ARROW TSA PREDICTIVE ECOSYSTEM MAPPING (PEM) 2003 RE...

ARROW TSA PREDICTIVE ECOSYSTEM MAPPING

(PEM) 2003 RE-ITERATION

Improvements to the PEM model

FINAL PROJECT REPORT

For:

Colin Pike Bell Pole Ltd.

4661 60th Street Salmon Arm BC V1E 1X2

(250) 832-1180

By:

Maureen Ketcheson MSc R.P.Bio Tom Dool BES Lawson Bradley

Keyes Lessard For Tech Gareth Kernaghan For Tech

JMJ Holdings Inc.

Suite 208 – 507 Baker Street Nelson, BC V1L 4J2

(250) 354-4913

March 31, 2003.

ARROW TSA PREDICTIVE ECOSYSTEM MAPPING (PEM) 2003 RE-ITERATION

JMJ Holdings Inc. suite 208 – 507 Baker Street, Nelson, BC V1L 4J2 (250) 354-4913 March31, 2003

Table of Contents 1.0 INTRODUCTION ....................................................................................................................... 1

2.0 METHODS................................................................................................................................. 1

2.1 INPUT LAYER DEVELOPMENT AND TESTING................................................................... 1 2.2 INPUT DATA QUALITY ASSESSMENT REPORT................................................................ 1

2.2.1 PROJECT OBJECTIVES ............................................................................................... 2 2.2.2 THEMATIC INPUT LAYER DOCUMENTATION AND QUALITY ASSESSMENT ......... 5

2.2.2.1 TRIM 1 DATA ........................................................................................................... 6 2.2.2.1.1 TRIM 1 – TNTL LAYER.................................................................................. 6 2.2.2.1.2 TRIM 1 – TWTR LAYER ................................................................................ 7 2.2.2.1.3 TRIM 1 – TDEM LAYER................................................................................. 8

2.2.3 FOREST INVENTORY FOREST COVER MAPPING .................................................... 9 2.2.4 ARROW TSA PLOT DATA............................................................................................ 10 2.2.5 BIOGEOCLIMATIC SUBZONE MAPPING ................................................................... 11 2.2.6 JMJ HOLDINGS BIOTERRAIN MAPPING ................................................................... 12 2.2.7 INPUT PROCESSING AND QUALITY ASSESSMENT................................................ 13

2.2.7.1 TRIM 1 INPUT LAYERS......................................................................................... 15 2.2.7.1.1 INPUT LAYERS GENERATED FROM TRIM TNTL INPUT LAYER ............... 16

2.2.7.1.1.1 MASK GRID – MASK ................................................................................ 16 2.2.7.2. INPUT LAYERS GENERATED FROM TRIM TDEM INPUT LAYER.................... 16

2.2.7.2.1 SLOPE - SLP ................................................................................................... 16 2.2.7.2.2 ASPECT – AS.................................................................................................. 19 2.2.7.2.3 LANDSCAPE SHAPE - SHP ........................................................................... 20 2.2.7.2.4 SLOPE POSITION........................................................................................... 23

2.2.7.3 INPUT LAYERS GENERATED FROM TRIM TWTR INPUT LAYER .................... 27 2.2.7.3.1 HYDROGRAPHIC FEATURES – TWTR......................................................... 27

2.2.7.4 INPUT LAYERS GENERATED FROM FOREST COVER..................................... 28 2.2.7.4.1 ALPINE FOREST - AF..................................................................................... 28 2.2.7.4.2 FOREST COVER ROCK INPUT LAYER – FC_ROCK ................................... 29 2.2.7.4.3 URBAN AREAS – UR ...................................................................................... 29 2.2.7.4.4 ALPINE ECOSYSTEMS - ALP ........................................................................ 30

2.2.7.5 AVALANCHE PATH AND RUNOUT ZONE INPUT LAYER - AV .......................... 30

3.0 KNOWLEDGE BASE DEVELOPMENT AND TESTING........................................................ 33

3.1 FIRST RUN RASTER SITE SERIES ALLOCATION ........................................................... 33 3.2 SECOND RUN RASTER MODIFICATIONS ........................................................................ 34

