetation could be successfully discriminated. As the MED group was not significantly different this must indicate a gradation of the vegetation composition to Table 10- Cluster analysis for ALF. C and R4 were used as grouping variables giving three classes. Numbers within each group represent the training sites for each category. LOW ALF (n=14) 1, 6, 26, 31, 49, 59, 65, 70, 79, 85, 92, 93, 98, 99 MED ALF (n=36) 3, 11, 15, 16, 17, 18, 21, 27, 33, 35, 36, 40, 42, 44, 63, 72, 73, 76, 89, 95, 100, 103, 106, 112, 114, 118, 123, 125, 127, 129, 133, 134, 135, 137, 140, 145 HIGH ALF (n=34) 8, 9, 12, 22, 29, 34, 38, 41, 43, 45, 46, 53, 54, 56, 57, 58, 60, 62, 65, 68, 80, 82, 86, 87, 90, 101, 102, 109, 111, 115, 121, 122, 130, 142 102 Table 11 - Comparison of three classes grouped using C and R4 for spectral, plant and soil parameters. Mean separation was accomplished using a non-para-metric t test (0(= 0.05). Means within rows followed by the same letter are not significantly different. CLASS Parameters LOW ALF (n = 14) MED ALF (n = 36) HIGH ALF (n = 34) Spectral y x s %CV 138.1a 17.3 12.5 125.5a 9.5 7.6 117.3b 12.8 11.0 m R4 x s %CV x s %CV x s %CV 131.9 93.6 22.1 8.7 16.7 9.3 All groups significantly different 153.3a 12.2 7.9 167.5bc 11.9 7.1 0.95 0.75 0.06 0.05 6.0 7.2 All groups significantly different 70, 11. 15.8 169.4c 12.0 7.1 60 05 5 Plant ALF x s %CV SALT x s %CV GRASS x s %CV POPR x s %CV BRIN x s %CV 1.1a 1.2 112.7 34.2a 28.6 83.7 38.6a 28.7 74.8 21.4a 21.6 100.9 7.4a 12.0 161.1 18.8bc 14.7 78.0 7.1bc 12.9 181.0 59.9bc 22.9 38.3 39.1b 22.2 56.9 18.1bc 14.9 82.0 21.9c 19.9 90.8 7.7c 17.2 223.0 58.3c 23.8 40.8 38.6a 25, 65, 17.4c 12.9 74.0 Table 11 (Continued) CLASS Parameters LOW ALF MED ALF HIGH ALF (n = 14) (n = 36) (n = 34) HOJU x s %CV 10.3a 8.7 85.0 PUNU x s %CV 18.6a 20.5 110.2 Soil PHW x s %CV 9.14a 0.92 10.0 EC x s %CV 37a 23 89.7 ECA "x s %CV 5. 1. 33. 38a 79 3 ENA x s %CV 9.21a 6.75 73.3 EK x s %CV 1.49a 0.72 48.3 1.3bc 4.1 323.0 2.1c 6.5 317.0 2.8b 7.2 257.0 3.7a 9.0 243.0 7.65bc 0.71 9.3 7.71c 0.75 9.7 0.29bc 0.26 90.7 0.32c 0.35 109.9 8.48bc 1.70 20.0 8.01c 2.22 27.8 1.12bc 2.00 179.3 1.96c 3.42 174.2 0.78bc 0.30 38.3 0.810c 0.39 248.0 M H L L M + . . . . + . . . . + . . . . + . . . . + . . . . + . . . . + . . . . + . . . . + . . . . + . . . . + . X . . + . . . . + . . . . + . . . .+ .450 .550 .650 .750 .850 .950 1 .05 400 .500 .600 .700 .800 .900 1.OO 1.1 R4 ALF (Medicago sativa) c lass i f icat ion in 2-D pixel value vector space - 3 c lasses . L (LOW ALF), M (MEDIUM ALF) and H (HIGH ALF) c lasses were determined by c luster analysis with C and R 4 (m/y) as group ing variables. The center of each c lass represents the mean pixel value and the d iameter represents p lus and minus one standard deviat ion. 105 encompass more actively photosynthesizing site vegetation. The MED group was significantly different from HIGH (P < 0.05). Magenta was significantly different (P < 0.05) for all three groups with the LOW group exhibiting the highest red and the HIGH group the lowest red. The variability of the clustered groups was low and group separation was good as evidenced by the low %CV (L = 16.7, M = 9.3, H = 15.8). These results indicated that the High ALF group, mainly composed of ALF and GRASS, was actively absorbing red wavelengths while red is being reflected from senescent and stressed vegetation in the LOW ALF group. This trend has been reported by many workers (Tucker et al 1980; Murtha 1981; Hall et_al. 1983; Thomson et_al. 1985; Watson and Van Ryswyk 1986). C(NIR) was not significantly different between the MED and HIGH groups indicating a reverse trend to that of Y. Although these groups were not significantly different for the C parameter they exhibited the lowest within group variation (% CV _< 10.0) (See Table 11). The data derived from the Optronics scanner is based on a scale of 0 (black) - 255 (white). M (Red) has the widest range (45 - 168 = 123) followed by Yellow (Green) (86 - 165 = 79) and C (NIR) (127 - 191 = 64). DN's exhibited by the cases on the study site do not optimally use the full range of the Optronics data. Despite the excellent clustering of the groups, as indicated by the low coefficients of variation, the narrower distribution of the Y and C data results in groups which are not significantly different (p < 0.05). This problem was addressed by the addition of R 4 (M/Y) to the analysis. Clustering by R 4 and C included all the dye-layers for maximum group separation. R 4 was significantly different (p < 0.05) for all three groups as were R 2 and R3. Variation for all R 4 analyses was low (CV < 8%). In the two-dimensional diagram of the mean +_ one standard deviation presented in Figure 24, the three groups show good separation although the C means of HIGH and MED are not significantly different (p < 0.05). Cases 15, 17, 42 and 44 exhibited HIGH C values but were retained in the MED group on the basis of their R 4 values (See 106 Appendix 2). HIGH and MED group separation would likely be improved by the re-classification of these units. 5.622 Plant Parameters ALF, SALT, GRASS and HOJU were significantly different (p < 0.05) between all groups except MED and HIGH. BRIN and PUNU were significantly different between LOW and MED. DIST and HOJU were not significantly different between any of the groups. All groups exhibited linear trends. Salt tolerant species SALT, PUNU, DIST and HOJU exhibited higher % canopy cover in the LOAV ALF class and lower % canopy cover in the HIGH ALF class. The reverse of this trend was noted in ALF, BRIN and POPR. Species such as DIST, HOJU and PUNU were not represented at many case locations, so many zeros are recorded. This resulted in large standard deviations and coefficients of variation. An examination of the ranges for each group provided a clearer picture. For example DIST means are 39-24-5 for the LOW-MED-HIGH ALF groups. PUNU means are 67-30-33 for L-M-H, and HOJU means are 26-23-29. Distributions of DIST and PUNU indicate higher concentrations in the LOW ALF class with decreasing levels in the MED and HIGH classes. PUNU has higher concentrations in all classes. DIST appears to be poorly adapted to the HIGH ALF soils, while HOJU is more generally distributed. HOJU was noted by Best et_al. (1978) to inhabit soils ranging from marshy to arid conditions, although it is generally described as halophytic. Some potentially mis-classified cases were also identified.' The HIGH ALF class encompasses soils with low pH, EC, ENa and EK; high ECa. Cases 12, 29, 122 and 130 were found to have soil and vegetation characteristics indicative of saline soils yet on the basis of C and R 4 reflectance, they were placed in the HIGH ALF class. These cases will be further examined in Section 5.63. BRIN exhibited a trend toward equal populations, about 0 - 50% in groups M and H, and 0 - 3% in L. These data indicated a decreased preference for saline areas and a 107 general adaptation of BRIN to other conditions. POPR was evenly distributed over the study site with maximum populations of 69-73-85 % canopy cover in the LOW, MED and HIGH classes respectively. Means were more equal, though they tended to be lower in the LOW ALF group. ALF ranges (0-3, 0-56 and 0-60) (Table 11) and means of the LOW, MED and HIGH groups indicate a linear trend; standard deviations and %CV are dispersed. As ALF populations varied across this study site in response to factors other than salinity, these variations were not unexpected. However, LOW ALF (high saline) soils were successfully delineated from MED and HIGH, while the latter groups were not significantly different from one another. These results indicated the significantly different DN's exhibited by ALF, especially in terms of R^ and M (Table 11), may cause spectral separation not accounted for by population differences. ALF was in some cases (eg. Field 4) stunted and chlorotic in appearance. Such plants would exhibit higher R DN's. Tucker et_al. (1980) found moisture-stressed ALF with a similar appearance exhibited higher R- reflectance. In addition, the maximum ALF population at any case was 60%. The population also included components of GRASS and SALT which exhibited similar statistical trends in terms of SD and %CV. By nature of these classifications all species are represented in the three classes. The DN's successfully determine marked linear trends in the populations of all the indicator species. All plant parameters were significantly different (p < 0.05) between the LOW and MED and LOW and HIGH classes. MED and HIGH classes were not significantly different (p < 0.05) between any plant parameters. 5.623 Soil Parameters PHW, EC, ECa, ENa and EK were significantly different (P < 0.05) for all groups but MED and HIGH ALF (Table 12). PHW showed the least variation within groups with all CV's < 10%. ECa was the next best discriminating parameter with a % CV of 20 - 34%. EK (%CV = 40 - 50 %) was substantially more variable than PHW or ECa. EC and ENa were more widely distributed. 108 PHW, EC and ENa are indicative of sodic rather than saline or saline-sodic soils. Sodic soils commonly have high pH values and high levels of ENa (Bohn et_al. 1979). EC values are < 4 mmho cm" .^ PHW and ECa are the best discriminating soil parameters as they exhibit the lowest within group variation. These parameters best define classes of soil capability which are in turn defined by the composition of vegetation communities they support. EC and ENa were also useful discriminating parameters, but in this study the range of values encountered in each class limit were wider than for PHW and ECa. EK appears to behave similarly to ENa in study site soils. The highest EK levels were found in the saline LOW ALF unit. HIGH EK may act in tandem with ENa in restricting growth of ALF, GRASS, BRIN and POPR. There is evidence (See Section 5.32) that low EK levels such as those found in Field 5 limit ALF winter survival. ECa was the most important cation in determining soil capability for alfalfa on this study site. ECa in the HIGH group showed more variability (%CV = 27.8) than MED (%CV = 20.0) which contributed to the lack of significance observed between the MED and HIGH classes. ECa was the only soil parameter with which ALF was strongly correlated. At the lower ECa levels in the LOW ALF group (ECa = 5.38 c moles +-kg_1) ALF is virtually excluded (ALF = 1.1%) while high ECa levels are found in the HIGH ALF class. Salinity effects may result from an interaction of ECa and ENa. Low levels of ECa, in conjunction with high ENa may characterize soils with low capability for alfalfa production. PHW was the least variable parameter within groups. The LOW ALF groups mean pH (9.14) was significantly (P < 0.05) higher than the MED (7.65) and HIGH (7.71) groups. The MED groups maximum was 9.7 while the maximum for the HIGH group was 9.97. The MED group represents a transition in soil types between the LOW and HIGH groups. Saline units which exhibit vegetation intermediate between the two extreme classes were expected in this group. Nevertheless, inspection of the data revealed that Cases 12, 29, 122, 130 exhibited the high pH values more typical of the LOW group. 109 5.63 Cluster Analysis of A L F - C and R 4 - 3 Classes: Second Analysis The above-mentioned cases bordered saline areas in the field and were found to include vegetation indicative of the L O W and H I G H classes. The 5 X 5 (2.5 m X 2.5 m = 6.025 2 m ) pixel matrix reflectance value (DN) is an average of reflectance for these 25 0.5m X 0.5m pixels. In this situation the reflectance more strongly represents the adjacent H I G H A L F unit. The green SED (Carex spp.) in these cases may also contribute to NIR reflectance. These four mis-classified cases represent 11.8% of the cases in the H I G H group of 4.8% overall. This frequency of mis-classification is consistent with the confidence level of the mean separation tests (p < 0.05). Thomson et_al. (1985) discriminated vegetation classes successfully at the 95% confidence level using Thematic Mapper data. Table 12 presents the recalculated means, standard deviations and % C V with four cases (12, 29, 122, 130) removed from the H I G H group; Case 80 which exhibited saline soil properties and salt-tolerant vegetation although the pH was < 9, was removed from the M E D class. Case 134 was also removed as it contained several missing plant parameters. Cases 15, 17, 42, 44, were classified in the M E D group, although their C(NIR) D N was more typical of the H I G H group. The brighter Y DN's (x = 130) of these cases was more suggestive of the senescent vegetation in the L O W group. Al l cases had high G R A S S components (43, 65, 73, and 46%) so the high C D N was most probably due to dry, senescent grass foliage. Thomson et al. (1985) and Watson and Van Ryswyk (1986) reported similar trends in Y reflection from dry, senescent foliage. These units were grouped in the M E D group on the strength of their R 4 values. S A L T was the only parameter which was significantly different (p < 0.05) following the second analysis between all alfalfa capability classes. Separation between classes was improved, in that the means of the groups were more widely separated and the % C V was reduced in the M E D and H I G H groups for A L F , GRASS, POPR, BRIN, P H W , E C , ECa , 110 Table 12 - Computer re-classification of ALF. The three categories of alfalfa were re-classified following removal of some anomalous cases. Mean separation was accomplished using a non-parametric t test (ex. = 0.05). Means within rows followed by the same letter are not significantly different. CLASS Parameters LOW ALF MED ALF HIGH ALF (n = 14) (n = 35) (n = 29) Spectral y x 138.1a 125.5a 117.1b s 17.3 9.5 13.8 %CV 12.5 7.6 11.7 m x 131.9 93.6 70.0 s 22.1 8.8 11.9 %CV 16.7 9.4 17.0 All groups significantly different c x 153.3a 167.9bc 169.1c s 12.2 11.9 12.9 %CV 7.9 7.1 12.0 R4 x 0.95 0.75 0.60 s 0.06 0.06 0.05 %CV 6.0 8.0 12.0 All groups significantly different Plant ALF x 1.1a 17.5bc 24.6c s 1.2 15.7 20.2 %CV 112.7 89.7 82.1 SALT x 34.2 7.3 0.90 s 28.6 13.0 2.38 %CV 88.9 177.8 265.4 All groups significantly different GRASS x 38.6a 61.6bc 66.2c s 28.7 21.1 22.1 %CV 74.8 33.4 35.0 POPR x 21.4a 40.5bc 42.4c s 21.6 21.9 25.3 %CV 100.9 54.1 59.5 Table 12 (continued) CLASS Parameters LOW ALF MED ALF HIGH ALF (n = 14) (n = 35) (n = 29) Spectral BRIN x 7.4a 18.7bc 19.2c s 12.0 14.8 13.0 %CV 161.1 79.1 67.7 HOJU x 10.3a 1.3bc 0.07c s 8.7 4.2 0.37 %CV 85.0 318.0 537.7 PUNU x 18.6a 2.9bc 0.6c s 20.5 7.3 2.1 %CV 110.2 252.9 386.0 Soil PHW x 9.14a 7.66bc 7.47c s 0.92 0.72 0.41 %CV 10.0 9.3 5.4 EC x 1.37a 0.29bc 0.20c s 1.23 0.26 0.11 %CV 89.7 90.1 56.0 ECa x 5.38a 8.46bc 8.36c s 1.79 1.72 1.91 %CV 33.3 20.3 22.9 ENA x 9.21a 1.12bc 0.99c s 6.75 2.03 1.92 %CV 73.3 180.9 193.0 EK x 1.49c 0.78bc 0.73c s 0.72 0.30 0.19 %CV 48.3 38.5 26.0 112 and EK. Group means and coefficients of variation more clearly depicted distribution trends in DIST, PUNU and ENa than in the first C and R 4 classification. Removal of the mis-classified units (5/83 = 6.02%) improved the numerical trends between the three groups and resulted in significant differences between groups based on the SALT parameter. Although the statistical separation between groups of the other key soil and plant parameters was not improved, this exercise did provide more distinct trends between the three classes and greatly aids the establishment of a continuous classification of parameters. The spectral signatures remained unchanged as a result of removing mis-classified cases. The successful identification of wrongly classified cases, and their removal, reinforces the necessity of ground truthing in digital CIR air photo analysis. Ustin et al. (1986) found spectral properties of semi-arid shrub communities changed as a result of physiological and phenological changes. Ground truthing was necessary to determine the timing of these changes and relate them to sequential Thematic Mapper scenes during the course of the growing season. Tueller (1982) also noted the importance of ground truthing data in interpretation of MSS data of rangeland scenes. 