3.2.1 SECOND RUN TRIM WATER FEATURE MODIFICATIONS....................................... 35 3.2.2 SECOND RUN VECTOR BIOTERRAIN MODIFICATIONS ......................................... 35 3.2.3 SECOND RUN VECTOR FOREST COVER MODIFICATIONS................................... 36

3.3 PEM MODEL OVERVIEW................................................................................................... 36

4.0 RESULTS................................................................................................................................ 38

4.1 PEM MODEL MAP ENTITY ALLOCATION BY BEC VARIANT ........................................... 38 4.2 GOODNESS OF FIT AND ACCURACY ASSESSMENT ..................................................... 43

4.2.1 CHI-SQUARED TEST OF ECOSYSTEM UNIT PROPORTIONS................................ 43 4.2.1.1 ESSF wc1 TEST OF ECOSYSTEM UNIT PROPORTIONS................................. 43 4.2.1.2 ESSF wc4 TEST OF ECOSYSTEM UNIT PROPORTIONS................................. 45 4.2.1.3 ICH dw TEST OF ECOSYSTEM UNIT PROPORTIONS....................................... 47 4.2.1.4 ICH mw2 TEST OF ECOSYSTEM UNIT PROPORTIONS................................... 49 4.2.1.5 TEST OF ECOSYSTEM UNIT PROPORTIONS FOR ALL SUBZONES.............. 52

4.2.2 PROPORTION OF DOMINANT ENTITY CORRECT .................................................. 56 4.2.2.1 ESSF wc1 PROPORTION OF DOMINANT ENTITY CORRECT ......................... 56



4.2.2.2 ESSF wc4 PROPORTION OF DOMINANT ENTITY CORRECT ......................... 58 4.2.2.3 ICH dw PROPORTION OF DOMINANT ENTITY CORRECT............................... 59 4.2.2.4 ICH mw2 PROPORTION OF DOMINANT ENTITY CORRECT............................ 61 4.2.2.5 OVERALL PROPORTION OF DOMINANT ENTITY CORRECT........................... 63

4.2.3 THE PERCENT OVERLAP TEST................................................................................ 64 4.2.3.1 ESSF wc1 PERCENT OVERLAP TEST ............................................................... 64 4.2.3.2 ESSF wc4 PERCENT OVERLAP TEST ............................................................... 65 4.2.3.3 ICH dw PERCENT OVERLAP TEST .................................................................... 66 4.2.3.4 ICH mw2 PERCENT OVERLAP TEST ................................................................. 68

4.2.4 ON POINT ACCURACY ASSESSMENT...................................................................... 70

5.0 REFERENCES CITED ............................................................................................................ 76

List of Figures

Figure 1. TRIM 1 Mapsheets required for the Project ..................................................................... 3 Figure 2. Arrow TSA PEM Mapping Area........................................................................................ 4 Figure 3. Calculation of Curvature................................................................................................. 21 Figure 4. Arrow 2003 PEM Model Overview. ................................................................................ 37

List of Tables Table 1. Thematic Input Layers and Input Data Layers ............................................................... 14 Table 2. Slope Classes.................................................................................................................. 17 Table 3. Slope Input Layer Confusion Matrix ................................................................................ 17 Table 4. Statement of Overall Accuracy........................................................................................ 18 Table 5. Aspect Classes................................................................................................................ 19 Table 6. Aspect Input Layer Confusion Matrix .............................................................................. 19 Table 7. Statement of Overall Accuracy........................................................................................ 20 Table 8. Distribution of Curvature Values...................................................................................... 22 Table 9. Landscape Shape Classes.............................................................................................. 22 Table 10. Distribution of Slope Position Values ............................................................................ 25 Table 11. Slope Position Classes.................................................................................................. 26 Table 12. Slope Position Input Layer Confusion Matrix ................................................................ 26 Table 13. Statement of Overall Accuracy of Slope Position.......................................................... 26 Table 14. TRIM 1 Feature Codes and Corresponding Hydrographic Features ............................ 27 Table 15. Alpine Forest Areas Classes ......................................................................................... 28 Table 16. Forest Cover Rock Input Layer Classes........................................................................ 29 Table 17. Urban Areas Classes..................................................................................................... 29 Table 18. Alpine Ecosystem Input Layer Classes ......................................................................... 30 Table 19. Arrow TSA 2003 PEM Results ...................................................................................... 39 Table 20. Distribution of Site Series in ESSF wc1 Plots – Accuracy Data.................................... 43 Table 21. Distribution of Map Entities and Site Series in the ESSF wc1 PEM Polygons.............. 43 Table 22. Chi Squared Test of Ecosystem Proportions in the ESSFwc1 – Accuracy Data.......... 44 Table 23. Distribution of Site Series in ESSF wc1 Plots – Model Building Data........................... 44 Table 24. Chi Squared Test of Ecosystem Proportions in the ESSFwc1 – Model Building Data . 45 Table 25. Distribution of Site Series in ESSF wc4 Plots – Accuracy Data.................................... 45 Table 26. Distribution of Site Series in the ESSF wc4 PEM Polygons.......................................... 46 Table 27. Chi Squared Test of Ecosystem Proportions in the ESSFwc4 – Accuracy Data.......... 46 Table 28. Distribution of Site Series in ESSF wc4 Plots – Model Building Data........................... 47 Table 29. Chi Squared Test of Ecosystem Proportions in the ESSFwc4 – Model Building Data . 47 Table 30. Distribution of Site Series in ICH dw Plots – Accuracy Data......................................... 47 Table 31. Distribution of Site Series in the ICH dw PEM Polygons............................................... 48 Table 32. Chi Squared Test of Ecosystem Proportions in the ICH dw – Accuracy Data .............. 48 Table 33. Distribution of Site Series in ICH dw Plots – Model Building Data................................ 49



Table 34. Chi Squared Test of Ecosystem Proportions in the ICH dw – Model Building Data ..... 49 Table 35. Distribution of Site Series in ICH mw2 Plots – Accuracy Data...................................... 50 Table 36. Distribution of Site Series In ICH mw2 PEM Polygons ................................................. 50 Table 37. Chi Squared Test of Ecosystem Proportions in the ICH mw2 – Accuracy Data........... 51 Table 38. Distribution of Site Series in ICH mw2 Plots – Model Building Data............................. 51 Table 39. Chi Squared Test of Ecosystem Proportions in the ICH mw2 – Model Building Data .. 52 Table 40. Summary of Test of Ecosystem Proportions by Subzone – Accuracy Data ................. 52 Table 41. Summary of Test of Ecosystem Proportions by Subzone – Model Building Data ........ 52 Table 42. Anthropogenic and Non-Vegetated Plots and PEM Area ............................................. 53 Table 43. Site Series in the Mesic 01/04 Class............................................................................. 53 Table 44. Site Series in the Xeric 02 Class ................................................................................... 53 Table 45. Site Series in the Subhygric 03 Class ........................................................................... 54 Table 46. Site Series in the Submesic/Subxeric Class ................................................................. 54 Table 47. Site Series in the Hygric 04 / 05 Class .......................................................................... 54 Table 48. Site Series in the Subhygric-Subhydric Class............................................................... 55 Table 49. Pooled Chi Square Test ................................................................................................ 55 Table 50. Partially Grouped Chi-Square Test ............................................................................... 56 Table 51. ESSF wc1 Dominant Entity Correct Calculation ........................................................... 57 Table 52. ESSF wc1 Dominant Entity Correct .............................................................................. 57 Table 53. ESSF wc4 Dominant Entity Correct Calculation ........................................................... 58 Table 54. ESSF wc4 Dominant Entity Correct .............................................................................. 59 Table 55. ICH dw Dominant Entity Correct Calculation................................................................. 60 Table 56. ICH dw Dominant Entity Correct ................................................................................... 61 Table 57. ICH mw2 Dominant Entity Correct Calculation. ............................................................ 62 Table 58. ICH mw2 Dominant Entity Correct ................................................................................ 63 Table 59. Summary of Dominant Entity Correct Tests Weighted Average ................................... 64 Table 60. Summary of Dominant Entity Correct Tests Unweighted Average ............................... 64 Table 61. ESSF wc1 Percent Overlap Calculations...................................................................... 65 Table 62. ESSF wc4 Percent Overlap Calculations...................................................................... 66 Table 63. ICH dw Percent Overlap Calculations ........................................................................... 67 Table 64. ICH mw2 Percent Overlap Calculations. ....................................................................... 68 Table 65. Summary of Percent Overlap Test ................................................................................ 70 Table 66. Distribution of Polygon Deciles...................................................................................... 70 Table 67. Plot and Polygon Comparison....................................................................................... 71 Table 68. Relationship of Correct Calls to Deciles ........................................................................ 75