5.7 Spectral Signatures Tueller (1982) defined a spectral signature as combinations of reflectance levels from different spectral regions. In this study each class has a spectral signature composed of three data-numbers representing reflectance in the green, red and nearinfrared regions. The spectral signatures for a 3-group classification (LOW, MED, HIGH) of saline soil capability classes are presented in Table 13. Parameters included were Y, M and C. Y was included as the MED and HIGH groups were significantly different (p < 0.05) whereas they were not for C. C complements Y in that it separated the LOW and MED classes where the Y parameter was not different. M and R 4 (M/Y) were significantly different for all groups. Table 13 - Digital numbers in the three film dye - layers (y, m, c) for the three ALF classes. Range = 0-225 where 0 = black and 255 = White. CLASS  Parameter LOW ALF MED ALF HIGH ALF y 138 126 117 m 132 94 70 c 153 167 169 114 5.8 Continuous Soil and Vegetation Classes Continuous classes of soil and plant parameters were devised from the second classification and shown as histograms of parameter values vs. Class Frequency . Only those parameters which showed the best separation have been graphed. Means of classes generated by the cluster analysis are not always statistically significant. LOW ALF and HIGH ALF classes are different (P < 0.05) but MED ALF may not be different from either LOW ALF or HIGH ALF. This may reflect the nature of the MED class as a transition grouping between the LOW and HIGH classes. This argument does not suggest that all the MED units (45%) are necessarily saline seepage areas. Rather it indicates that this class should be examined further as it contains units with a lower capability for alfalfa production, possibly due to soil salinity. 115 5.81 P lant Parameters His tograms representing the mean and +_ 1 standard deviation of A L F , S A L T and G R A S S population densities (Figures 25, 26, 27) visual ly depict the distribution of A L F , S A L T and G R A S S across the site. These parameters produced the best defined indicators in terms of group separation and wi th in class var iabi l i ty . Separations between classes noted in the text refer to percentage differences between the histograms of _+ 1SD for each class. Th is technique recognizes that considerable variation exists in the distribution of individuals of a species so it concentrates on the population trend of the species ( A L F etc.) and species group parameters ( S A L T , G R A S S ) . A L F shows distinction between H and M (> 32%) and between M E D and L O W and H I G H (2.3 - 4.5%) (Figure 25). The wide variat ion in the H I G H A L F class is part ly due to low A L F populations caused somewhat by salinity and by other stress factors such as winterki l l , fertility or soil moisture conditions. The L O W A L F ( H I G H S A L T ) unit in S A L T shows a wide variat ion in trends. L O W S A L T values in soil units are due to the existence of A L F , G R A S S , F O R B or S E D within the population. This would support the successional nature of S A L T species intrusion into soil units as P H W , E C , E N a and E K rise while E C a is lower. The G R A S S parameter showed separation between the L O W and M E D , L O W and H I G H groups. The L O W C L A S S range was actually 0-92 which indicates some edge pixel contribution from M or H groups. However , the _+ 1SD range (10-60%) indicates higher tolerance of G R A S S than A L F to saline soils. The M E D and H I G H ranges are v i r tua l ly identical indicating G R A S S adaptation to a wider variety of soil conditions. This would explain the tendency of G R A S S to outcompete A L F in mixed stands (Chamblee 1972). 5.82 Soil Parameters Figures 28-32 depict histograms for soil parameters P H W , E C , E C a , E N a and E K . Separation noted is by unit of measurement for the particular parameter. P H W shows the best group delineation in terms of soil or plant parameters in that it had the narrowest S D Figure 25 - Continuous classification - ALF. Population trends of three classes -LOW (LOW ALF), MEDIUM (MEDIUM ALF) and HIGH (HIGH ALF) depicted as histograms showing the class value and plus and minus one standard deviation. Class Frequency ( % ) 2.4 7.3 34.2 60.0 Canopy Cover ( % ) Figure 26 - Continuous Classification - SALT. Population trends of three classes -LOW ALF (High SALT), MEDIUM ALF (Medium SALT) and HIGH ALF (LOW SALT) depicted as histograms showing the mean class value and plus and minus one standard deviation. Class Frequency ( £ ) 36.6 61.9 63 2 100 Canopy Cover ( X ) Figure 27 - Cont inuous Classification - G R A S S . Population trends of three c lasses - LOW ALF, MEDIUM A L F and HIGH A L F are dep icted as histograms showing the mean class value and plus and minus one standard deviation. Figure 28 - Cont inuous Classification - PHW. Distribution of PHW values of three c lasses - LOW ALF, MEDIUM A L F and HIGH A L F are dep ic ted as histograms showing the mean class value and plus and minus one standard deviation. and %CV of all the parameters (Figure 28). Although some overlap occurs with LOW (8.2-8.3) and HIGH (7.1-7.9) there is a distinct area of MED (7.9-8.2) where PHW could signify a critical region of transition between soils with HIGH capability for alfalfa production and soils with LOW ALF capability due to sodic soils. EC (Figure 29) has narrow ranges in the HIGH and MED classes. The + S D of the HIGH class is 0.31 mmho cm"1. This overlaps with the LOW GROUP (0.2-0.3l)mmho cm"1. The HIGH ALF group is excluded pastasi EC of about 0.31 mmho cm"1 though the class max was 0.65. The LOW group also overlaps the MED group (0.2-0.6 mmho cm"1). The overlap of the continuous classes indicates the narrow distribution of EC values on the study site. These values differ from relative plant growth data of Bresler et _al. (1982). However, this data was obtained from uniformly salinized field plots having nearly constant salinity with depth. Bower (1968) has shown EC to increase with sampling depth. Values point to a transition range but the wide variation in EC of the LOW group indicates EC does not distinguish between groups as well as PHW. LOW EC values are characteristic of sodic soils (Bresler et_al. 1982; Bohn et_al. 1979). ECa exhibits strong separation between the LOW and HIGH groups (Figure 30) and the LOW and MED groups. MED and HIGH have similar distributions. A threshold value of approximately 7.5 cmoles + kg"1 marks the lower limits of the MED and HIGH classes. This may represent a minimum amount of ECa which must be exceeded for HIGH alfalfa soil capability. ENa exhibits wide variation in the LOW group but a very distinct barrier (2.9-3.1 cmoles +• kg"1) separates MED and HIGH groups (Figure 31). Both the MED and HIGH groups max values were 8.0-8.2 c moles + kg' 1. This is evidence of ALF surviving at higher ENa levels but the low mean indicates generally low ENa levels in HIGH ALF capability soil units. EK was significantly different (p < 0.05) between LOW and. MED, and LOW and 121 Class Frequency (%) 100 l 0.20 0.29 1-37 4.00 Electrical Conductivity (mmhos-cm~ 1 ) Figure 29 - Cont inuous Classification - E C . Distribution of electrical conductivity values (EC ) of three c lasses - LOW ALF, MEDIUM A L F and HIGH ALF are depicted as histograms showing the mean class value and plus and minus one standard deviation. C l a s s Frequency ( £ ) I C C n 37 IS — 0 . 0 L O W MEDIUM HIGH 5 3 6 6.36 6.46 Exchangeable Ca (cmoles + - kg _ 1 ) Figure 30 - Continuous Classification - ECa. Distribution of exchangeable calcium (ECa) values of three classes - LOW ALF, MEDIUM ALF and HIGH ALF depicted as histograms showing the mean class value and plus a n d minus one standard deviation. Class Frequency {%} 100 i 0.99 1 12 9.20 16.00 Exchangeable Na (c• moles + kg~l) Figure 31 - Cont inuous Classification - ENa. Distribution of exchangeable sodium (ENa) values of three c lasses - LOW ALF, MEDIUM A L F and HIGH ALF depicted as histograms showing the mean class value and plus and minus one standard deviation. 124 HIGH groups but was not significantly different between MED and HIGH groups (Figure 32). Higher levels seem to exclude ALF (LOW ALF class). In this way EK behaves similar to ENa. There is strong evidence based on Site History and the photo comparison (Sec. 5.3) linking A L F winterkill with low EK levels. Photo 8911 (Fields 4 and 5) had the lowest EK levels (Table 4, page ) and the highest alfalfa winterkill (See Table 7). EK levels in PH8911 were 0.70 cmoles -r-kg"1 while those of the HIGH ALF were 0.73 cmoles -t-kg"1. These numbers cannot be taken as absolutes but they do point to an important threshold which must exist in EK values. 5.9 Cluster Analysis of SALT - C and R3 - 3 Classes SALT exhibited strong correlations with C, R3 and R 4 as well as the saline soil indicator properties (see Figure 33). Inverse relationships between SALT and ALF have been demonstrated so clustering on spectral reflectance indicative of SALT may designate soils with LOW ALF capability. An.alternative classification utilizing C and Rg as clustering parameters was performed to determine if this procedure produced improved classification over the C and R 4 vs. ALF classification. Classes retain the same names; LOW ALF (HIGH SALT), MED ALF (MED SALT) and HIGH ALF (LOW SALT). Results of cluster grouping are presented in Table 14. C vs. Rg is depicted in Figure 33. The clustered groups are LOW ALF (9/83 = 10.0%) MED ALF (24/83 = 29.0%) AND HIGH ALF (51/83 = 61%). These groups differed from the C and R 4 classification. The LOW and MED groups are smaller and the HIGH group has now become the largest. These data imply that 61% of the sampled cases are highly capable for ALF production. Y was not included in this clustering due to its low correlation with SALT. This may represent a loss of information. C exhibited the highest correlation of the single band parameters. Rg exhibited the highest correlation of the products and ratios so it was also selected as a grouping parameter. Rg (M/C) does not include Y as did R 4 (M/Y). Table 14 - Cluster analysis for SALT. C and R3 were used as grouping variables giving three classes. Numbers within each group give the training sites for each category. LOW ALF MED ALF HIGH ALF (n=9) (n=51) (n=24) 1, 26, 31, 49, 70, 85, 92, 98, 99 3 , 8, 9, 12, 15, 6, 11, 16, 21, 27, 17, 18, 22, 29, 54, 57, 59, 63, 33, 34, 35, 36, 65, 72, 76, 79, 38, 40, 41, 42, 89, 93, 100, 103, 43, 44, 45, 46, 114, 123, 127, 53, 56, 57, 60, 129, 134, 137, 62, 67, 68, 73 140 80, 82, 86, 87, 90, 95, 101, 102, 106, 109, 111, 112, 115, 118, 121, 122, 125, 130, 133, 135, 142, 145 Figure 32 - Cont inuous Classification - EK. Distribution of exchangeable potassium (EK). Distribution of exchangeable potassium (EK) values of three c lasses - LOW ALF, MEDIUM A L F and HIGH A L F depicted as histograms showing the mean class value and plus and minus one standard deviation. 127 Figure 33 - SALT (Salt-tolerant grasses) classification in 2-D pixel value vector space - 3 classes L (LOW ALF-HIGH SALT), M (MEDIUM ALF -MEDIUM SALT) and H (HIGH ALF - LOW SALT) classes were determined by cluster analysis of C and R 3 (c/y). The center of each class represents the mean pixel value and the diameter represents plus and minus one standard deviation. 120 •5.91 Spectral Parameters Y and M show similar trends as in the C and vs. A L F classification (3 Classes) (Table 15). However the C parameter successfully delineates only the HIGH group separately from LOW and MED. The SD and % CV are low for the groups but the two-dimensional clustering with Rg has resulted in similar C values for the LOW and MED groups. This parameter is not useful for the detection of saline soils in this classification. Rg is significant for all three groups. 5.92 Plant Parameters The A L F group was significantly different (p < 0.05) for all parameters. All groups were lower in absolute value than in the C and R^ comparison and may provide less meaningful continuous classifications. The HIGH group of C and Rg should include more of the HIGH values and show higher absolute values, lower SD and % CV if it is to be superior to the C and R^ classification. The MED group had a much wider SD and % CV than in C and R 4 . Thus the C and Rg classified ALF parameter provided a less realistic continuous classification than the C and R^. SALT provided good groupings in terms of trends and absolute values but only the LOW and HIGH groups were significantly different (p < 0.05). The GRASS parameter showed good group separation in terms of low SD and % CV but none of the groups were significantly different (p < 0.05). Individual grass species showed no significant differences between groups, excepting BRIN. Differences between the LOW vs. MED and HIGH groups were significantly different (p < 0.05) in the BRIN parameter. Six of the cases in the LOW group were zeros, the other three values were 28%, 4% and 4%; thus, this was not a representative class. HOJU, DIST, PUNU and POPR were not significantly different between groups. The lack of group significance observed between all grass species parameters is a reflection of the important contribution of Y (green) reflectance to DN distinction of Table 15 - Comparison of three classes of SALT using C and R3 for spectral, plant and soil parameters. Mean separation was accomplished using a non-parametric t-test ( #. = 0.05). Means within rows followed by the same letter are not significantly different. CLASS Parameters HIGH SALT MED SALT HIGH SALT (LOW ALF) (MED ALF) (HIGH ALF) n = 9 n = 24 n = 51 Spectral y x 146.9a 118.3bc 123, ,1c s 14.7 10.6 13, .1 %CV 10.0 8.6 11, .1 m x 113.7 94.7 79. ,4 s 18.7 16.7 13, .7 %CV 13.0 17.7 17, .3 All groups significantly different c X 156.2a 154.1a 173, .lb s 9.4 12.8 7, .9 %CV 6.0 8.3 4, ,6 R3 x 0.92 0.62 0, .46 s 0.08 0.11 0, .07 %CV 9.2 17.4 15, .5 All groups significantly different Plant ALF x 0.7 11.1 21. ,5 s 0.7 13.4 18, .8 %CV 100.0 120.1 87, ,4 All groups significantly different SALT x 32.1a 15.8ab 6, ,4b s 25.0 24.4 14, .4 %CV 77.8 153.9 224, .1 BRIN x 3.9 16.8 19. .0 s 8.8 12.0 17. ,4 %CV 226.9 71.2 91, .7 Table 15 (Continued) CLASS Parameters HIGH SALT MED SALT HIGH SALT (LOW ALF) (MED ALF) (HIGH ALF) (n = 9) (n = 24) (n = 51) Soil PHW x s %CV 9.33a 0.93 9.9 EC x s %CV 1.79a 1.29 72.0 ECa x s %CV 5.21a 1.8 24.5 ENa x s %CV 11.1a 7.4 66.4 7.68bc 0.76 9.9 7.91c 0.82 10.4 0.30bc 0.31 104.4 8.20bc 2.02 24.6 0.38c 0.40 103.2 7.83c 2.30 29.2 1.67bc 3.05 182.1 2.12c 3.15 148.8 131 different vegetation groups. This analysis was unable to provide a useful continuous classification due to few significant groups in the plant parameters. The inclusion of Y in R 4 (M/Y) in C and R 4 vs. ALF clearly produced superior DNs for the indicator species. 