List of Appendices (contained on CD) APPENDIX I – Pivot Tables APPENDIX II – Knowledge Tables APPENDIX III – Second Run Rule Sets APPENDIX VI – Plot GIS Attributes APPENDIX V – Area Summaries APPENDIX VI – Project Metadata


JMJ Holdings Inc. suite 208 – 507 Baker Street, Nelson, BC V1L 4J2 (250) 354-4913 March 31, 2003

page 1

1.0 INTRODUCTION The original Arrow PEM model (Ketcheson et al 2001) was an innovative undertaking designed to model site series and support an assignment of site index based on the SIBEC approach (BC Forests Productivity Council, Site Productivity Working Group 2001). The 2001 Arrow TSA PEM model was moderately successful at predicting site series, with an independent accuracy of 51.1% and an overall model goodness of fit of 60.3% (Serrouya pers com., 2002). Review by MOF personnel in Victoria indicated that the model did not achieve a high enough accuracy score to be used for the basis of a site index adjustment. MOF requires a PEM model with a percent dominant correct and percent dominant overlap score of at least 65% for that activity (Meidinger 2001). Subsequent to the completion of the Arrow PEM, JMJ Holdings Inc. has continued to refine our bioterrain based PEM process with good results. We achieved percent dominant correct and percent dominant overlap scores of 67% and 71% respectively (Lloyd, D. 2002b) in the Dry Belt Okanagan TSA PEM project (Ketcheson et al 2002). We proposed to the Arrow IFPA committee that we could re-visit the original Arrow PEM model and incorporate our improvements to the modeling process with the end goal being an accuracy score of greater than 65%. This would better facilitate the PEM model being used in support of an assignment of site index that could be used in Timber Supply Analysis activities. What follows is the documentation of the methods and results of modifications and improvements to the original Arrow model. 2.0 METHODS 2.1 INPUT LAYER DEVELOPMENT AND TESTING The original Arrow TSA PEM did not have an input data quality assessment component. There were input layers used in the 2001 PEM model which may have had a poor relationship to the field data. These layers, in some cases were weighted heavily in the knowledge bases, this likely affected the accuracy of the original model. The original input layers were tested and improved where necessary for the 2002 Arrow PEM model.

2.2 INPUT DATA QUALITY ASSESSMENT REPORT The second phase of Predictive Ecosystem Mapping in the Arrow TSA began November 2002 with the review of Geographic Information Systems (GIS) input layers used in the previous iteration. Layers were tested against available field data, and improved where needed. This document will discuss the changes and detail the accuracy of the new layers.



page 2

2.2.1 PROJECT OBJECTIVES The task required of this phase of the project included: • The assessment of existing input layers from Year 1. • The assessment of improved thematic input layers for use in the PEM process • The creation of GIS input layers • The assessment of GIS input layers for use in the PEM process For purposes of clarification it should be stated here that this phase of the project involved the use of 4 Thematic Input Layers for assessment and collected field data. • Ministry of Sustainable Resource Management TRIM 1 mapping • Ministry of Forests Forest Cover Data • Ministry of Forests Biogeoclimatic Mapping • Arrow TSA Predictive Ecosystem Mapping Project Field Plot Data The final PEM used 12 input layers: • Mask • Biogeoclimatic zones • Slope • Aspect • Landscape Shape • Slope Position • Hydrography • Urban • Forest Cover Rock • Alpine Forest • Alpine Ecosystems • Avalanche paths



page 3

The original mapping extents for thematic input layers ranged from localized plot data to provincial data sets. All input layers with the exception of the Arrow TSA PEM Plot Data are contiguous layers covering 130 TRIM 1 mapsheets shown below in Figure 1. Figure 2 shows the Arrow TSA PEM Mapping Area. Figure 1. TRIM 1 Mapsheets required for the Project 082E.010 082E.069 082F.013 082F.051 082F.084 082K.015 082K.054 082E.020 082E.070 082F.014 082F.052 082F.085 082K.021 082K.061 082E.029 082E.079 082F.015 082F.053 082F.091 082K.022 082K.062 082E.030 082E.080 082F.021 082F.061 082F.092 082K.023 082K.063 082E.039 082E.089 082F.022 082F.062 082F.093 082K.024 082E.040 082E.090 082F.023 082F.063 082F.094 082K.031 082E.049 082E.098 082F.024 082F.064 082F.095 082K.032 082E.050 082E.099 082F.025 082F.065 082K.001 082K.033 082E.059 082E.100 082F.031 082F.071 082K.002 082K.034 082E.060 082F.001 082F.032 082F.072 082K.003 082K.041 082E.069 082F.002 082F.033 082F.073 082K.004 082K.042 082E.070 082F.003 082F.034 082F.074 082K.005 082K.043 082E.079 082F.004 082F.035 082F.075 082K.011 082K.044 082E.080 082F.005 082F.041 082F.081 082K.012 082K.051 082E.089 082F.011 082F.042 082F.082 082K.013 082K.052 082L.010 082L.040 082L.070 082L.100 082L.009 082K.064 082L.019 082L.049 082L.079 082N.002 082K.083 082K.071 082L.020 082L.050 082L.080 082N.003 082K.092 082K.072 082L.029 082L.059 082L.089 082N.004 082K.093 082K.073 082L.030 082L.060 082L.090 082N.013 082K.094 082K.074 082L.039 082L.069 082L.099 082N.014 082K.082 082K.081 082E.090 082F.012 082F.043 082F.083 082K.014 082K.053



page 4

Figure 2. Arrow TSA PEM Mapping Area.



page 5

It should be noted that these mapsheets represent the extents of the entire Arrow TSA study area. Currently the areas subject to study include the following biogeoclimatic subzones. They are: AT ESSF dc 1 ESSF dcp 1 ESSF vc ESSF vcp ESSF wc 1 ESSF wc 4 ESSF wcp 4 ICH dw ICH mw 2 ICH vk 1 ICH wk 1 ICH xw IDF un 2.2.2 THEMATIC INPUT LAYER DOCUMENTATION AND QUALITY ASSESSMENT Consultation with the client and the Ministry of Sustainable Resource Management indicated that there was sufficient data of a quality high enough to carry out a PEM project. These thematic input layers included • Ministry of Sustainable Resource Management TRIM 1 mapping • Ministry of Forests Forest Cover mapping • Ministry of Forests Biogeoclimatic Mapping • Arrow TSA Predictive Ecosystem Mapping Project Field Plot Data • JMJ Holdings Bioterrain Mapping



page 6

2.2.2.1 TRIM 1 DATA TRIM 1 data was acquired from the client in the spring of 2001. Several different data layers were used from the TRIM 1 thematic mapping. • TDEM – Elevation points and 3D breaklines • TWTR – Hydrographic Features • TNTL – 1:20,000 Neatlines This information was then processed and used to produce • Mask • Aspect • Landscape Shape • Slope Position • Slope • Hydrography More information on the processing of this information for the generation of input layers can be found in Section 2.2.7 “Input Processing” of this document.