5.93 Soil Parameters The C and Rg classification grouped soil parameters PHW, EC, ECa and ENa with significant separations between LOW and MED and LOW and HIGH groups. However, the SD and %CV were higher for the groups classified by C and R 4 for all parameters. Linear trends were established but were less useful in establishing continuous classifications. 5.94 Mis-classified Units Mis-classified units are those which are classified in one group on the basis of their DN, but which have soil characteristics of another unit. The number of mis-classified units is a measure of the discriminating power of the model. Cases 27, 54, 58, 59 and 65, the cases with abnormally low C values, were all classified in group - MED ALF despite their Rvalues being distributed similar to that of the R 4 parameter. The C and R 4 classification distributed these cases among the three groups. These cases will have a minimal effect on the DNs of the three groups in this classification. However they do influence values in the C and Rg classification. The MED group had the largest SD and % CV of all three groups (see Figure 33) Digital numbers were not as distinct as with the C and R 4 classification. This may be due to a loss of discriminating power with removal of the Y parameter. Examination of the HIGH ALF class (n = 51) reveals eleven mis-classified units in terms of soil parameters (eg. pH > 8.2) (11, 29, 35, 40, 42, 43, 80, 121, 122, 125, 130). This represents 21.5% of the class (11/51) or 14.3% of all cases (11/83). Cases 12, 40, 42, 43 and 121 all have significant ALF and/or GRASS components so they should have been in the MED group while the remaining cases should have been in the LOW group. This is 132 evidence that an alfalfa capability classification system based on C and Rg develops less accurate continuous classes of vegetation and soil. 5.95 Comparison of C and R 4 and C and Rg Classifications Clearly spectral parameters C and R 4 provided an improved classification over the C and Rg analysis (5.0% of total mis-classified for C and R 4 vs. 14.3% for C and Rg). In terms of spectral parameters for use in the supervised computer-assisted classification of the study site, the C and R 4 analysis utilized green (Y) reflectance where Y was not included in C and Rg; C successfully discriminated LOW and MED and LOW and HIGH while C was only significantly different (p < 0.05) between LOW and HIGH in C and Rg. M and Rg, M and R 4 were significantly different for all groups in their respective analysis. Other workers (Murtha 1981; Thomson et al. 1985; Smith et al. 1987) found Rg (NIR/R) to be more useful than in this study. C and R 4 provided superior classes of plant parameters with most groups being significantly different (p < 0.05) between LOW and MED; SALT between all classes. Those parameters which were not significantly different between MED and HIGH maintained linear trends useful in the development of continuous classes. C and Rg showed trends but had fewer significantly different groups. The classification systems had similar results in soil parameters in that the LOW and MED classes; (p < 0.05) and the LOW and HIGH classes were significantly different for most parameters and the HIGH group maintained a linear trend with the other groups. However, the C and R 4 derived classes had lower % CV and SD. C and R 4 developed DNs whiqh will be used in the final phase of the study site classification, the supervised computer classification of photo images. 5.96 Additional Soil and Plant Parameters Soil parameters PHOS, TN, ORGC, BD and the plant parameter BIOMASS were evaluated as they were strongly linked to parameters included in the analysis and 133 therefore may contribute further information to the alfalfa capability assessment (See Table 16). PHOS was not significantly different (p < 0.05) between the groups but the trend showed higher available PHOS levels for the LOW ALF unit than for the HIGH ALF unit. PHOS was not correlated with ECa, although these are alkaline soils, but it was positively correlated with PHW. This relationship was reflected in the greater PHOS availability in the LOW ALF (HIGH pH, HIGH SALT) group. Provincial soil interpretations (Neufeld 1980) multiply Olson (Black 1965) PHOS levels by 2.5 - 3 times to gain equivalence with Bray test interpretations for the province of B.C. (van Lierop, 1987 - personal communication, October 1987). PHOS values are LOW - MED throughout the study site. (Appendix 2) The ECarENa ratio was an excellent parameter for separating the LOW ALF group but the MED and HIGH groups were not significantly different (p < 0.05). The MED and HIGH groups have much higher means and wide SD and % CV. This ratio is composed of two inversely related parameters, ECa and ENa. ENa in particular shows wide variation in the MED and HIGH group. This contributes to the wide range of values in this parameter. TN and ORGC are indicators of the soil's ability to supply nitrogen and other nutrients (Brady 1984). These parameters also provide information on past biomass production. The contribution of roots and leaf litter to the soil is a function of that biomass production. A subset of cases for which biomass data was available (n = 20) was grouped by cluster analysis of C and R 4 (Table 16). BIOM was significantly different (p < 0.05) between the LOW and HIGH classes only but a strong linear trend was evident between all groups with the lowest BIOM values in the LOW class (260.3 kg ha"1) and the highest values in the HIGH class (412.9 kg ha"1). The LOW and HIGH classes had smaller SD and % CV than did the MED class (Figure 34). Group membership was similar to that of the full data set; LOW 25% (5/20), MED 45% (9/20) and HIGH 30% (6/20). These factors 134 Table 16 - Classification of ALF for associated soil parameters. C and R4 were used as grouping variables to evaluate secondary soil parameters associated with the three primary ALF classes. CLASS Parameters LOW ALF (n = 14) MED ALF (n = 35) HIGH ALF (n = 29) PHOS x s %CV 18.5 8.0 43.2 14.5 9.0 62.0 12.8 5.7 44.5 TN x s %CV 0.13 0.05 38.5 0.15 0.04 26.7 0.16 0.03 18.8 0RGC x s %CV 1.76 0.70 39.8 1.97 0.49 24.9 2.15 0.40 18.6 Ca:Na x s %CV 2.07 4.42 213.5 38.39 31.45 81.9 34.92 27.62 79.0 BIOM x s %CV (n = 5) 260.3a 74.2 28.5 (n = 9) 374.2b 155.7 41.9 (n = 6) 412.9b 95.5 23.1 135 Class Frequency (X) 100 T 45 37 16 -260.3 374.2 412.9 Biomass (kg-ha.- l) LOW M E D I U M H I G H 600.0 Figure 34 - Continuous Classification - BIOM. Distribution of biomass values of three c lasses - LOW ALF, MEDIUM A L F and HIGH A L F depicted as histograms showing the mean class value and plus and minus one standard deviation. 136 again indicate the variability of soil conditions in the MED group. Biomass has been successfully predicted by spectral parameters in studies by a number of workers (Richardson et al. 1983; Thomson etal. 1985). Spectral parameters M, C and R 4 had moderately strong correlations with BIOM. Y was not correlated. BIOM exhibited very strong positive correlations with ALF; strong positive correlations with E, TN, ORGC and BRIN. Strong negative correlations were noted between BIOM and SALT, POJU and PUNU. This limited data set is too small to make field predictions. However, the relationships of BIOM with key parameters shows promise for similar future studies. Strong evidence for a significant association between biomass production and soil salinity was observed in the linear trend between groups and the relatively narrow standard deviations within groups when compared with other parameters. A larger data set would be necessary to define the relationship. 6.0 RESULTS OF COMPUTER-ASSISTED SUPERVISED CLASSIFICATION 6.1 Supervised Classification The DNs presented in Table 13 were derived from analysis of the data bank of spectral, plant and soil parameters sampled at 84 case locations selected from the series of ground-truthing locations. This supervised classification was intended to separate the data set into the maximum number of classes for which the parameters were significantly different (p < 0.05). In this study, the number of classes was governed, in part, by the size of the management unit, the production capability of the unit and the feasibility for management or reclamation strategies. The system used four categories; LOW ALF, MED ALF, HIGH ALF and UNCL (unclassified). The map production function of the Meridian system allows the interpreter to assign a palette of colours to the 0 - 255 (0 = black; 255 = white) grey scale. The range of pseudocolours on the palette can be compressed and moved to the desired range of intensities to highlight small changes within this range (McDonald, Dettwiler amd Assoc. 1986). The spectral parameters for the different classes exhibited linear trends. Y and decreased from LOW ALF to HIGH ALF; M and C increased from LOW ALF to HIGH ALF. All categories had at least three parameters in which all groups were significantly different (p < 0.05). Thus, each class had a mean and range (assigned by the computer) for each parameter. Each group was then assigned a colour, chosen to highlight contrast between groups (UNCL = black; LOW = Yellow, MED = Green, HIGH = Red). The range of digital number values exhibited for study site soil and vegetation did not fully utilize the 0 - 255 gray scale. Although exposed sodic soil "slicks" and white, saline crusts can be found within the images they were not included with sufficient frequency in the data set to establish separate classes. DN's for these areas thus fell into the unclassified category. Similarly unclassified were exposed, light soil on adjacent poor-condition rangelands, shadows in the adjacent forest canopy and bare eroded soil on the creek bank to the south of the study site. Features within a field such as stone piles, landclearing debris and isolated trees can be misinterpreted as sodic or saline spots. Visual inspection of paper prints or positive transparencies can clarify these errors. R^ was found to exhibit too large an unclassified region to produce a useful map despite the difference (p < 0.05) among classes. The range of possible values outside those exhibited by the sample cases was too broad to be accommodated by the capability of the system. The results of a supervised cluster analysis of digitized colour-IR images PH 8882, PH 8909, PH 8883 and PH8911 are presented in Figures 35-38. In PH 8882 (Field 1) (Figure 35) characteristic features include an UNCL region consisting of trees and stonepiles in the N portion of the field, two dirt roads intersecting at right angles in the SW portion and a large area of black, exposed sodic soil in the SE portion. The LOW ALF unit consisted of saline soils bordering the N and NE portion, and senescent native grasses in the S and SW. Encroaching saline/sodic spots were located in the S portion of the field. MED ALF areas surrounded these sodic zones in the S and were adjacent to them in the N PH 8882 - Supervised classification of a digitized colour infrared image of Field 1. HIGH ALF, MEDIUM ALF, LOW A L F and unclassif ied c lasses are coloured red, blue, yellow and black respectively. 139 and SE. The lower elevation areas located in the SW portion of PH 8882 are composed of MED ALF sites. The NW sector was categorized as HIGH ALF. Two cases in PH 8882 were re-classified by the computer. Cases 12 and 29 were located (See Table 17) adjacent to seeps and included in these units. These cases were among five removed from the C and analysis. Image PH 8909 (Field 2) (Figure 36) had an area of UNCL and MED ALF in the NW corner and an area of UNCL in the SW quarter. The NW corner consisted of vegetation of lower density and bare soil. The extensive area in the SW corner consisted of land-clearing debris with some small patches of bare soil. LOW ALF was represented by the large saline area on the WT side, a site at the S end where the image adjoins that of PH 8882 and scattered points in the NW corner. Some areas of senescent native range vegetation (Stipa comata, Trin and Rupr., Stipa spartea, Trin. and Poa sandbergii, Vasey) were included in this group on the N and W borders. With the exception of some scattered patches throughout the site and concentrations of HIGH ALF N and S of the saline area on the W side, the remainder of the scene is MED ALF. Three units in this image were re-classified from MED ALF to LOW A L F and one unit from HIGH ALF to LOW ALF. These cases (34, 35, 40 and 42, respectively) were located on the W side of Ph 8909 (Field 2). The computer-based classification had delineated this as an area of increasing salinity. The features of PH 8883 (Field 3, Field 4) (Figure 37) included two large UNCL areas mixed with LOW ALF in the NW and NE portion of the image. The NE portion consisted of a road, exposed soil and SALT vegetation. The LOW ALF in the NW sector was a combination of salt tolerant vegetation (SALT) and senescent grass species (GRASS) as well as alfalfa. The UNCL adjacent to this area was a pond. A large area of LOW ALF extended to the S of this unit, with two additional areas of LOW ALF found along the S boundary of the image. These are areas of less dense, senescent vegetation, primarily grass species with some alfalfa. The NE corner showed mottling of UNCL with HIGH 140 Table 17 - Supervised computer classification of training sites. Assignment of training sites within each category of ALF is shown for the four photographs used in the analysis. Photograph Class PH8882 PH8909 PH8883 PH8911 LOW 1, 6, 26, 49 59, 65, 70 ALF 31 79, 85, 92 93, 98, 99 From M-L* 35, 40, 42 63 125, 129, 133 From H-L* 12, 29 80 121, 122, 130 MED ALF 3, 11, 15, 16, 17, 21, 27 33, 36, 44 72, 73, 76, 89, 95, 100, 103 106, 118, 135, 145 112, 123, 137, 114 127 140 From H-M* 68, 90 121 HIGH ALF 9, 22, 29 38, 45, 41, 46, 43, 53 60, 62, 67, 82, 86, 87, 101, 102 111, 115, 142 * Cases which were re-classified during supervised computer analysis of the training site data. The cases indicated were moved from medium (m) to low (L), high (H) to low (L) or high (H) to medium (M) in the respective photographs. Figure 36 - PH 8909 - Supervised classification of a digitized colour infrared image of Field 2. HIGH ALF, MEDIUM ALF, LOW A L F and unclassified (UNCL) c lasses are co loured red, blue, yellow and black respectively. Figure 37 - PH 8883 - Supervised classification of a digitized colour infrared image of Field 3 (right side) and the eastern half of Field 4. HIGH ALF, MEDIUM ALF, LOW A L F and unclassif ied (UNCL) c lasses are coloured red, blue, yellow and black respectively. 143 ALF. This portion of the image included both small patches of sodic soils and shadowing effects as a consequence of the sun angle. The photos were taken between 11:19 and 11:35 AM. Shadowing from adjacent forest cover was noted at the eastern boundary of PH 8883. There was evidence from the image position studies of a tendency for under-exposure on the extreme Right (East) of the photos. The abnormally low C values for Cases 54, 56, 57, 58 (Appendix 2) were consistent with this exposure effect. Large areas of MED ALF dominated Field 4. Areas of HIGH ALF were also visible within the field. The orientation of these HIGH ALF areas suggested they may reflect a response to the surface application of fertilizer (11-55-0) in 1985. (See Site History, Section 2.1) Cases 63 and 80 in PH 8883 were re-classified from MED ALF to LOW ALF. Cases 68 and 90 were re-classified from HIGH ALF to MED ALF. PH 8911 (Field 4 and 5) (Figure 38) was characterized as a large, mixed unit of UNCL, LOW ALF, MED ALF and HIGH ALF in the N central portion of the image. The mottled UNCL, MED A L F and HIGH ALF area in the NW section of this unit was an aspen grove. Surrounding the aspen to the E, S and W is mixed UNCL and LOW ALF indicative of salt tolerant vegetation and exposed soil. Saline areas can be seen extending S from this area. Other LOW ALF areas were found in the NE portion of the image. The UNCL areas in the SE portion of the image represented sparse grass vegetation with some exposed light soil. Areas of HIGH ALF were found in the W end of the site. These areas also appeared to be showing a response to surface-applied fertilizer. Other areas of HIGH ALF were located on the E side of the image, centered around the main irrigation line. The remainder of this field was MED ALF. Cases 125, 129 and 133 were re-classified from MED ALF to LOW ALF. Cases 121 and 109 were moved from HIGH ALF to LOW ALF and MED ALF, respectively. P H 8911 - Supervised classification of a digitized colour infrared image of the western half of Field 4 (right side) and Field 5. HIGH ALF, M E D I U M ALF , LOW A L F and unclassif ied (UNCL) c lasses are co loured red, blue, yellow and black respectively. 145 6.2 Filtered Image Visual presentation of the computer-generated classifications can be improved by the use of an image enhancement function. For this study, a 5 X 5 pixel filter was used to accomplish image "smoothing" (McDonald Dettwiler and Assoc. 1986). The filter removed isolated pixels or pixel groups which detracted from the visual appearance of the mapped classes. (Figures 39-42) Generally, this procedure resulted in improved clarity as "noise" pixels were removed. For example, the shadow/field boundary was improved in PH 8883 (Figure 41) and the unit appeared more contiguous. Occasionally, information is lost using image enhancement. For example, in PH 8909 (Figure 40) filtering removed many isolated pixels or pixel groups from the W side as anomalies; however, the information from soil data for the cases (Appendix 2) in this area of the image have indicated that this was, in fact, a region of increasing salinity. 6.3 Supervised Classification - Other Applications The preceding sections of this study have dealt with the development and results of a supervised classification based on ground truth data. Spectral information gathered at a series of training sites was used to create a set of DNs for the three classes of vegetation and soil parameters (LOW ALF, MED ALF, and HIGH ALF). This information was then used to develop a supervised classification of the remainder of the study area. A supervised classification can also be carried out with the aid of "a priori" knowledge of the spectral differences between the desired classes. Using this information, a greater number of groups can be separated. . Further partitioning of the UNCL category which was composed of values outside the range of recorded values from the study site is possible. Spectral reflectance in the C dye-layer for sodic soils and saline patches were lower and higher, respectively, than the general range. Additional information can be imparted by the maps when these sub-sets of the UNCL regions are distinguished. 146 Figure 3 9 - PH 8882 - Supervised classification of a digitized colour infrared image of Field 1 using a 5 x 5 pixel filter for image "smoothing". HIGH ALF, MEDIUM ALF, LOW A L F and unclassif ied (UNCL) classes are coloured red, blue, yellow and black respectively. Figure 40 - P H 8909 - Supervised classification of a digitized colour infrared image of Field 2. Using a 5 x 5 pixel filter for image "smoothing". H I G H ALF, MEDIUM ALF, LOW ALF and unclassified (UNCL) classes are coloured red, blue, yellow and black respectively. 148 Figure 41 - PH 8883 - Supervised classification of a digitized colour infrared image of Field 3 and the eastern half of Field 4 using a 5 x 5 pixel filter for image "smoothing". HIGH ALF, MEDIUM ALF, LOW ALF and unclassif ied (UNCL) c lasses are coloured red, blue, yellow and black respectively. 149 Figure 42 - PH 8911 - Supervised classification of a digitized colour infrared image of the western half of Field 4 and Field 5 using a 5 x 5 pixel filter for image "smoothing". HIGH ALF, MEDIUM ALF, LOW A L F and unclassif ied (UNCL) c lasses are co loured red, blue, yellow and black respectively. For large scale studies of the type described here a supervised classification with ground control is superior to a supervised classification of the area in which no ground control information is incorporated into the system. Unsupervised classifications would not provide quantifiable data in a study of this nature. 151 7.0 CONCLUSIONS The results of this investigation confirmed that remote sensing techniques can be used to detect and evaluate saline soils. Quantification of variations in spectral signature is the key to successful application of the technique; it is imperative that variation in the film-filter-object-image system be clearly understood to facilitate interpretation. Understanding the capabilities and limitations of the technique used is the first step in any such study (Murtha 1981). The remote sensing system used permitted interpretation of CIR aerial photographs using the same computer-based analytical system that is available for Multispectral Scanner (MSS) Data. In this study, vegetation density values were located within the linear portion of the exposure curve, indicating good exposure for alfalfa. The IR balance of the film (22) used was more appropriate for higher altitudes; the increased sensitivity of the C dye-layer produced an image with intensified magenta tones. This effect may account for the narrower range of values observed in this study for the C parameter and may have resulted in some loss of discriminating power. The importance of image window positioning was also taken into account. Measurement of edge pixels was avoided as far as possible to eliminate the confounding factors associated with exposure at photo edges as discussed by Moore (1980). The eastern portion of PH 8883 had four cases with low C values. These may be due to shadowing or vignetting effects. Data from different aerial photographs may be included in the same analysis if the factors which influence image formation are first considered. Careful analysis of the four photographs used in this investigation was used to characterize and select the appropriate data set for subsequent analysis of saline soil stress effects on vegetation. Parameter measurements of the spectral-plant-soil system consisted of the best available information within the resources of the project. A larger ground truthing data set for biomass, soil sampling by textural layer and radiometer correlation data would 152 have supplied useful additional information. Canopy cover estimates based on leaf area rather than the point estimate technique may have produced stronger spectral-plant correlations. All parameters used in the classification analysis were selected on the strength of their correlations with other parameters in the system. Using this approach no pre-conceived categories were imposed on the system. This analytical approach was also applied to the cluster analysis of the ground truth data. The DNs were developed on the principle of vegetation adaptation to edaphic conditions (in this study - categories of saline soil). The system successfully discriminated DNs which were indicative of classes of vegetation which are, in turn, indicative of gradations of saline soil conditions. Parameters C and R^ proved to be the most strongly-related to alfalfa. As a primary aim of the study was to evaluate remote sensing techniques in designating alfalfa capability classes of saline soils, these parameters were used to cluster a series of sites into three categories. The groups of spectral parameters were used to determine digital numbers (DN) in the Y, M and C dye-forming layers, representing green, red and NIR reflectance for use in the Meridian PC - image analysis system. The plant and soil parameter trends and groups were used to determine continuous classifications for the three categories. Related plant and soil parameters were included in the analysis at this point. The parameter means were significantly different (P < 0.05) between LOW ALF and MED ALF, LOW ALF and HIGH ALF, in most cases and between MED ALF and HIGH ALF in some cases. The continuous classification utilized the range of available experimental values (expressed in terms of +_ 1SD and % CV) to devise classes applicable to management or reclamation strategies. The LOW ALF class indicated areas of saline and/or sodic soils marked by high pH, EC, ENa and EK and by low ECa. This category represented those areas which require reclamation strategies in order to become productive. The MED A L F class included areas of developing salinity, areas of lower 153 alfalfa populations and sites which included grass species which may or may not have been senescent. These MED ALF capability areas were of lower capability and likely require fertility management, more sensitive irrigation scheduling or simple reseeding to become productive. HIGH ALF areas represented, in most cases, those sites with high capability for alfalfa production. Representatives of this class may have contained high populations of grass species in lieu of alfalfa which had declined for reasons other than salinity. The limited biomass data set indicated higher production in these areas which were characterized by neutral pH, low EC, ENa and EK and high ECa. The fertility indicators TN and ORGC, showed negative correlations with saline soil indicators and positive trends with alfalfa. They were useful indicators of the productivity potential of the soil. TN levels appeared generally low to medium in comparison to common ratings for alfalfa production in forage agriculture. Phosphorus was positively correlated with pHW reflecting the prevalence of calcareous soils. More available phosphorus was present in the LOW ALF unit than in the HIGH ALF group. The phosphorus range for HIGH ALF sites was 7-19 ppm (x _+ 1SD). Olson's test data for BC calcareous soils provided values which must be increased by 2.5 -3x to approximate Bray test values (Neufeld 1986). Using the higher factor, correction yielded a range of 21 - 57 ppm. Soils with these phosphorus levels would require 28 - 90 kg ha"1 supplemental P2O5 f ° r satisfactory alfalfa production. Higher levels of available phosphorus were found in the LOW ALF and MED ALF categories reflecting the units investigated required some supplementary phosphorus for optimum production of alfalfa. This may indicate shortages of PHOS at all sites. Basic soils are noted for moderate fixation of P as Ca phosphates (Bohn et al. 1979). Greater available PHOS at the LOW and MED sites may have resulted from measurement of some of this fixed P. The DNs supplied to the Meridian PC-image analysis system were used to produce capability maps of the study site depicting the three alfalfa production classes plus an additional category of UNCL. Pixel classification data were not available from the system 154 but the cases were grouped LOW ALF - 18%, MED ALF - 43%, HIGH ALF - 37%. Computer-assisted supervised classification of the data improved the groups to produce a distribution of LOW ALF - 31%, MED ALF - 42%, HIGH ALF - 28%. This distribution appears more representative as the randomly stratified training sites did not include areas of visible salinity symptoms. This pilot study demonstrated that computer-based analysis of digitized CIR aerial photographs may be a valuable tool in the identification and evaluation of saline soils. 155 8.0 RECOMMENDATIONS: The following are a summary of recommendations regarding future studies of this nature: A. Film - Filter - Image Analysis 1. Colour transparencies were shown to provide greater exposure latitude and superior ability to discriminate subtle spectral differences than paper prints, confirming their suitability for quantitative densitometric analysis. 2. The effect of image fall-off requires that photograph edges be excluded from use in densitometry measurements.Data sets from different photographs or portions of photographs should not be used in classification work until the variation between images has been evaluated. 3. Processing may add confounding effects to image interpretation through the tendency toward underexposure and/or overexposure near the edges of the photograph.Such processing effects and those of vignetting may be reduced by avoiding edge measurements. B. Sampling and Data Collection 1. Stratified random sampling produces good general classes of alfalfa capability. Such classifications may increased in number and significance with pre-typing of some areas (eg. saline crusts, bare soil, sodic soil) and including these samples in the analysis. 2. Radiometer data should be collected at the sites as resources permit. These additional data would enable the development of more species-specific DNs (radiometer readings of homogeneous vegetation groups) as well as permit better calibration of image and site data. 3. Plant population estimate correlations with soil and spectral parameters could be improved upon with use of canopy cover data based on leaf area of a species vs. point estimates. These data would produce a more accurately quantifiable bare soil component as well . 4. Soils should be sampled by soil horizon for chemical and physical parameters where resources permit. Subsoil chemical and physical information m a y provide greater insight into salinit) ' effects throughout the rootzone.Water movement and its effect on sal ini ty encroachment could also be better interpreted. 5. Use of field markers for locating cases would facilitate location of t raining sites during spectral data collection and analysis. C. Reclamation/Management Due to the nature of data acquisition only l imited recommendations can be made regarding appropriate reclamation/management strategies. This study has evaluated study site soils and classified them into three categories of alfalfa management units related to the severity of sal inity effects. Addi t ional soil and vegetation data would be an essential pre-requisite for the provision of opt imum production recommendations for alfalfa cultivation. 157 Literature Cited Alberta Agriculture. 1980. Irrigation in Alberta. Agdex 560-1. Bauer, M.E., Vanderbilt, V.C., Robinson, B.F. and C.S.T. Daughtry. 1980. Spectral properties of agricultural crops and soils measured from space, aerial, field and laboratory sensors. Proceedings of the International Archives of Photogrammetry, XXIII (B7):56-73. Beckett, P.H.T. and R. Webster. 1971. Soil variability: a review. Soil and Fert. 34(1):1-15. Bernstein, L. 1975. Effects of salinity and sodicity on plant growth. Ann. Rev. Phytopathol. 13:295-312. Best, K.F., Banting, J.D. and Bowes, G.G. 1978. The biology of Canadian weeds. 31. Hordeum jubatum. Can. J. Plant Sci. 58:699-708. Black, C.A. 1965. Methods of soil analysis Part 1 - Physical properties, Part 2 -Chemical and microbial properties. American Society of Agronomy, Inc. Madison, Wise. 1592 pp. Bohn, H.L., McNeal, B.L. and O'Connor, G.A. 1979. Soil Chemistry. John Wiley and Sons, Toronto, 313 pp. Bower, C.A., Ogata, G. and Tucker, J.M. 1968. Sodium hazard of irrigation waters as influenced by leaching fraction and by precipitation or solution of calcium carbonate. Soil Sci. 106:29-34. Brach, E.J. and Mack, A.R. 1977. Differentiation of selected annual field crops throughout the growing season by their spectral reflectance properties. Can. J.R.S. 3(l):55-65. Brady, N.C. 1984. The Nature and Properties of Soils. 9th Edition. MacMillan Publishing Co., Inc. New York, 750 pp. Bresler, E., McNeal, B.L. and Carter, D.L. 1982. Saline and Sodic Soils - Principles and  Dynamics-Modelling. Springer-Verlag, New York, 236 pp. Bridges, E.M. 1970. World Soils. Cambridge University Press, London, 89 pp. Brinker, R.C. 1969. Elementary Surveying 5th ed. International Textbook Co., Scranton, Pa. 620 pp. British Columbia Ministry of Agriculture and Fisheries, 1986. Alfalfa Winterkill - 1986. Williams Lake District Office. British Columbia Ministry of the Environment. 1981. Chilcotin River 92 0/NE. 1:50,000 Topographic Map. 158 Canada Soil Survey Committee, Subcommittee on Soil Classification. 1978. The Canadian system of soil classification. Can. Dep. Agric. Publ. No. 1646, Ottawa, 164 pp. Carter, D.L. 1981. Salinity and plant productivity. In: Chemical Rubber Co., Handbook Series on Nutrition and Food. Chamblee, D.S. 1972. Relationships with other species in a mixture. In Alfalfa Science  and Technology ed. C H . Hanson. American Society of Agronomy, Madison, Wise. 812 pp. Chatervedi. L., Carver, K.R., Harlan, J . C , Hancock, G.D., Small, F.V. and Dalstead, K.J. 1983. Multispectral remote sensing of saline seeps. IEEE Transactions of Geoscience and Remote Sensing. Vol. GE-21, No. 3. Chorley, R.J. and Kennedy, B.A. 1971. Physical Geography - A Systems Approach. Prentice Hall International Inc., London, 370 pp. Crown, P.H. 1979. Crop identification in a parkland environment using aerial photography. Can JRS 5(2):128-135. Dalstead, K.J. , Worcester, B.K. and Brun, L.J. 1979. Detection of saline seeps by remote sensing techniques. Photogramm. Eng. 45(3):285-291. Daubenmire, R.L. 1965. Plants and Environment. John Wiley and Sons, Ltd. 422 pp. Dixon, W.J. 1985. BMDP7D - Description of groups with histograms and ANOVA. BMDP6D - Bivariate scatter plots. BMDP2R - Stepwise regression. BMDP3D -Comparison of two groups with t-tests. BMDP5D - Histogram and univariate plots. Statistical Software Manual. University of California Press. Berkeley, Ca. 732 pp. Driscoll, R.S., Reppert, J.N. and Heller, R.C. 1974. Microdensitometry to identify plant communities and components on colour infrared aerial photos. J. Range Man. 27(l):66-70. Egan, W.G. 1985. Photometry and Polarization in Remote Sensing. Elsevier Scientific Publishing Co., New York. 503 pp. Ehlig, C F . 1960. Effects of salinity on four varieties of grapes grown in sand culture. Proc. Am. Soc. Hortic. Sci. 74:661-670. Elgin, J.H., Tucker, C.J., McMurtrey, J.E., III, Holben, H. 1979. Relationships between ground-collected spectral radiation and crop canopy variables. Agr. Abst., Mtg, Am. Soc. Agron., Ft. Collins, Co. Everitt, J.H., Gerberman, A.H., and Cuellar, J.A. 1977. Distinguishing saline from non-saline rangelands with skylab imagery. Photogramm. Eng. and RS. 43(8): 1041-1047. Fischer, W.A. 1975. History of remote sensing. In Manual of Remote Sensing. R.G. Reeves (ed.). Am. Soc. of Photogramm. Falls Church, Va. 2144 pp. 159 Fleming, J. 1978. Exploiting the variability of aerochrome infrared film. Photogramm. Eng. 44(5):601-605. Fowler, D.B. 1981. Fall growth and cold acclimation of winter wheat and rye on saline soils. Can. J. PI. Sc. 61:225-230. Fritz, N. 1967. Optimum methods for using infrared-sensitive colour film. Photogramm. Eng. 33:1128-1138. Gates, D.M. 1968. Sensing biological environments with a portable radiation thermometer. Appl. Opt. 7:1803-1809. Gausman, H.W. 1974. Leaf reflectance of near-infrared. Photogramm. Eng. 40(2): 183-191. Goh, T.B., Pawluk, S., and Dudas, D.. 1986. Adsorption and release of phosphate in Chernozemic and Solodic/Solonetzic soils. Can. J. Soil Sci. 66:521-529. Grove, A.R. and Carlson, G.E. 1972. Morphology and anatomy, in Alfalfa Science and  Technology, ed. C H . Hanson. American Society of Agronomy. Madison, Wise. 812 pp. Hall. P.M., McLean, J.A. and Murtha, P.A. 1983. Dye-layer density analysis identifies new Douglas-fir beetle attacks. Can. J . For. Res. 14(2):279-282. Halvorson, A.D. and Reule, C.A. 1980. Alfalfa for hydrologic control of saline seeps. Soil Sci. Soc. Am. J. 44:370-374. Hillel, D. 1980. Fundamentals of Soil Physics. Academic Press, Toronto. 413 pp. Hitchcock, C L . and Cronquist, A. 1973. Flora of the Pacific Northwest. University of Washington Press, Seattle, WA, 730 pp. Holland, S.S. 1976. Landforms of British Columbia: A Physiographic Outline. B.C. Department of Mines and Petroleum Resources. Bulletin 48, 38 pp. Jackson, R.D. 1986. Remote sensing of biotic and abiotic plant stress. Ann. Rev. Phytopathol. 24:265-287. Kelley, P.S. 1983. Digital analysis in remote sensing: from data to information. In: Remote Sensing of the Environment. 2nd edition. B.F. Richadson, Jr., (ed.) Kendall/Hunt Pub. Co. 496 pp. Knipling, E.B. 1970. Physical and physiological basis for the reflectance of visible and near-infrared radiation from vegetation. Remote Sensing Environ. 1:155-159. Kowalenko, C.G. 1985. Implications of soil variability on fertilizer assessments for crop production in Proc. 9th B.C. Soil Science workshop. MOE Tech. Report 16. Victoria, B.C. Ladouceur, G., Allard, R., and Ghosh, S. 1986. Semi-automatic survey of crop damage using colour infrared photography. Photogramm. Eng. and R.S. 52:111-115. 160 Levitt, J. 1972. Response of plants to environmental stresses. Academic Press, New York. Lillesand, T.M. and Kiefer, R.W. 1979. Remote Sensing and Image Interpretation. John Wiley and Sons, Toronto, 612 pp. Lo, T.H.C., Scarpace, L. and Lillesand, R.M. 1986. Use of multi-temporal spectral images in agricultural land-cover classification. Photogramm. Eng. and R.S. 52:535-544. Lord, T.M. and McLean, A. 1969. Aerial photo interpretation on British Columbia rangelands. J. Range Man. 22(l):3-9. Maas, E.V. and Hoffman, E.J. 1977. Crop salt tolerance - current assessment. J. Irr. and Drainage. Div. ASAE 103:115-134. McDonald Dettwiler and Associates. 1986. Meridian PC Programming Guide. Version 2.0, October, 1986. McGraw, J.F. and Tueller, P.T. 1983. Landsat computer-aided analysis techniques for range vegetation mapping. J. Range Man. 36(5):627-631. McLean, A. 1979. Range Management Handbook for British Columbia. Agriculture Canada, Kamloops, B.C. 104pp. Miller, M.R., Brown, P.L., Donovan, J.J. , Bergatina, R.N., Sonderegge, J.L. and Schmidt, F.A. 1981. Saline seep development and control in the North American Great Plains - hydrogeological aspects. Agr. Water Man. 4:115-141. Moon, D.E. and Schreier, H. 1985. Soil variability: does it invalidate our interpretations? in Proc. 9th B.C. Soil Science Workshop. MOE Technical Report 16. Victoria, B.C. Moore, R.E. 1980. Film filter combinations for aerial colour photography. Supply and Services Canada. Stock SMP-1263. Murtha, P.A. 1981. Detection and analysis of vegetation stresses. In Uses of Remote  Sensing in Forest Pest Damage Appraisal. Ed. R.J. Hall. Northern Forest Research Center, Canadian Forestry Service, Environment Canada, Edmonton, Alberta. Murtha, P.A. 1982. Detection and analysis of vegetation stress pg. 141-158. In Johannsen, C L . and J.L. Saunders, eds. Remote Sensing for Resource Management. Soil Cons. Soc. Am. Ankeny, Iowa, 665 pp. Neufeld, J.H. 1980. Soil testing methods and interpretations. British Columbia Ministry of Agriculture and Fisheries. 80-2. Ott, L. and Mendenhal, W. 1985. Understanding Statistics. 4th Edition. Duxbury Press, Boston, Mass. 493 pp. 161 Patterson, J.M. and Whitaker, R.A. 1982. U.B.C. CGroup-hierarchical grouping analysis with optional contiguity constraint, University of British Columbia Computing Center, Vancouver, B.C. Prairie Farm Rehabilitation Administration. 1983. Land Degradation and Soil Conservation Issues on the Canadian Prairies. Agriculture Canada, Regina. 326 pp. Proctor, J . and Woodell, S.R.J. 1975. The ecology of serpentine soils. Adv. in Ec. Res. 9:225-366. Richardson, A.J. , Everitt, J.H. and Gausman, H.W. 1983. Radiometric estimation of biomass and nitrogen content of alicia grass. R.S. Env. 13:179-184. Ringwood, P. 1981. Resurvey of parts of the exterior rectilinear boundaries of Sandy Harry Indian Reserve No. 4 and Alkali Lake Reserve No. 4A, Lillooet District, Province of B.C. #67340. Salisbury, F.W. and Ross, C.W. 1978. Plant Physiology. 2nd ed. Wadsworth Publishing Co., Belmont, CA. 422 pp. Saxton, K.E. , Rawls, W.J., Romberger, J.S. and Papendick, R.I. 1986. Estimating generalized soil-water characteristics from texture. Soil Sci. Soc. Am. J. 50:1031-1036. Schreier, H. and Zheng, F. 1986. Using remote sensing technology for field application of variable rates of fertilizer. Potash and Phosphate Institute. Atlanta, Georgia; Cochrane, Alta. Senate Committee on Agriculture, Fisheries and Forestry. 1984. Soil at Risk. Committees and Private Legislation Branch, The Senate of Canada, Ottawa, Ontario. 129 pp. Shukla, A.K. 1979. Specific ion effects on crops. Plant and Soil. 52:457-459. Sokal, R.R. and Rohlf, F.J. 1987. Biostatistics. 2nd ed. pg. 213. W.R. Freeman and Company, New York, 363 pp. Sommerfeldt, T.G. 1985. Landsat salinity mapping. Twenty-Second Annual Soil Science Workshop Proceedings. Sommerfeldt, T.G., Thompson, M.D. and Prout, M.D. 1984. Delineation and mapping of soil salinity in Southern Alberta from Landsat data. Can. J . of R.S. 10(2): 104-110. Smith, S.M., Schreier, H. and Wiart, R. 1987. Agricultural field management with microcomputer based GIS and image analysis systems. 2nd Annual International Conference, Exhibits and Workshops on Geographic Information Systems. Thompson, M.D. 1979. Remote sensing in operational range management programs in western Canada. In Proceedings of the 11th International Symposium on Remote Sensing of the Environment. (V2):1101-1110. 162 Thompson, M.D., Dams, R.V., McCartney, D.H., Chu, G., and Beaubier, G.M. 1984. Development of remote sensing systems and technology transfer for management of Saskatchewan prairie and parkland pastures. Can. J. of R.S. 10(2): 149-166. Thomson, K.P.B., Gosselin, C , Adams, B.W. and Sutherland, I. 1985. Thematic Mapper data used for rangeland management in rough fescue grasslands in Western Canada. Can. J.R.S. 11(2):162-176. Tipper, H.W. 1971. Glacial Geomorphology and Pleistocene History of Central British Columbia. Geological Survey of Canada. Bulletin 196, Ottawa, 89 pp. and 8 maps. Tucker, C.J., Elgin, J.H., Jr., McMurtrey, J.E., III. 1980. Relationship of crop radiance to alfalfa agronomic values. Int. J. Remote Sensing. l(l):69-75. Tueller, P.A. 1982. Rangeland Remote Sensing - An Approach. RNRF Symposium on the Applications of Remote Sensing to Resource Management. Am. Soc. Photogramm. Little Falls, VA. United States Salinity Laboratory Staff. 1954. Diagnosis and Improvement of Saline and Alkali Soils. The United States Department of Agriculture. Handbook 60. Ustin, S.L., Adams, J.B., Elvidge, C D . , Rejmanzk, M., Rock, B.N., Smith, M.O., Thomas, R.W. and Woodward, R.A. 1986. Thematic Mapper studies of semiarid shrub communities. Bioscience 36(7):446-452. Valentine, K.W.G., Sprout, P.N., Baker, T.E. and Lavkulich. 1978. Soil Landscapes of British Columbia. Resource Analysis Branch Ministry of the Environment. Victoria, 197 pp. Valentine, K. W.G. 1984. Soil Association of the Chilcotin River Region. 93 0/NE. (1:50,000 map). Agriculture Canada, Vancouver, B.C. Valentine, K.W.G. and Lidstone, A. 1985. Specifications for soil survey intensity (survey order) in Canada. Can. J. Soil Sci. 65:543-553. Vander Pluym, H.S.A. 1985. Locating the dominant recharge sources of dryland saline seeps. Twenty-second Alberta Soil Science Workshop Proceedings, Lethbridge, Alta. Vickery, P.J., Hedges, D.A. and Duggin, M.J. 1980. Assessment of the fertilizer requirement of improved pasture from remote sensing information. R.S. of the Env. 9:131-148. Watson, E.K. and van Ryswyk, A.L. 1986. Remote Sensing - Grasslands. Agriculture Canada, Kamloops, B.C. Wilding, L.P. 1984. Spatial variability: its documentation, accomodation and implications to soil surveys, in Soil Spatial Variability. Proc. of a Workshop of the ISSS and the SSSA D.R. Nelson and J. Bouma (eds.) Las Vegas, Nev. Center for Agricultural Publishing and Documentation. Wageningen, The Netherlands. 163 APPENDICES APPENDIX 1. Description of soil orders relevant to the study site The following information was extracted from the Canadian System of Soil Classification (Canada Soil Survey Committee 1978). CHERNOZEMIC ORDER The general concept of the Chernozemic order is that of well to i m p r e f e c t l y drained s o i l s having s u r f a c e horizons darkened by the accumulation of organic matter from the decompositin of x e r o p h y t i c or mesophytic grasses and forbs r e p r e s e n t a t i v e of g r a s s l a n d communities or of g r a s s l a n d - f o r e s t communities with a s s o c i a t e d shrubs and f o r b s . The major area of Chernozemic s o i l s is the c o o l , s u b a r i d to subhumid I n t e r i o r P l a i n s Region of Western Canada. Minor areas of Chernozemic s o i l s occur in some v a l l e y s and mountain slopes in the C o r d i l l e r a n Region extending in some cases beyond the tre e l i n e . Most Chernozemic s o i l s are f r o z e n during some period each winter and t h e i r s o l a are dry at some pe r i o d each summer. Th e i r mean annual temperature is high-er than 0*C and u s u a l l y l e s s than 5.5'C, but some Chernozemic s o i l s in dry v a l l e y s of B r i t i s h Columbia have higher tempera-tures . The s p e c i f i c d e f i n i t i o n is as f o l l o w s : S o i l s of t h Chernozemic order have an A horizon in which organic matter has accumulated (Ah, Ahe, Ap) that meets the requirements of a chernozemic A h o r i z o n . A chernozemic A has the f o l l o w i n g pro-p e r t i e s : 1. It is at l e a s t 10 cm t h i c k . 2. Its c o l o r value is darker than 5.5 dry and 3.5 moist, and i t s chroma is less than 3.5 moist. 3. Its c o l o r value i s at l e a s t one Munsell unit darker than that of the IC horizon.. 4. In s o i l s d i s t u r b e d by c u l t i v a t i o n or other means, the Ap horizon must be t h i c k and dark enough to provide 15 cm of su r f a c e m a t e r i a l that meets the c o l o r c r i t e r i a given in 2 and 3 above. 5. It c o n t a i n s between 1 and 17% organic carbon and i t s C/N r a t i o is l e s s than 17. 6. C h a r a c t e r i s t i c a l l y i t has s u f f i c i e n t l y good s t r u c t u r e that i t is n e i t h e r both massive and hard nor s i n g l e grained when dry. 7. Its base s a t u r a t i o n ( n e u t r a l s a l t ) is more than 80% and c a l -cium is the dominant exchangeable c a t i o n . '8. It is r e s t r i c t e d to s o i l s having a mean annual temperature of O'C or higher and a s o i l moisture subclass d r i e r than humid. 165 Chernozemic s o i l s h o r i z o n . They do not p o d z o l i c B, ev idence „. 3 meet the c r i t e r i a of G l e y s o l i c the s u r f a c e . may have an Ae h o r i z o n and a Bm or a Bt have any of the f o l l o w i n g : s o l o n e t z i c B, of g l e y i n g s t r o n g l y enough expressed to s o i l s , permafrost w i t h i n 2 m o f There are four great groups r e c o g n i z e d Brown Dark Brown Bl ack Dark Gray C o l o r va lue (dry) 4.5 to 5.5 chroma > 1.5 subar id to s e m i a r i d c o l o r va lue (dry) 3.5 to 4.5 chroma > 1.5 sem i ar i d c o l o r va lue (dry) < 3.5 chroma 1.5 or l e s s subhum i d c o l o r va lue ( d r y ) < 4.5 e luv i al A subhum i d H o r i z o n Sequences Great Group Brown, Dark Brown and Black Dark Gray Subgroup O r t h i c Rego C a l c a r e o u s E1uvi ated S o l o n e t z ic G1 eyed Gleyed Rego G leyed C a l c a r e o u s Gleyed E l u v i a t e d Gleyed S o l o n e t z i c O r t h i c Rego C a l c a r e o u s S o l o n e t z ic G leyed G leyed Rego Gleyed C a l c a r e o u s G leyed S o l o n e t z i c Ah, Bm, Cca or Ck Ah, C or Cca or Ck AT, Bmk, Cca or Ck ""Ali.^AT, Bt j or Bt_, Cca or " T k ~~ Ah, Ae, Bnjt j , Csa or Ck "AT, Tftngj, C k g j A~h~, Ckgj AT, Bmkgj, Ckgj "AT, A j e , B t j g j or Btg j , ~~CkgT" Ah_, Ae, Bnjt j g j , Csagj Ahe, Ae, Bm or Bt_j_ or Bt_, Cca or "CT Ahe, CCa or Ck Ahe, Bmk, Cca or Ck Ahe, A~e7 B n j t j , Cs or Ck ATe, Bmgj, Ckgj ATe, "TFoJ  ATe, Bmkgj, Ckgj ATe, Ae, B n j t j g j , Ckgj or Csgj 166 LUV I SOL IC ORDER S o i l s of the L u v i s o l i c order g e n e r a l l y have l i g h t c o l o r e d e l u v i a l h o r i z o n s and they have i l l u v i a l B horizons in which s i l i c a t e c l a y has accumulated. These s o i l s develop c h a r a c t e r -i s t i c a l l y in well to i m p e r f e c t l y drained s i t e s , in sandy loam to c l a y t e x t u r e d , b a s e - s a t u r a t e d parent m a t e r i a l s under f o r e s t veg-e t a t i o n in subhumid to humid, mild to very co l d c l i m a t e s . However, depending on the combination of s o i l environmental f a c t o r s , some L u v i s o l i c s o i l s occur under condi-t i o n s o u t s i d e the range i n d i c a t e d as c h a r a c t e r i s t i c . For exam-p l e , some L u v i s o l i c s o i l s develop in a c i d parent m a t e r i a l s and some occur in f o r e s t - g r a s s 1 and t r a n s i t i o n zones. L u v i s o l i c s o i l s occur widely in Canada, from the southern e x t r e m i t y of Onta r i o to the zone of permafrost and from New-foundland to B r i t i s h Columbia. The l a r g e s t area of these s o i l s occurs in the c e n t r a l to northern I n t e r i o r P l a i n s Region under deciduous, mixed and c o n i f e r o u s f o r e s t . The Gray L u v i s o l s of that area u s u a l l y have well developed, p l a t y Ae horizons of low chroma, Bt horizons with moderate to strong p r i s m a t i c or blocky s t r u c t u r e s , c a l c a r e o u s C horizons and s o l a of high base s a t u r a -t i o n ( n e u t r a l s a l t e x t r a c t i o n ) . Gray L u v i s o l s of the A t l a n t i c P r o v i n c e s , on the other hand, commonly have Bt horizons of weak s t r u c t u r e , and low to moderate base s a t u r a t i o n . The Gray Brown L u v i s o l s of southern O n t a r i o and some v a l l e y s of B r i t i s h Colum-bia c h a r a c t e r i s t i c a l l y have f o r e s t mull Ah h o r i z o n s , moderate to strong blocky s t r u c t u r e d Bt horizons and c a l c a r e o u s C ho r i z o n s . These are two great groups r e c o g n i z e d : Gray Brown L u v i s o l f o r e s t mull Ah Ae, Bt mean annual so i 1 temperature >8*C Gray L u v i s o l may or may not have Ah Ae, Bt u s u a l l y mean annual s o i l temperature <8"C Horizon Sequences Great Group Subgroup Gray Brown L u v i s o l O r t h i c Ah , Ae_, Bt_, Ck B r u n i s o l i c 7[F, Ae, J3m or B_f, Ae, B_t, BC, Ck P o d z o l i c LFH, Ah, Ae, B_f, Ae, B_t, BC, Ck Gleyed Ah, Aegj, B t g j , Ckg Gleyed B r u n i s o l i c AT, Bmgj , Aeg j , B t g j , Ckg Gleyed P o d z o l i c M, Ae, B f g j , Aegj, B t g j , Ckg Orth ic Dark Gray B r u n i s o l i c LFH, Ae, AB, _Bt, C or Ck LFH, TK or Ahe, Ajs, B_t, C or ~ LFH, Bm or B_f, Ae, Bit_, BC, C or Ck P o d z o l i c LFH, Ae, Bf, Ae, Bt_, BC, C or Ck S o l o n e t z i c LFH, Ae, AB, B t n j , BC, C or Tk F r a g i c LFH, Ahe, Ae, Bt_, Btx. or BCx, C Gleyed LFHT"Ae, B t g j , Cg Gleyed Dark Gray LFH, TK or Ahe, Ae, B t g j , eg Gleyed B r u n i s o l i c LFH, Bm or Bf, Aegj, Btgj , BCgJT Cg Gleyed P o d z o l i c LFH, Ae, Bf, Aegj, B t g j , BCgj, Cg Gleyed S o l o n e t z i c LFH, Ae, ABgj, B t n j g j , Cgj or Csag Gleyed F r a g i c LFH, Ahe, Aegj , Btgj , Btxgj or BCxg, Cg SOLONETZIC ORDER S o i l s of the S o l o n e t z i c order have B hor i zons that are very h a r d when dry and that swel l to a s t i c k y mass of very low perme-a b i l i t y when w e t . T y p i c a l l y the s o l o n e t z i c B hor i zon has p r i s -mat ic or columnar m a c r o s t r u c t u r e that breaks to hard to extreme-l y h a r d , b l o c k y peds with dark c o a t i n g s . They occur on s a l i n e parent m a t e r i a l s in some areas of the s e m i - a r i d to subhumid I n t e r i o r P l a i n s in a s s o c i a t i o n with Chernozemic s o i l s and to a l e s s e r extent with L u v i s o l i c and G l e y s o l i c s o i l s . Most S o l o n -e t z i c s o i l s are a s s o c i a t e d with a v e g e t a t i v e cover of g r a s s e s a n d f o r b s ; some occur under t r e e cover but i t is thought that the t r e e s d id not become e s t a b l i s h e d u n t i l s o l o d i z a t i o n was wel l under way. S o l o n e t z i c s o i l s are thought to have developed from parent m a t e r i a l s that were more or l e s s u n i f o r m l y s a l i n i z e d with s a l t s high in sodium. Leach ing of s a l t s by descending r a i n w a t e r p r e -sumably r e s u l t s in d e f l o c c u l a t ion of the sodium s a t u r a t e d c o l -l o i d s . The p e p t i z e d c o l l o i d s are a p p a r e n t l y c a r r i e d downward and d e p o s i t e d in the B h o r i z o n . F u r t h e r l e a c h i n g r e s u l t s in d e -p l e t i o n of a l k a l i c a t i o n s in the A h o r i z o n , which becomes a c i -d i c , and a p l a t y Ae h o r i z o n u s u a l l y d e v e l o p s . The u n d e r l y i n g B h o r i z o n u s u a l l y c o n s i s t s of d a r k l y s t a i n e d , f u s e d , i n t a c t colum-n a r p e d s . S t r u c t u r a l breakdown of the upper part of the B h o r i z o n a p p a r e n t l y o c c u r s at an advanced stage of development as e / c h a n g e a b l e s o d i u m i s leached d o w n w a r d . A t t h i s s t a g e , t h e i c l o n e t z i c B u s u a l l y breaks r e a d i l y to b l o c k y peds c o a t e d w i t h • A h i t e s i l i c i o u s p o w d e r . Complete d e s t r u c t i o n of t h e s o l o n e t z i c 2 h o r i z o n w o u l d b e t h e most advanced stage of s o l o d i z a t i o n . T h e r a t e of e v o l u t i o n through the ' s t a g e s ' of develoment d e p e n d s ..pon p r o p e r t i e s of t h e parent m a t e r i a l ( s a l t c o n t e n t , h y d r a u l i c c o n d u c t i v i t y ) i n a d d i t i o n to c l i m a t i c f a c t o r s . M o s t s o l o n e t z i c s o i l s in Canada have a n e u t r a l to a c i d i c A h o r i z o n i n d i c a t i n g that some s o l o d i z a t i o n has o c c u r r e d . S o i l s w i t h s t r o n g l y a l k a l i n e A h o r i z o n s , an e a r l y stage of S o l o n e t z i c s o i l f o r m a t i o n , are uncommon in Canada. As s o l o d i z a t i o n pro-c e e d s the h o r i z o n s of s a l t and l ime a c c u m u l a t i o n m o v e d o w n w a r d f r o m the B to the C h o r i z o n . In most S o l o n e t z i c s o i l s the s a t u -r a t i o n e x t r a c t of the C h o r i z o n has a c o n d u c t i v i t y of more t h a n 4 m i 1 1 i m h o s / c m . S o l o d i z a t i o n is a r r e s t e d where s a l i n e ground-w a t e r is w i t h i n c a p i l l a r y reach of the so lum, and res a 1 i n i z a t i o n m a y occur in groundwater d i s c h a r g e a r e a s . There are three great groups reco g n i z e d : 169 So 1 onet z lacks a continuous Ae > 2 cm Solodize d Solonetz Ae > 2 cm, i n t a c t columnar, Bn or Bnt Solod Ae > 5 cm, d i s t i n c t AB, d i s i n t e g r a t -ing Bnt Horizon Sequences Great Group Solonetz S o l o d i z e d Solonetz Solod Subgroup Brown Dark Brown Black A l k a l i n e Gleyed Brown Gleyed Dark Brown Gleyed Black Brown Dark Brown Black Dark Gray Gray Gleyed Brown Gleyed Dark Brown Gleyed Black Gleyed Dark Gray Gleyed Gray Brown Dark Brown Black Dark Gray Gray Gleyed Brown Gleyed Dark Brown Gleyed Black Gleyed Dark Gray Gleyed Gray Ah , Bnt, Csk Ah, Bn or Bnt, Csk Ah, Bnt, Csk "AT, "BnT Csk AT, TTrTgj, Cskgj AT, Bngj, Cskgj AT, Bntgj, Cskgj Ah, Ae, Bn or Bnt, Csk AT, AT, "BT or Bnt, Csk AT, AT, Jjnt or _B_n, Csk ATe, Ae, Bnt, Csk ATe, AT, TTnT, Csk Ah, Aegj, Bngj, Cskgj AT, Aegj, Bngj or Bn t g j , CsFg Ah, Aegj, B n t g j , Cskgj Ahe, Aegj, Bntgj, Cskg Ahe, Aegj, Bntgj , Cksg Ah, Ae, AB, Bnt, Ck, Csk AT, AT, AT, BTt , Ck, Csk ATe,~Ae ,~~A"B,~~BTt, Ck, Csk ATe, AT, AT, TTnT, Ck, Csk ATe, AT, AT, "BTT, Ck, Csk ATT Aegj ,~A"Bgj , Bntgj , ~~CsFg Ah, Aegj, ABgj, Bntgj , CsFg Ah, Aegj, ABgj, Bntgj , Cskg Ahe, Aegj, ABgj , Bntgj , C~skg Ahe, Aegj, ABgj , Bntgj , Cskg 170 Master Mineral Horizons Mineral horizons are those which contain less organic matter than that specified for organic horizons (30% organic matter). A - This is a mineral horizon or horizons formed at or near the surface in the zone of the removal of materials in solution and suspension, or a maximum in situ accumulation of organic matter, or both. Included are: 1) horizons in which organic matter has accumulated as a result of biological activity. 2) horizons that have been eluviated of any or all of clay, iron, aluminum, organic matter or all(Ae). 3) horizons having the characteristics of 1) and 2) above but transitional to B or C. 4) horizons markedly disturbed by pasture or cultivation (Ap). B - This is a mineral horizon or horizons characterized by one of the following: 1) an enrichment in silicate clay, iron, aluminum or humus, alone or in combination (Bt, Bf, Bfh, Bhf and Bh). 2) a prismatic or columnar structure that exhibits pronounced coatings or stainings and significant amounts of exchangeable Na (Bn). 3) an alteration by hydrolysis, reduction or oxidation to give a change in colour or structure from horizons abov. a horizon characterized by the removal of clay, iron, am or organic matter alone or in combination. It is used with A. f - a region enriched with hydrated iron. Used with B; B, h, g. g - a horizon characterized by gray colours, or prominent mottling, or both, indicative of permanent or periodic intense reduction h - a horizon enriched with organic matter. It is used alone; with A and e; with B; or with B and f. j - modifier of n and t to denote an expression of, but failur. e - a horizon characterized by the removal of clay, iron, aluminum or organic matter alone or in combination. It is used with A. f - a region enriched with hydrated iron. Used with B; B, h, g. g - a horizon characterized by gray colours, or prominent mottling, or both, indicative of permanent or periodic intense reduction 171 h - a horizon enriched with organic matter. It is used alone; with A and e; with B; or with B and f. j - modifier of n and t to denote an expression of, but failure to meet, the specified limits of the suffix it modifies (placed to the right). k - denotes presence of carbonates. m - a horizon slightly altered by hydrolysis, oxidation or solution, or all three to give a change in colour or structure or both - used with B. n - a horizon in which the ratio of exchangeable Ca to exchangeable Na is 10 or less. When used with B it must also exhibit: prismatic or columnar structure, dry coatings on ped surfaces, hard to very hard consitence when dry. s - a horizon with salts, including gypsum, which may be detected as crystals or veins, surface crusts of salt, plant stress, salt-tolerant plants. Used with C and k. sa - a horizon with secondary enrichment of salts more soluble than calcium and magnesium carbonates, where the concentration exceeds that present in the unenriched parent material. > 10 cm thick. ECe > 4 mmhos cm"1. Used with C. a horizon enriched with silicate clay. It is used with B. (Bt) alone and with other suffixes (Bnt) 172 APPENDIX 2. Training Site Data The raw data matrices for the 84 training sites used in the analyses described in this thesis are presented in the following pages. The data for each case number are listed in the sequence indicated below: Case#/Row#/E/ALF/SALT/GRASS/FORB/SED,'BLUE/GREEN/ , Red/pHW/pHCEC/BD/BIOM/Row^AVP/TN/ORGC/ ACa/AMg/AK/ECa/EMg/ENa/EK'iRow#/SAND/SILT/ CLAY/POPR/BRIN.TJlST/HOJU/POJLl/PUNU/ARFR/Ca:Na/y/ m/c/Code/R2/R3/R4/PR 1/ 173 1 . 0 0 1 . 0 0 9 0 5 . 4 5 1 . O O 4 0 . 0 0 0 . 7 6 8 . 8 7 7 . 38 0 . 8 1 1 179 . 2 7 4 . 5 0 11 . 11 1 . 0 6 4 . 75 1 1 . 1 1 0 . 0 13 . 0 0 5 . 0 0 0 . 0 0 . 0 131 . 0 0 1 5 3 . 0 0 1 . O O 0 . 9 0 o . 8 6 3 . o o 1 . 0 0 9 0 3 . 5 5 14 . O O 0 . 0 0 . 6 0 6 . 9 4 6 . 9 1 0 . 2 0 1 0 0 8 . 19 8 . 2 5 11 . 11 0 . 5 8 7 . 2 5 11 . 9 3 0 . 0 6 6 . 0 0 17 . 0 0 0 . 0 0 . 0 8 4 . 0 0 172 . 0 0 2 . 0 0 0 . 7 0 0 . 4 9 6 . 0 0 1 .00 9 0 6 . 16 0 .0 2 6 . 0 0 0 . 7 2 8 . 9 4 7 . 9 5 0 . 7 4 1 3 5 1 . 6 7 8 . 5 0 15 . 6 4 2 . 0 2 9 . 5 0 15 . 6 4 0 . 0 12 . 0 0 19 . 0 0 0 . 0 12 . 0 0 1 1 5 . . 0 0 167 . 0 0 1 0 0 0 8 0 0 . 6 9 8 . . 0 0 1 . 0 0 9 0 7 . 0 4 3 0 . 0 0 0 . 0 0 . . 5 9 7 . 6 6 7 . 2 9 0 . 2 5 1 127 . . 27 7 . . 0 0 12 . 5 2 0 . 5 1 7 . 2 5 1 1 . . 9 3 0 . , 0 34 . 0 0 21 . 0 0 0 . 0 0 . 0 6 7 , . 0 0 175 . O O 3 . 0 0 0 . 6 8 0 . 3 8 9 , 0 0 1 . 0 0 9 0 9 . 6 1 3 5 . 0 0 0 . 0 0 . 6 1 6 . 8 0 6 . 5 2 0 . 13 1 0 8 0 . o o 4 . . 2 5 9 . 4 6 0 . 9 7 6 . 5 0 10 . 7 0 0 . , 0 28 . 0 0 3 0 .OO 0 , . 0 0 . 0 6 5 . . 0 0 174 . 0 0 3 . 0 0 0 , 6 7 0 . 3 7 1 1 OO 1 . 0 0 9 0 4 . 7 1 2 5 0 6 0 . 0 0 0 . 8 5 6 . 7 7 6 4 0 0. . 12 1 2 6 1 . 16 8 2 5 9 . 4 6 0 . 8 2 8 2 5 9 , . 4 6 1 0 . 4 0 12 • OO 1 .OO 7 . , 0 0 2 3 . .OO 103 . . 0 0 163 , O 0 2 0 0 0 74 0 . 6 3 12 . 0 0 1 , .OO 9 0 5 . , 14 2 0 . , 0 0 4 1 . . 0 0 0 . 6 6 8 6 8 7 . . 4 0 0 . . 8 4 1 0 8 0 . 9 3 3 . 7 5 9 . 4 6 0 . 6 6 4 . 0 0 9 . . 4 6 12 . 6 0 14 . 0 0 11 . 0 0 1 . OO 2 3 . . 0 0 7 2 . 0 0 174 , 0 0 3 , OO 0 . 6 9 0 . . 4 1 15 . OO 1 OO 9 0 4 , 7 1 2 9 . OO 0 0 0 . 5 2 7 . . 13 6 . . 4 0 0 . 25 1 0 2 5 . OO 6 . 2 5 1 0 . . 7 0 0 . , 5 9 6 . 2 5 1 1 . 11 1 1 . 7 0 2 2 . 0 0 1 6 . OO 0 0 0 0 9 4 . 0 0 1 8 7 . . 0 0 1 , , 0 0 0 . 74 0 . . 5 0 16 . OO 1 0 0 9 0 4 . , 31 5 . 0 1 . OO 0 . 6 6 7 9 4 7 . 0 7 0 . 3 7 1 1 8 5 . 0 5 7 . 2 5 12 . , 7 6 0 . 5 7 8 . 0 0 1 3 . . 17 0 . 0 3 5 , 0 0 5 5 . OO 0 . 0 0 . , 0 107 . 0 0 168 . 0 0 2 . 0 0 0 . 77 0 . . 6 4 17 . OO 1 . . 0 0 9 0 3 . 8 9 18 . 0 0 0 . . 0 0 . 6 4 7 , . 0 6 6 . 6 3 0. 18 1 0 8 0 . 15 9 . 2 5 10 , . 2 9 0 . 8 2 9 . 2 5 1 0 . 7 0 0 . 0 4 2 . . 0 0 2 0 . 0 0 0 . 0 0 . 0 9 0 . OO 182 OO 2 . 0 0 0 . 74 0 . 4 9 18 . 0 0 1 . . 0 0 9 0 7 . 6 3 1 2 . 0 0 7 . 0 0 0 . 5 9 . 7 . , 12 6 . 7 5 0 . 15 1 158 . 61 9 . , 5 0 10 2 9 0 . 8 8 1 0 . OO 1 0 . 7 0 0 . 0 6 9 . OO 5 . OO 0 . 0 6 . 0 0 8 9 . 0 0 168 , 0 0 2 . 0 0 0 . 7 2 0 . 5 3 21 , OO 1 . . 0 0 9 0 3 . 7 7 14 . 0 0 2 8 . 0 0 0 6 1 8 . 0 7 7 . , 0 3 0 . 5 7 0 . , 0 9 . 2 5 12 , . 7 6 0 . 6 4 9 . 2 5 12 . 7 6 0 . 0 5 2 . 0 0 4 . 0 0 0 . 0 1 . 0 0 7 9 . 0 0 156 . 0 0 2 . 0 0 0 . 71 0 . 51 24 . 0 0 27 . 0 0 2 . 0 0 0 . 9 9 1 , 0 3 0 . 0 2 . 0 0 22 . 0 0 0 . 14 2 . 4 0 6 . 3 0 1 . 0 5 3 . 0 0 0 O 0 . 0 5 . 0 0 0 . 0 0 . 0 0 . 7 5 137 . 0 0 0 . 9 6 1 7 9 4 7 . 0 0 8 5 . 0 0 0 . 0 0 . 0 1 . 37 1 . 6 4 0 . 0 2 . 0 0 9 . 0 0 0 . 17 2 . 3 0 0 . 2 0 0 6 3 3 . 0 0 0 . 0 0 . 0 0 , 0 0 . 0 0 . 0 36 . 2 5 121 . 0 0 0 . 6 9 1 0 1 6 4 . 0 0 3 7 . 0 0 15 . 0 0 18 . 0 0 1 . 1 1 1 . 2 4 0 . 0 2 . 0 0 35 . 0 0 0 . 22 2 . 9 0 5 . 8 7 2 . 11 3 . 0 0 0 . 0 0 . 0 2 . 0 0 12 . 0 0 0 . 0 1. . 6 1 133 . 0 0 0 . 8 6 1 5 2 9 5 . 0 0 5 9 . 0 0 11 . 0 0 0 . 0 1. . 4 4 1 . 7 2 0 . 0 2 . 0 0 15 . 0 0 0 . 17 2 . 4 0 0 . 7 4 0 . 5 5 3 . 0 0 0 . . 0 0 . 0 0 . . 0 0 . 0 0 . 0 9 . 8 0 1 1 9 . 0 0 0 . 5 6 7 9 7 3 . 0 0 6 4 . 0 0 3 . 0 0 0 . 0 1. . 4 7 1 . 7 4 0 . 0 2 . . 0 0 22 o o 0 . 13 1 . 9 0 8 . 2 6 1 . 0 2 3 . 0 0 0 . 0 0 . 0 O . 0 0 . 0 0 . 0 0 . . 7 9 1 16 .OO 0 . . 5 6 7 5 4 0 , . 0 0 1 3 . . 0 0 0 . 0 0 0 . 0 0 1. . 2 2 1 . 16 2 9 8 . . 8 0 2 . 0 0 14 . 0 0 0 . . 15 2 . 10 0 . 18 0 . 8 2 3 . 0 0 4 6 . . 4 0 4 3 . 2 0 0 , . 0 3 0 . . 0 0 4 . 0 0 4 5 . 8 3 121 .OO 0 . 8 5 1 2 4 6 3 . . 0 0 31 . . 0 0 2 . 0 0 3 . 0 0 1. , 4 3 1 . 6 9 4 2 6 7 0 2 0 0 17 . 0 0 o , , 16 2 . 4 0 7 . 3 9 0 6 6 3 . 0 0 28 . . 9 0 58 . 5 0 0 . , 0 12 OO 0 . 0 0 . . 5 4 1 2 0 . 0 0 0 , 6 0 8 6 4 0 . 0 0 4 3 . OO 2 8 . OO 0 . 0 1. 4 8 1 7 6 3 6 1 . 6 0 2 . OO 4 0 0 0 . 18 2 . 5 0 1 . 0 4 0 5 9 3 . 0 0 2 6 . 3 0 6 2 0 0 0 . 0 0 . 0 0 . . 0 6 . .01 139 , OO 0 . 6 8 1 3 0 6 6 . 0 0 9 0 . 0 0 4 . 0 0 0 . 0 1 . 31 1 . 4 0 0 . 0 2 . OO 22 . 0 0 0 . 15 2 . 2 0 0 . 54 0 . 6 0 3 . . 0 0 0 . 0 0 . . 0 0 . 0 4 . 0 0 . . 0 14 . 81 1 3 0 . 0 0 0 . 8 2 1 3 9 1 0 . 0 0 6 5 . OO 17 . 0 0 0 . . 0 1 . 4 6 1 . 6 3 0 . 0 2 . 0 0 12 , , 0 0 0 . 17 2 . 2 0 • 0 . 24 0 . 8 4 3 . . 0 0 o. 0 0 . 0 0 . 0 0 . 0 0 , , 0 38 . 5 4 134 . OO 0 . 6 7 1 2 0 6 0 . 0 0 7 7 . 0 0 3 . . 0 0 1 . . 0 0 1 . . 2 7 1 . 51 0 . 0 2 . 0 0 5 . 0 0 0 . 18 2 . 3 0 0 . 18 0 . 8 9 3 . . 0 0 0 . 0 0 . 0 0 . 0 0 . 0 0 . . 0 5 5 . 5 5 121 , OO 0 . 74 1 0 7 6 9 . OO 5 6 . 0 0 2 . OO 0 , o 1 . . 4 8 1 , . 7 0 0 . 0 2 . 0 0 22 . o o 0 . 17 2 , . 2 0 2 . 15 0 . 6 4 3 . o o 0 . 0 0 . . 0 0 . 0 1 5 . 0 0 0 . 0 4 . , 3 0 1 10 , . 0 0 0 . 72 8 6 9 0 . 0 0 22 .00 1 . .00 902 0 54 7 , 76 7 10 .25 1 1 .52 0 0 .0 23 OO 26 83 .00 177 , ,00 3 26 OO 1 . .00 902 0 ,91 9. 81 8 4 .50 7 . 82 0 13 .40 23 OO 27 127. .00 146. .00 1 27 . OO 1 . OO 900 0. .74 7, 68 7 8 .50 1 1 . 93 0 0 0 54 . 00 16 81 . .00 127 . 00 2 29 . o o 1 . .00 903 0. .60 9. ,50 8 4 . 25 12 . 3 4 2 0. .0 0. ,0 0 73. ,00 170, ,00 1 31 00 1 . 00 906 0. 69 7 . .43 7 6 . . 25 10. . 70 0 0. 0 58 . 00 0 168 . o o 175 . 00 1 33 . .00 1 . 00 915 0. .62 6 89 6 10. 00 7 . 4 1 1 0. 0 51 . ,00 4 111, o o 174 . .00 2 34 . 00 1 . ,00 906 0. .55 6 . .93 6 9 . . 50 10 . 29 0 0. ,0 28 00 14 87 . OO 179. ,00 2 35. .00 1 .00 910 0 . 54 9 . 39 8 9 . .50 13 . 17 1 0. O 15 ,00 18 97 .00 175 .00 1 36 .OO 1 .OO 916 C . .51 7 . .58 6 7 .50 1 1 . . 1 1 0 8 .90 17 .00 13 87 , .OO 178. .00 2 38 .00 1 . o o 907 0 . 50 7 . 35 7 7 .50 1 1 . 1 1 0 0 .0 19 . o o 16 81 .00 180 .00 3 40 .00 1 .OO 914 0 .52 8 .98 7 6 .OO 15 . 22 1 0 .0 26 . o o 2 89 .OO 177 .00 1 4 1 .00 1 . o o 918 0 .60 7 .40 7 6 .50 4 .94 0 2 .30 24 .00 32 85 .00 182 .OO 3 13 4 1 . 00 2 , 00 03 0 37 1057 , 20 50 10. ,75 12 . 35 OO 0. 0 2 . 00 o o 0. 71 0. ,47 97 1. OO 5. OO 64 2. 15 1020 90 92 5. OO 9. 88 OO 0. 0 3. 00 00 0 92 0. 87 68 21 . OO 5 . OO 38 0. 23 1 100. 08 95 10. .00 12 . 34 OO 0. 0 4 . OO 00 0. 76 0. 64 26 0. 0 65 . 00 29 0. 92 1220. 80 56 4 . 25 13 . 17 0 0. .0 29. 00 00 0. 68 0. .43 31 1. OO 1 . 00 35 0. .42 1086 . , 17 84 6 . 75 1 1 . ,93 0 0. 0 0. 0 00 o , 93 0. 96 01 36. .00 0. 0 64 0. . 13 1 163 . , 43 06 12 . •25 9 . 47 00 0. .0 0. .0 00 0 .80 0, .64 09 14 . ,00 8 . 00 63 0. . 16 1 127 , 56 64 9 . 50 9 , .47 OO 2 .OO 0. .0 00 0. .77 0. ,49 45 56, .00 3 .00 03 0 .91 1 123. , 26 51 9 .50 13. . 17 OO 3 .OO 0. .0 00 0 . 77 0. .55 16 47 .OO 3. .00 90 0 . 16 1264 . 30 75 8 .25 1 1 .52 OO 3 OO 0 .0 00 0 .72 0. .49 63 58 .00 0 .0 00 0 . 16 867 .88 73 7 .50 1 1 . 1 1 OO 0 .0 0 .0 00 0 .75 0 .45 91 27 .OO 3 .00 53 0 .50 1259 .69 47 6 .00 15 .64 00 3 .00 0 .0 00 0 . 73 0 .50 57 28 .00 0 .0 56 0 . 15 1 185 . 13 60 10 .OO 8 .23 00 0 .0 0 .0 .00 0 .72 0 . 47 55 . 00 2 . 00 0 0. 0 2 . 00 11 0. 46 0. 60 3 0. 0 0. 0 1 0. 6610375. 00 51 . 00 6 . 00 0 0. 0 2. .00 1B 15 . 20 1 . . 14 3 30. O 0. 0 0. 0 9 5 ' 17018 00 70. OO O. 0 0 351 . 60 2. .00 10 0. 17 0, 95 3 0 . 0 0. 0 5 0. 84 7857 . .00 22 . OO 5 , o o 3 0 . 0 2 ,00 23 6 . 09 2 .68 3. 3. 00 30. ,00 . 2 0. 63 8395. 00 65 . 00 5 . CO 21 0. 0 2, .00 10 1 . 74 0 .97 3 0. .0 0 .0 0 1 . 0327384. .00 55. 00 3, .00 0 0. 0 2 .00 18 0. 13 1 , .39 3 0. 0 0. .0 0 0, ,8015429 .00 43. 00 8 .00 27 0. 0 2 o o 9 0. , 19 0 .64 3 0. ,0 6 .00 0 0. 63 12006 .00 33. ,00 2 .00 6 684 , ,50 2 .00 42 8 70 1 .60 3 0 .0 0 .0 0 0 ,7212998 .00 30, .00 4 .00 0 0. O 2 • OO 7 0, .63 0 .90 3 0 .0 0 .0 0 0, .671 1223 .OO 35 .00 0 .0 7 0 .0 2 .00 12 0 .72 0 .81 3 0 .0 0 .0 0 0 .6010935 .00 28 .00 29 .OO 13 568 .00 2 .00 22 1 .65 1 .79 3 0 .0 0 .0 0 0 .68 1 1570 .00 56 .OO 15 .00 0 0 .0 2 .00 10 0 . 15 0 .96 3 0 .0 0 .0 0 0 .6511135 .00 0 1. 48 1 . 76 00 0. 13 . 1 . 90 00 0 . 0 0. 0 00 23. 37 125. 00 0 1. 2 1 1 . 23 00 0. 10 1 . 40 00 27 . 60 59 . 00 0 0 . 33 134 . 00 o 1. 53 1 . 73 00 0. 13 1 . 70 00 0. 0 0. 0 00 58. 82 97 . 00 00 1. 62 1 . 85 00 0. 20 2. 70 o o 0 . 0 0 0 00 o . 70 1 15. 00 o o 0. 78 0. 83 00 o . 15 2. 20 o o 0. 0 0. 0 0 3 . 88 163 . 00 0 1. 07 1 . 30 o o 0. 22 2 . 90' 00 0. 0 0. 0 0 94. 23 139. 00 00 1. ,27 1 . 60 o o 0. 15 2 , o o o o 0. ,0 0. .0 0 50. .00 138 . o o 00 1 , .28 1 .61 o o 0 . 24 3 .40 00 0 .0 0 .0 o 1 .09 134 .00 0 1 .21 1 .53 o o 0. .08 2 . 30 00 29 .40 61 .70 0 • 13 .09 129 .00 o o 1 .27 1 .63 o o 0 . 15 1 .90 00 0 .0 0 .0 0 10 .31 135 .00 00 1 .28 1 .64 00 0 . 14 1 .70 00 0 .O 0 .0 0 3 .64 130 .00 0 1 .20 1 .48 00 0 . 18 2 .60 .00 47 . 10 50 .60 .0 4 1 .66 131 .00 175 42 00 1 .00 922 . 35 19 .OO 0 .0 0 . 74 8 .75 7 .41 0 . 37 1066 . 14 5 .50 10 .70 0 .38 7 . 75 15 .64 O .0 (S3 .00 10 .00 0 .0 0 .0 107 .OO 191 .00 1 .00 0 .76 0 .56 43 .00 1 ,00 912 .06 52 .00 0 .O 0 .54 8. 86 7 .40 0 .65 1 164 .23 4 .25 6 58 O. 58 5. .75 9 88 0 .0 6 .00 42. 00 0. .0 0 0 62 .OO 175 .OO 3 .oo 0. .68 0 .35 44 .OO 1 .00 913. 22 35 .00 17 .00 0 .81 7. .40 6 .76 0 . 15 916 .57 B .75 10 .70 0 .51 9 .25 1 1 . . 1 1 0 .0 38 .00 7. .00 8 .00 8 .00 91 .OO 187 .00 2 . .OO 0 . 74 0 49 45 .00 1 .00 914 ,53 59 .00 0 .0 0 .52 7 . 15 6 63 0 . 13 1096 .25 10 .50 9 .05 0 82 11 .00 9 .05 0 .0 30 OO 11 . .OO 0 .0 0 .0 79. .00 187 .00 3. .00 0 .72 0 .42 46 . OO 1 .00 919 .57 54 OO 1 . .00 0. .49 7. . 14 6 . 6 1 0 . 12 1186, .27 9 .75 8 64 O. 65 9 .75 9, .05 9 . . 10 9. OO 17 . OO 1 .00 0 .0 80. 00 181 , .00 3. OO 0. 72 0. .44 49 . .OO 1 . .OO 912 . 86 0 .0 50, .00 0, 84 9 28 8. ,30 0. 82 1123 .50 5, 38 15. .23 1 . 56 5 .38 16 .05 6. ,30 9. .00 1 . OO 9 OO 14 , OO 1 15 . 00 149 . .00 1 . OO 0 .85 0. .77 53 , OO 1 .OO 916 . 68 60, OO 0. 0 0. 53 7 .40 6. ,82 0, . 16 1264 78 10. 63 10. 49 1 . .01 10. 63 10. 49 0. ,0 20, OO 5 . 00 0 ,0 o. .0 84 . ,00 186. 00 .3 • ,00 0. .73 0. .45 54 . OO 1 . oo 903 . 86 42 . .00 0 0 0. 67 7. 15 6. 25 0. 21 1042. 73 10. OO 9 . 88 O. 50 10 25 10. 29 0. 0 22. .00 30. 00 0. ,0 0. .0 52. 00 .138 . .00 3. 00 0. 63 0. .38 56 . OO 1 . oo 904 . 51 26. OO 0. ,0 0. ,61 7 . .56 6. 40 0. . 13 1042. ,73 9 . 25 9 05 0. 73 9. . 25 9. 05 0. 0 43. OO 7 . OO 0. 0 0. .0 5 1 . OO 150. OO 3 . OO 0. 62 O. 34 57 . 00 1 . ,00 898 . 27 12. 00 0. 0 0. 62 7 . ,45 6. 21 0. , 14 1150. 23 6 . OO 6 . 99 0. 43 8 . 25 10. 29 0. 0 30, .00 25. 00 0. 0 0. .0 54 . CO 158 . OO 3 . 00 0. ,65 0. 34 58 . 00 1 . 00 895. 53 3. ,00 0. 0 0. 72 7 . . 32 6 . 20 0. 18 1 150. 23 10. ,75 8. 23 0. 58 1 1 . ,50 9. 05 0. 0 81 . 00 4 . 00 0. 0 0. 0 54 . OO 130. ,00 3. 00 0. 66 0. 42 59. 00 1 . 00 900. 01 0. ,0 79. 00 0. 80 9. . 13 8 . 38 0. 15 1487 , .22 3. 63 17 . 28 2. 24 3. 63 17 . ,70 0. 0 3 . OO 1 . OO 16 . 00 16 00 109 . ,00 131 . ,00 1 . 00 0. 87 0. 83 73 .00 6 .00 0 .0 1 . 14 1 .31 0 .0 2 .00 8 .OO 0 . 12 1 .60 2 . 28 0 .70 3 .OO 0 .0 0 .0 0 .0 0 .0 0 .0 3 .40 145 .00 0 .7415515 .00 48 .00 0 .0 0 .0 1 .42 1 .81 0 .0 2 .00 22 .00 0 . 17 2 .80 6. 36 0 . 78 3 .OO 0 .0 0 .0 0 .0 0 .0 0 .0 0 .83 1 19 .00 0 .52 7378. .OO 46 .00 2 CO 0. .0 1 . 17 1 .30 0 .0 2 .00 6 CO 0 . 14 1 .80 0 .26 0 .59 3 .00 0 .0 0 .0 1 .00 0 .0 0 .0 34 .58 138 .00 o .6612558 .00 41 .00 0 .0 0 .0 1. 27 1 .59 0 .0 2 .00 9. OO 0. . 16 2 . 10 0 . 17 0 .86 3. .00 0. 0 0 .0 0 .0 0 .0 0 .0 64 .70 134 .OO 0 .5910586 .00 27 .OO 2 .00 0 .0 1. .30 1 .64 550 .50 2, .00 6. .00 0. , 15 2 .00 0 . 15 0 .72 3 ,00 26. . 20 64 .70 0. .0 0. .0 0, .0 65. ,00 130 .00 o. 6210400 .00 14 , .00 13 .00 16 , oo 1 . oo 1 . 10 206 .00 2 .00 10. .00 0, . 11 1 .60 7 .39 1 .74 3 .00 33 ,80 59 .90 8 .00 16 .OO 0. .0 0. .73 127 .00 0. .9114605 .OO 25. OO 3 .OO 12 . .00 1 . , 26 1 .63 0 .0 2 .00 15. .00 0 . 18 2 .40 0 .26 1 . . 12 3 .00 0. ,0 0 .0 0, .0 0 .0 0. 0 40. 8 8 135 .OO 0. ,6211340. .00 52 . OO 6 .00 0. .0 1 . . 34 1 .62 0. .0 2. .00 13. .00 0. , 16 2 .20 0. . 17 0, , 54 3. .00 0. .0 0 .0 c. ,0 0 .0 0. ,0 60. , 29 87 .00 0. .60 4524 .00 50. OO 0. .0 23 CO 1 , 39 1 .70 325. 80 2 .00 8. ,00 0, , 14 1 .80 0. 13 0. .77 3. oo 0. 0 0 .0 0. .0 0. .0 0. 0 7 1 . ,51 93 .00 0. 55 4743 , ,00 55 . 00 4 . 00 29 . 00 1 . ,42 1 . 72 0. 0 2, .00 12. 00 0. . 16 2 .20 0. 17 0. .63 3. 00 0, 0 0 .0 0. .0 0. 0 0. 0 48. .53 102. .00 0. 53 5508 . oo 85. 00 5. ,00 0. 0 1 . . 10 1 .41 0. 0 2 . ,00 9 . oo 0. 21 2 . 70 0. 17 0. ,65 3. oo 0. .0 0, ,0 0. 0 0. 0 0. 0 67 . .65 86 .00 0. 63 4644 . OO 4 . 00 0. .0 5. 00 1 . ,25 1 , .41 241 . . 10 2 . ,00 15. oo 0. 22 2 .60 6. ,52 2. 28 3 . 00 0, ,0 0, .0 7 . 00 40, OO 0. .0 0. ,56 114. OO 0. ,9612426. .00 176 60 . 00 1 .00 903 0 .61 7 .44 6 7 .OO 9 . 88 0 0 .0 75 .00 17 68 .00 162 OO 3 62 .00 1 .00 897 0 .61 7 .49 6 7 .00 10 .70 0 0 .0 32 .00 25 65 .00 163 .00 3 63 o o 1 .oo 895 0 64 8 .55 7 4 . 25 11 . 1 1 O 0 .0 33 .00 0 79 oo 156 oo 1 65, .00 1 .00 901 0. .88 9 .93 8 4 . 75 7 .41 1 7 , .30 7 . 00 12 112, ,00 132 .00 1 67 , ,00 1 . .00 897 0, 49 7 . 48 7 9. , 75 9 47 1 0. .0 48 . 00 6 59 , 00 172 . OO 3 68 OO 1 .00 896 0. 54 7 . 36 6 6 . 25 10. 29 0 0 . ,0 79 . .00 0 73 . 00 172 .00 2 70. OO 1 .00 901 0. 99 9 .92 9 3. ,50 5 . 76 2 0 . ,0 28 .00 2 155. ,00 152 .OO 1 72. ,00 1 . .00 896 0/60 7 . 68 6 5. .00 9 .47 0 0. .0 25. 00 25 96 . .00 162 .00 2 73 . .00 1 .00 894 0 . 58 7 .06 6 7 . OO 9 . 47 0 0 . .0 30 .OO 20 87 , .00 167 .00 2 76 . 00 1 .OO 902 0, .61 7. .29 7 6 .00 10 .70 0 0, .O 29 .00 28 94 .00 164 .00 2 79 , .00 1 .OO 892 0 .64 7 .45 6 5 . 75 10 . 70 0 8 .20 69 .00 0 108 .OO 156 .00 1 80 .00 1 .00 892 0 .76 9 .97 8 3 .75 8 .64 0 10 . 0 0 2 1 . 0 0 8 74 .00 168 .00 1 25 7 . 00 0. 0 87 0 . 15 1241 .72 50 8 . 25 1 1 . 1 1 00 0. 0 0. .0 00 0. 65 0. .42 03 2 . 00 0. ,0 74 0. . 15 1 105 . 99 50 7 . 50 1 1 , ,52 00 0. 0 0 . .0 00 0, .67 o . 40 90 4 . 00 2 7 . OO 45 0 . 63 1 170. ,00 66 4 . 75 12. . 34 0 24. 00 1 .00 OO 0. 71 0 . 51 69 3 . 00 76 . OO 83 1 . .60 1419 . 62 47 5. 25 9 .05 00 0, 0 9 00 oo 0, 89 0, .85 76 12. ,00 0 ,0 06 0, ,20 1 124 . 18 10 10. 13 10. 08 o o 0. O 0. .0 00 0 . 64 0. .34 50 15 . 00 0 .0 87 0. 38 1070 . 18 35 7 . OO 11. . 1 1 0 0 . ,0 0 . 0 00 0. .70 0 . 42 17 1. OO 52 .OO 03 2 .20 1589 .29 81 3 . 50 7 . 82 00 8 ,00 1 , .00 OO 0. ,96 1 . .02 95 8 . OO 24 . .00 81 0. .55 1437 . 50 50 5. .75 10. 70 00 0. 0 0 . .0 00 0. . 76 0 .59 22 5 .0 14 .00 62 0. . 15 1215, .OO 73 8 .25 10 . 29 OO 0. .0 0 .0 00 0 .73 0. .52 53 38 . OO o .0 00 0. 21 1 137, .04 49 8 OO 12 .35 OO 0. .0 0 .0 00 0. , 76 0 .57 75 3 . 00 0 .0 62 0. . 14 1232 , .27 73 7 . 00 1 1 .93 0 0 .0 26 .00 00 0 .79 O .69 05 0 .0 47 .OO 95 1 . 3 0 1478 .99 66 5 .OO 1 1 . 52 O O 0 . 0 13 .OO OO 0 . 70 0 . 44 93 00 0 0 0 0. .0 2 . 00 14 0 . 30 0. 58 3 0 0 0. 0 0 0. .64 7208 . ,00 60. .00 9 ,00 29 0 .0 2 .00 12 0. . 39 0 ,58 3 0. 0 0. .0 0 0 . .59 7 150. 00 36 .00 18 .00 15 0. .0 2 .00 22 5. .00 0 .77 3 0. 0 2 .00 0 0. 72 8690, ,00 19. .00 0, .0 0 0 o 2 ,00 35 11 . 96 1 , .60 3 0, ,0 67 .00 0 0. .9613104 oo 55 00 33 oo 0 0, .0 2 . oo 14 0. .22 1 .23 3 0 ,0 0 .0 0 0, .54 6490 OO 79. ,00 6 00 0 0. ,0 2. OO 9 0. .98 0. . 42 3 0. .0 0 .0 0 0. .60 8833 OO 30 .00 6 .00 1 1 0, .0 2 .00 18 12 .83 2 .98 3 0. .0 43 .00 0 1 .0622630 .00 61 .00 7 .00 0 0 .0 2 .OO 6 1 . . 74 0 .55 3 12 .00 12 .00 0 0 .7811808 .00 54 .00 5 .00 22 0 .0 2 .00 15 0 .20 0 .93 3 14 .00 0 .0 0 0 .7110614 .00 59 .00 3 .OO 0 0 .0 2 .00 12 0 . 20 0 .64 3 0 .0 0 .0 0 0 .76 1 1656 .00 87 .00 4 .00 0 386 .70 2 .00 12 0 . 4 1 0 .90 3 0 .0 0 .0 0 0 .8713392 .00 29 .00 19 .00 0 315 . 10 2 .00 12 13 . 26 0 .84 3 5 .00 29 .00 0 0 .63 8658 .OO 0 1. 48 1 . 80 00 0. 16 1 . 80 00 0. 0 0 . .0 0 27 . 50 106 . 00 00 1. 01 1 , 34 00 0 . 18 2 , .40 00 0. .0 0 . 0 0 19. 23 1 10. 00 00 1. 03 1 , ,30 00 0. , 13 1 , .60 00 0 . 0 0, .0 0 0, .95 1 10, 00 0 1. . 17 1 , ,23 OO 0. .09 1 , ,00 OO 31 . .50 61 . , 20 0 0 . 44 1 17 . .00 0 1. 10 1 , 56 00 0. , 27 3 . 30 00 0, .0 0 . 0 0 46 . .04 1 10. 00 0 0, 9 9 1 . . 35 00 0. 16 1 . .80 00 0. 0 0. .0 0 7 . 14 121 . .00 00 1. 22 1 . 25 00 0. ,07 0 .80 00 0. .0 0 .0 0 0. .27 146 .00 0 0 .93 1 .23 oo 0. , 13 1 .80 00 0 0 0 .0 0 3 . 30 123 .00 oo 0 .94 1 . 27 00 0 •15 2 .00 00 0 .0 0 .0 0 4 1 . 25 122 .OO 0 1 .OO 1 . 28 00 0 . 1 1 1 .40 oo 0 .0 0 .0 0 40 .00 124 .00 0 0 .83 1 .23 00 0 . 17 2 .20 00 32 .40 59 .40 0 17 .07 124 .00 0 0 . 89 1 .03 00 0 . 1 1 1 .60 oo 26 .00 64 . OO 0 0 .38 1 17 .00 82 .00 1 . 0 0 9 0 4 . 9 9 0 56 7 . 5 2 6 . 7 3 3 . 7 5 9 88 0 . 37 7 . 6 0 64 . 0 0 21 . OO 74 . OO 174 . 0 0 3 . 0 0 8 5 . 0 0 1 . 0 0 9 01 , 9 8 0 . 8 3 10 . 24 9 10 3 . 0 0 4 . 12 2 . 3 7 12 8 0 0 0 0 . 0 158 . OO 1 6 0 . . 0 0 1. OO 8 6 . 0 0 1 . . 0 0 9 0 3 . . 9 6 0 . 6 2 7 . 6 6 6 82 3 . 8 8 7 . 82 0 , , 4 5 1 1 . 8 0 - 62 . 0 0 35 , 0 0 71 . OO 159 . 0 0 3 . 0 0 87 . . 0 0 1 , 0 0 9 0 4 6 2 0 . ,57 7 . 3 3 6 . 3 5 7 . , 5 0 9 0 5 0 . 6 5 8 . 2 0 12 . 0 0 35 OO 72 . 0 0 164 . 0 0 3 OO 8 9 . OO 1 , . 0 0 8 9 7 . 58 0 , , 6 3 7 . .71 7 . , 2 0 8 . 2 5 9 . 88 0 . , 7 5 0 . , 0 25 0 0 44 . 0 0 1 0 9 . OO 157 . . 0 0 2 . OO 9 0 . . 0 0 1 , . 0 0 8 9 3 . 2 8 0 . 6 2 7 . . 5 8 7 . , 0 2 7 . , 5 0 9 . 47 0 . . 74 0 . 0 76 . OO 3 . . 0 0 81 0 0 171 . . 0 0 1 , . 0 0 92 . oo 1 . OO 9 0 3 . 9 0 0 . , 91 9 . 4 0 7 , 9 6 5 . , 0 0 9 47 0 . . 79 0 . , 0 16 . 0 0 0 . 0 127 . oo 148 . . 0 0 1 . 0 0 9 3 . , 0 0 1 . 0 0 9 0 4 . 15 0 . . 72 8 . 57 7 . 3 5 2 . 5 0 10 . 7 0 0 . 5 5 0 , . 0 16 . 0 0 37 . 0 0 109 . OO 154 .OO 1 .OO 9 5 . . 0 0 1 . 0 0 9 0 4 . 5 2 0 , , 5 6 7 . 57 6 . 8 0 8 . , 2 5 7 .4 1 0 . 6 6 0 . .0 66 . 0 0 8 OO 9 3 .00 174 . 0 0 2 . 0 0 9 8 . 00 1 . 0 0 9 0 4 . 0 8 1 . 14 8 . 6 5 7 . 35 6 . 0 0 8 . 2 3 1 . 10 0 . O 45 .OO 0 . 0 152 , . 0 0 165 . 0 0 1 .OO 9 9 .oo 1 . 0 0 9 0 2 . 8 4 0 . 8 6 10 . 4 0 9 . 2 2 5 . 5 0 2 . 3 0 0 . 5 5 O . 0 0 . 0 0 . 0 1 6 0 . 0 0 158 . 0 0 1 . 0 0 100 .OO 1 . 0 0 9 0 5 . 2 3 0 . 7 5 7 . 5 6 7 . 2 5 6 .OO 9 . 8 8 0 . 4 7 0 . 0 39 . 0 0 18 . 0 0 100 .OO 164 .OO 2 .OO 15 . 0 0 0 . 0 8 5 . 0 0 0 0 . 22 1302 . 9 0 0 . 0 2 4 . 75 1 1 . 9 3 1 . 0 4 0 0 . . 0 0 . 0 0 . 0 0 0 . 7 0 0 . 4 3 0 . 6 1 8 9 5 4 0 . 0 0 . 0 9 2 . 0 0 o 3 . 6 0 0 . O 2 5 9 . 7 0 2 3 . 2 5 5 35 21 . 52 2 3 8 . 0 0 25 0 0 1 . 0 0 27 1 . 0 3 0 . 9 9 0 . , 9 6 2 6 0 7 0 1 . 0 0 0 0 9 7 . 0 0 0 0 , . 2 3 1O04 9 8 4 9 5 . 7 0 2 4 . 75 10 . 2 9 2 ,41 0 0 . 0 0 . 0 0 0 0 0 . 6 9 0 . 4 5 0 . , 6 5 7 8 1 0 8 0 0 10 , . 0 0 47 , . 0 0 0 0 . 12 1 196 . 3 5 3 6 3 . , 5 0 2 8 25 9 . 8 8 0 , , 19 0 0 . 0 0 . 0 0 . . 0 10 0 . . 7 0 0 . . 44 0 6 3 8 2 8 0 9 . 0 0 0 . . 0 7 4 . . 0 0 17 0 . 18 1085 . 27 0 0 2 9 . 5 0 1 1 . 5 2 0 . 26 0 0 . 0 0 . 0 0 , . 0 0 0 . . 8 3 0 6 9 0 , 8 3 1 4 2 7 9 18 . . 0 0 0 . 0 7 9 . 0 0 3 0 . . 13 1 157 . 5 0 0 . 0 2 8 , 25 10 . 2 9 0 , . 15 0 0 , 0 0 . . 0 O . 0 0 0 . . 7 2 0 . 47 0 . 6 6 9 9 6 3 0 . 0 4 3 , 0 0 16 . 0 0 2 1. . 3 0 1224 . 62 2 0 8 . 10 2 5 . . 7 5 1 1 . . 1 1 7 , . 8 3 0 1 . 0 0 1 1 . 0 0 9 . 0 0 2 2 0 .91 0 . 8 6 0 . 9 5 1 7 0 1 8 3 . . 0 0 9 . 0 0 54 . 0 0 13 0 . 3 3 1 159 . 91 0 . 0 2 3 . 0 0 1 1 . 5 2 4 . 3 5 0 0 . 0 0 . 0 8 . 0 0 0 0 . 8 0 O . 71 0 . 8 9 1 3 4 0 7 25 . 0 0 0 . 0 74 • OO 1 0 . 10 1 190 . 76 O . 0 2 9 . 7 5 9 . 0 5 0 . 15 0 0 . 0 0 . 0 0 . 0 0 0 . 7 5 0 . 5 3 0 . 7 2 1 2 0 9 0 2 . 0 0 29 . 0 0 47 . 0 0 14 0 . 8 5 1 138 . 9 3 0 . 0 2 7 . 0 0 9 . 4 7 4 . 5 7 1 0 . 0 13 .OO 9 .OO 7 0 . 9 2 0 . 9 2 1 . 0 0 2 3 1 0 4 0 .O 6 9 . 0 0 0 . 0 0 4 .OO 1 2 3 0 . 6 4 0 . 0 2 5 . 5 0 3 . 4 2 22 . 3 9 0 28 .OO 14 OO 1 .OO 26 1 . 0 4 1 . 0 1 0 . 9 8 2 6 2 4 0 3 9 . 0 0 0 .O 57 .OO 0 0 . 3 0 1 0 0 0 . 0 0 0 . 0 2 7 . 0 0 11 . 1 1 0 . 7 4 0 0 . 0 0 . 0 0 . 0 0 0 . 7 6 0 . 6 1 0 . 8 0 1 2 5 0 0 0 0 . 0 1. 1 1 1 . 46 OO 8 . 0 0 0 . 13 1 . 6 0 5 0 3 . 0 0 37 . 0 0 5 5 . 4 0 0 0 . 0 4 . 57 121 . 0 0 0 0 0 0 . 0 0 . 82 0 . 94 OO 18 oo 0 . 13 1 . 6C 48 3 . 0 0 3 3 . 7 0 5 3 . 5 0 0 0 0 . . 0 0 . 15 165 . 0 0 OO 0 0 . 0 1. 14 1 . 47 0 0 14 . oo 0 . 19 2 . 5 0 56 3 . 0 0 1 8 . . 4 0 6 9 8 0 0 0 . 0 1. 97 110 . OO 0 0 0 3 2 . 0 0 1. 12 1. 4 9 OO 17 . 0 0 0 . 2 0 2 . 6 0 77 3 . . 0 0 26 . 10 6 5 . 7 0 0 0 0 . , 0 3 . 42 115 . OO 0 0 OO 0 . , 0 0 . 8 0 1 . 01 0 0 14 . , 0 0 0 . , 17 2 0 0 95 3 . . 0 0 0 . 0 0 . 0 0 0 . 0 3 6 . 54 131 . , 0 0 0 0 0 0 0 , . 0 1 . 26 1 . 4 6 0 0 . 8 . 0 0 0 . 12 1 6 0 9 0 3 . 0 0 0 . , 0 0 . 0 0 0 . 0 55 . OO 123 . OO OO 0 0 21 . 0 0 1 . 0 8 1 . 18 0 0 23 . 0 0 0 . . 13 1 . 7 0 9 6 3 oo 0 . . 0 0 0 0 0 17 oo 0 . 7 3 134 . 0 0 0 0 oo 0 . 0 1 . 14 1 . 3 3 0 0 14 .oo 0 . 1 1 1 . 4 0 7 0 3 . 0 0 0 . 0 0 . 0 0 12 . 0 0 0 . 6 9 123 . . 0 0 0 0 0 0 0 . 0 1 . 2 2 1 . 5 5 0 0 10 .oo 0 . 15 2 OO 9 0 3 . 0 0 0 . 0 0 . 0 0 1 . 0 0 64 . 6 7 130 . 0 0 0 0 0 0 8 . 0 0 1 . 2 8 1 .31 0 0 15 .oo 0 . 17 2 . 3 0 29 3 . 0 0 0 . 0 0 . 0 0 0 8 . 0 0 1 . 5 3 152 .OO 0 0 0 0 . 0 0 . 8 0 0 .91 OO 14 .oo • 0 . 0 5 0 . 5 0 72 3 . 0 0 0 . 0 0 . 0 OO O . 0 o . 2 5 164. OO 0 0 0 O . 0 1 . 15 1 . 3 5 OO 1 1 .oo 0 . 18 2 . 4 0 6 0 3 .oo 0 . 0 0 . 0 0 0 . 0 9 . 4 6 125 . 0 0 0 0 178 101 . o o 1 . 0 0 9 0 5 . 7 1 4 0 . 0 0 0 . 0 0 . 6 6 7 . 3 3 7 . 0 2 0 . 15 1 157 . 38 - 7 . 5 0 9 . 0 5 0 5 8 8 . 5 0 10 . . 2 9 0 . 0 4 7 .OO 11 .OO 0 . 0 0 O 78 .OO 178 . 0 0 3 . 0 0 o . 7 1 0 . 44 102 . 0 0 1 . 0 0 9 0 4 . 3 2 5 0 . 0 0 0 . 0 0 6 8 7 . 13 6 6 3 0 12 1327 , 41 8 . 2 5 8 . 2 5 0 . 6 3 9 .OO 9 . 4 7 O . 0 18 . 0 0 15 .OO 0 . 0 0 . 0 8 3 . 0 0 171 . 0 0 3 . 0 0 0 . 7 3 0 . 4 9 103 . 0 0 1 . 0 0 9 01 . 6 0 26 . 0 0 5 . 0 0 0 64 7 . 3 0 6 . 6 8 0 . 16 9 2 3 . 3 3 8 . 5 0 9 . 0 5 0 . 7 3 9 . 2 5 9 . 8 8 0 . 0 7 . 0 0 54 . 0 0 5 . 0 0 0 . 0 9 7 . OO 158 . 0 0 2 . 0 0 0 , . 78 0 . 61 1 0 6 . . 0 0 1 . 0 0 9 0 4 . 2 3 2 0 . , 0 8 . 0 0 0 78 6 . 9 5 5 . 8 4 0 . 14 1180 . 9 5 5. . 7 5 9 . 0 5 0 . 3 8 6 . 2 5 10 . 2 9 0 O 33 . 0 0 32 .OO 7 . 0 0 0 , o 9 9 OO 175 .OO 2 . 0 0 0 .71 0 . , 57 109 . . 0 0 1 . 0 0 901 . 38 3 . 0 0 0 . 0 0 . . 74 7 . 61 6 . 8 5 0 . . 28 1082 . 5 3 5 . . 5 0 9 . 0 4 0 . 6 5 6 . 2 5 10 . 29 9 , 2 0 8 5 . 0 0 6 .OO 0 0 0 . 0 7 6 . 0 0 167 . 0 0 2 . 0 0 0 . . 72 0 . . 4 6 1 1 1 . OO 1 .OO 8 9 8 . 4 9 15 o o 2 .OO 0 . 6 0 7 . 26 6 . 6 9 0 , . 14 1286 . 6 9 9 . . 2 5 8 64 0 . 5 0 10 . 0 0 9 88 0 . . 0 2 3 OO 56 OO 0 . . 0 0 . . 0 4 5 . . 0 0 164 . 0 0 3 . 0 0 0 .62 0 . . 2 7 1 1 2 . OO 1 , .oo 904 . 34 6 . oo o O 0 . , 0 6 83 6 . 4 3 0 , . 13 1114 , . 8 3 8 . . 2 5 7 . . 8 2 0 . 7 3 10 . 0 0 9 . , 0 5 0 . , 0 4 9 . 0 0 44 . 0 0 0 . 0 0 , , 0 9 7 . OO 177 . OO 2 . 0 0 0 . . 7 3 0 55 1 14 . OO 1 , . 0 0 9 01 4 8 10 , . 0 1 , . 0 0 0 . , 7 0 7 . . 9 5 7 , . 3 9 0 . 19 1 1 2 1 . 19 4 . 2 5 1 1 . . 52 0 38 4 . . 75 1 1 . . 5 2 6 . 6 0 6 6 OO 14 . , 0 0 0 . 0 0 . 0 9 0 OO 164 , .OO 2 . . 0 0 o, . 7 5 O, . 5 5 1 1 5 . OO 1 , . 0 0 9 0 0 . , 3 5 0 , 0 0 , 0 0 7 0 7 , 9 5 7 , , 3 9 0 . 19 0 . 0 5 OO 9 . . 0 5 0 . ,51 5 . 0 0 •9 . 0 5 0 . 0 8 3 . 0 0 12 0 0 0 . 0 0 . 0 5 5 . . 0 0 . 172 . OO 3 . . 0 0 0 . 6 6 0 . 32 1 18 . , 0 0 1 . 0 0 8 9 8 . , 42 3 . 0 0 0 . . 0 0. 6 8 7 . . 3 3 6 7 3 0 . 14 1 1 5 7 . 8 0 8 . 2 5 9 . 0 5 0 . 4 3 8 . 7 5 9 . 4 7 0 . , 0 6 4 . 0 0 28 . 0 0 0 . 0 0 . 0 8 4 . OO 166 . OO 2 . OO 0 . 71 0 . . 51 121 . , 0 0 1 , 0 0 8 9 8 . 4 5 7 . 0 0 3 . 0 0 0 . 6 6 8 . 4 0 7 . . 7 0 0 . 36 1 1 9 0 . 36 6 . , 5 0 1 3 . 58 0 . 8 6 7 . 7 5 1 3 . 5 8 0, . 0 6 0 , OO 14 0 0 0. 0 0 . 0 6 7 . , 0 0 171 , 0 0 2 . OO 0 . 6 8 0 . 39 1 2 2 . . 0 0 1 . 0 0 8 9 8 . 5 9 3 . 0 0 3 7 . 0 0 0. . 61 7 . 9 5 7 . 57 0 . 3 0 1 3 4 S . 4 0 1 0 . , 7 5 12 . 76 0 . 78 11 . 25 1 2 . 7 6 0 . 0 21 , OO 15 . OO 0 . 0 0 . 0 7 2 . . 0 0 166 . . 0 0 1 . , 0 0 0 . 6 9 0 . 4 3 6 0 . 0 0 0 . 0 0 . 0 1. , 25 1 .51 0 . 0 2 . 0 0 7 . 0 0 0 . 14 1 . 7 0 0 . 15 O . 6 9 3 oo 0 . 0 0 . 0 0 .O 0 O 0 . 0 56 6 7 126 . 0 0 0 . 62 9 8 2 8 .OO 3 3 . . 0 0 0 . . 0 14 , 0 0 1. , 11 1 , , 3 0 0 . 0 2 .oo 8 . 0 0 0 , 17 2 . 10 0 . 13 0 . 6 8 3 . 0 0 0 . 0 0 . 0 0 . 0 0 . 0 0 . 0 6 9 . 2 3 125 .OO 0 , 6 6 1 0 3 7 5 . 0 0 61 . 0 0 2 . 0 0 5 . 0 0 1 , 2 5 1 , . 67 0 . 0 2 . 0 0 17 . 0 0 0 , . 19 2 . 4 0 0 . 15 0 . 8 3 3 . 0 0 0 . 0 0 . 0 0 . 0 0 . 0 0 . 0 61 . 6 6 124 . 0 0 0 . . 78 12028 . 0 0 6 6 . 0 0 3 . 0 0 3 . 0 0 1, .01 1 . 13 0 . 0 2 . 0 0 11 . 0 0 0 . 19 2 . 3 0 1 , 13 0 46 3 . 0 0 0 . 0 0 . 0 0 . 0 0 . 0 3 . 0 0 5 . , 5 3 125 .OO 0 , 7 9 1 2 3 7 5 . 0 0 9 5 . 0 0 2 . 0 0 0 . 0 1. . 17 1 . , 4 0 0 , . 0 2 . 0 0 14 . 0 0 0 . , 17 2 . . 10 0 . 87 0 . 7 8 3 . 0 0 37 . OO 53 . 8 0 0 . 0 0 . 0 0 . 0 7 . , 18 1 20 . . 0 0 0 . 6 3 9 1 2 0 . 0 0 81 . 0 0 4 . 0 0 0 . 0 1 3 0 1 .67 0 . 0 2 . 0 0 11 , . 0 0 0 . . 17 2 . 10 0 , , 15 0 . 6 3 3 . 0 0 0 . 0 0 . 0 0 . . 0 0 . 0 0 . 0 6 6 . 67 102 . 0 0 0 . 44 4 5 9 0 . . 0 0 94 OO O . 0 0 . 0 0 . . 0 0 . 0 0 . . 0 2 . 0 0 14 . 0 0 0 . . 18 2. . 4 0 0 . . 13 0 . 9 0 3 . 0 0 0 . 0 0 . 0 0 . .0 0 . . 0 0 , . 0 7 6 . 92 130 . . 0 0 0 . 7 5 1 2 6 1 0 , , 0 0 8 0 . 0 0 2 . 0 0 0 . 0 1 . , 3 0 1 , . 6 4 2 4 3 . . 5 0 2 . . 0 0 9 . , 0 0 0 . 1 1 1 . 4 0 0 . 9 3 0 . 52 3 . . 0 0 36 . 10 57 . 3 0 0 . 0 0 , , 0 0 . 0 5 . 1 1 1 2 3 . OO 0 , . 7 3 1 1 0 7 0 . 0 0 9 7 . 0 0 0 . , 0 O , 0 1 . , 33 1 . 64 0 . 0 2 . . 0 0 14 0 0 0 , . 17 2 . 2 0 1 . 6 1 0 , ,51 3 . 0 0 0 . 0 0 . O 0 . 0 0 . 0 0 . 0 3 . 10 1 1 3 . 0 0 0 . 4 9 6 2 1 5 . 0 0 9 2 . 0 0 5 . 0 0 0 . , 0 0 . 98 1 , . 2 0 0 . 0 2 . . 0 0 1 1 . . 0 0 0 . 12 1 . 7 0 0 . 1 1 O. 5 0 3 . . 0 0 0 . 0 0 . 0 0 . 0 0 . 0 2 . 0 0 7 9 . 54 1 18 . 0 0 0 . 71 9 9 1 2 . . 0 0 8 6 . 0 0 3 . , 0 0 0 , , 0 1 . 26 1 . 5 7 0 . 0 2 . oo 14 , oo 0 . 12 1 . 7 0 1 . 39 0 . 96 3 . . 0 0 0 . 0 0 . 0 0 . 0 0 . 0 0 . 0 5 . 58 1 1 7 . 0 0 0 . 57 7 8 3 9 . OO 3 6 . 0 0 1 . 0 0 2 3 . 0 0 1 . 2 0 1 . 5 5 0 . 0 2 . 0 0 1 5 . 0 0 0 . 18 2 . 5 0 0 . 37 0 . 8 3 3 . 0 0 0 . 0 0 . 0 24 . OO 6 . 0 0 0 o 3 0 . 4 0 1 14 . 0 0 0 . 6 3 8 2 0 8 . 0 0 123 OO 1 .OO 897 .03 0 .69 7 .60 7 .00 8 . 25 9 .47 0 .43 0 .O 73 .OO 7 .00 94 .OO 150 .00 2 .00 125 .OO 1 .OO 898 .36 1 . 16 8 . 36 7 .83 10 . 50 12 .35 0 .40 6 .90 26 .00 15 OO 88 .00 174 .00 1 .00 127 .00 1 .00 896 .07 0 .59 7 .80 7 .34 1 1 .25 9 . 47 0 .83 8 .80 69 .00 17 .00 83 .00 156 .00 2 .00 129 .OO 1 .00 898 . 30 0 .84 9 .70 6 .95 8 . 25 9 .88 0 .55 0 .0 7 .00 1 .00 99 .OO 166 .00 1 .OO 130 .00 1 .00 897 . 63 1 .00 9 . 43 8 .66 5 .25 10 .29 1 . 4 1 0 .0 26 .00 2 .00 81 .OO 176 .OO 1,.00 133. .00 1 .00 893. .56 O. . 70 8 .09 7 , .59 8. .25 8 .64 0 .90 7 . .50 70 .00 21 . 00 89 . OO 167 .oo 1 . .00 01 , • OO 1 .00 889 . 19 0. .0 7 . 26 6 .56 9 . OO 12 . 35 0. .77 0. .0 0 .0 0, .0 90 . OO 155 .OO 2 OO 135 . OO 1 oo 900. . 19 0. . 74 7 .38 6 . 30 10. .00 7, .00 0. 68 0. .0 44 .00 6 .00 90. oo 173 oo 2 .00 137. 00 1 . .00 895. .22 0 . 98 7 . 19 6. 53 6 . 25 6. . 17 0, .45 0 . O 14 . OO 40. 00 111. 00 161 . .00 2 . 00 140. oo 1 . 00 . 895. .74 O. 65 7 . .29 6 . 70 8 . 50 7 . 00 0. 55 0 . 0 6. .00 29. 00 102 . 00 166 . .00 2. 00 142 . OO 1 . 00 892 . 56 0 . 85 7 . 17 6 . 57 9 . 25 6 . .58 0. 68 0 . 0 70. ,00 21 . ,00 78 . OO 173. 00 3. ,00 145 . .00 1 . .00 889 76 0. 69 7 . . 17 6. 61 10. OO 5. 76 O. .58 0. .0 70, ,00 14 . 00 87 . OO 172 . .00 2. 00 17 .00 0 .0 83 .00 1 0. 13 1075 .56 0 .0 2 8 .25 9 .47 0 . 13 0 0 .0 0 .0 O .0 0 0 . 77 0 .63 0 .8210810 5 .00 9 .00 4 1 OO 8 0 .37 1057 .80 254 . 70 2 10 .50 12 .35 1 . 48 0 0 .0 0. 0 0 .0 9 0 .73 0 .51 0 .6911176 4 .00 0 .0 86 .00 9 0 .27 1051 .78 0 .0 2 1 1 . 25 9 .47 0 . 15 0 0 .0 0. .0 0 .0 0 0 .71 0 . 53 0. .75 9130 20 .00 34 .00 41 .00 0 1 .40 1317 .04 224 .60 2 8 .25 9 88 7. .83 0 2 .00 3. .00 0 .0 27 0 .77 0 60 0 .7712672 9. .00 46 00 30. OO 0 1 .65 1262 . 54 0 .0 2 5 . 25 10. . 29 10 .87 1 5 .00 3 .00 5. .00 33 0 . 73 0, .46 0 .6310368 4 . 00 2 . ,00 92. .00 1 0 . 36 1204 . . 29 0 ,0 2 8 . 25 8 .64 0 . 15 0 0 .0 0, .0 0 .0 2 0 .74 0. ,53 0. .7211036 0 .0 0. 0 0 ,0 0 0 . 12 o. .0 0. .0 2 9. .00 12. . 35 0. .91 0 0 .0 0. .0 0 .0 0 0 .76 0 . 58 0. ,7610620 49 .00 0. .0 50. .00 0 0 . 13 1067 , . 35 0. .0 2 10, .00 7 . .00 0, ,09 0 0. .0 0. 0 0. .0 O 0. ,75 0. 52 0 .6911700 1 . .00 4 . 00 54. 00 12 0. , 1 1 104 1 . 66 0. 0 2 7 , .75 7 . 82 0. . 15 0 4 . OO 0. 0 0. 0 0 0, 83 0. 69 0. 8314763 12 . OO 1 . OO 54 . 00 16 0. 16 1 198 . 75 0. 0 2 8 75 7 . 00 0. 13 0 1 , ,00 0. 0 0. 0 0 0. .80 0. 61 0. 7713464 6 . 00 0 . 0 91 . 00 0 0. . 13 1027 . 50 0 . 0 2 9 .25 6 . 58 0. 11 0 0. 0 0. 0 0. 0 0 O. .73 0 . 45 0. 62 9828 13, 00 0. 0 84 . 00 0 0. 17 1 158 . 50 380. 80 2 10. OO 5 . 76 0. 13 0 0. ,0 0. 0 0. 0 0 0, 74 0 . 51 0. 6911049 00 0 .0 0 .90 1 . 10 00 6 .00 0 . 15 1 .80 43 3 .00 0 .0 0 .0 0 0 .0 63 .46 1 15 .00 00 00 21 .00 1 . 16 1 . 48 oo 1 1 .oo 0 . 1 1 1 .30 42 3 .00 26 .90 66 . 20 OO 0 .0 7, ,09 127 .00 00 00 0 .0 1 . 14 1 .50 00 18 .00 0 . 14 1 .90 91 3 .oo 36 .20 55 .00 0 0 .0 75 .00 1 10 .00 00 0 0 .0 1 . 10 1 . 35 00 12 .00 0 . 10 1 . 10 55 3 .00 0. .0 0 .0 00 0 .0 1 .05 128 .00 00 0 0 .0 1. . 14 1 . 30 00 37 .00 0 . 1 1 1 . 40 4 1 3, .oo 0 .0 0 .0 00 0 .0 0. 48 128 .00 00 00 0 .0 1, . 16 1 .51 oo 18 00 0, , 13 1 . 70 90 3 oo 31 .60 60 .90 00 0 .0 55 OO 124 .00 00 0 0 .0 0 .0 0 .0 oo 9 oo 0. . 1 1 1 . 50 78 3, .00 0, ,0 0 .0 0 0 .0 9 89 1 18 .00 00 0 0 .0 1 . , 36 1 . 70 OO 18 .00 0 . 20 2 .50 68 3. .00 0. 0 0 .0 0 0. .0 111. 1 1 130 .00 oo 00 27. 00 1. 16 1 . 33 oo 9. oo 0. 12 1 . . 30 60 3. 00 0. 0 0 .0 0 0. 0 51 , 67 133 . 00 00 oo 15 . oo 1 . 28 1 74 00 17 . oo 0. 12 1 . 50 61 3. oo 0. 0 0, .0 0 0. 0 67 . 30 132 .00 oo 0 0. 0 1 . 26 1 , ,62 00 34. .00 0. 12 1 80 68 3. 00 0. 0 0 .0 0 0. 0 84 . 09 126 , .00 00 0 0. .0 1 . 39 1 .80 00 46 . 00 0. 1 1 1 . .40 68 3 . 00 0. 0 0. 0 0 0. 0 76 . 92 127 . OO 00 

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.831.1-0097838/manifest

Comment

Related Items