2.2.2.1.1 TRIM 1 – TNTL LAYER

Citation: TRIM 1 Program Consultant/Department: Geographic Data BC Branch, Ministry of Sustainable Resource Management Publication scale: 1:20,000 Period of Compilation: 1997 Base Map Projection: Albers, NAD 83 Quality Control: As per Table 1 and Section 4.4.1.1. in the Standards for Predictive Ecosystem Mapping in British Columbia, Inventory Standard, (RIC, 1999) the neatline features in the TRIM 1 data were considered to be of adequate quality to complete a Predictive Ecosystem Mapping project at a scale of 1:20,000. Edge Matching: Edge matching was completed along mapsheet boundaries using the Arc/Info 8.1 APPEND process. No dangles were found along mapsheet boundaries. Edge Matching Error Minimum: 0 m shift along the X axis, 0 m shift along the Y axis. Edge Matching Error Average: 0 meters on the X axis. 0 meters along the Y axis.



page 7

Edge Matching Error Max: 0 meters on the X axis. 0 meters along the Y axis. Attribute/Label Matching: All neatline boundaries contained the appropriate labels and attributes Raster Sized: NA Adjusted Control Feature Shift: None

2.2.2.1.2 TRIM 1 – TWTR LAYER

Citation: TRIM 1 Program Consultant/Department: Geographic Data BC Branch, Ministry of Sustainable Resource Management Publication scale: 1:20,000 Period of Compilation: 1997 Base Map Projection: Albers, NAD 83 Quality Control: As per Table 1 and Section 4.4.1.1. in the Standards for Predictive Ecosystem Mapping in British Columbia, Inventory Standard, (RIC, 1999) the single line water features in the TRIM 1 data were considered to be of adequate quality to complete a Predictive Ecosystem Mapping project at a scale of 1:20,000. Edge Matching: Edge matching was completed along mapsheet boundaries using the Arc/Info 8.1 APPEND process. No dangles were found along mapsheet boundaries. Line work spanning map boundaries was joined to create seamless coverage. 100 lines were randomly selected from the input data and compared to the final output seamless coverage. Error was measured in terms of shift along the X axis (Easting) and the Y axis (Northing). Edge Matching Error Minimum: 0 m shift along the X axis, 0 m shift along the Y axis. Edge Matching Error Average: 0 meters on the X axis. 0 meters along the Y axis. Edge Matching Error Max: 0 meters on the X axis. 0 meters along the Y axis. Attribute/Label Matching: All water features were coded with the correct FCODE. Raster Sized: NA



page 8

Adjusted Control Feature Shift: None

2.2.2.1.3 TRIM 1 – TDEM LAYER Citation: TRIM 1 Program Consultant/Department: Geographic Data BC Branch, Ministry of Sustainable Resource Management Publication scale: 1:20,000 Period of Compilation: 1997 Base Map Projection: Albers, NAD 83 Quality Control: As per Table 1 and Section 4.4.1.1. in the Standards for Predictive Ecosystem Mapping in British Columbia, Inventory Standard, (RIC, 1999) the TDEM layer in the TRIM 1 data was considered to be of adequate quality to complete a Predictive Ecosystem Mapping project at a scale of 1:20,000. Edge Matching: DEM points were merged into one seamless Arc/Info point coverage. No edge-matching was required as there is inherent overlap in the data points for each mapsheet. It should be noted that in the zone of overlap between 2 given mapsheets there were some irregularities in the elevation values. This is to be expected as each mapsheet relies on different control points, which may be slightly higher or lower. These differences were of a low enough proportion that the process of converting to grid would have the effect of “smoothing” the surface and nullifying the differences. It should also me noted that as no 2 TDEM points occupy the same space they cannot be compared in an absolute fashion. In an effort to gauge the degree of difference between the respective sheets, contours with a 20m interval were created and compared in areas of overlap. 50 different meeting points along map sheet neatlines were selected between contour ‘pairs’ and the degree of difference was measured. Edge matching error is measured on an x-y plane as the contour pairs both represent the same elevation. Edge Matching Error Minimum: 2.1 meters Edge Matching Error Average: 6.3 meters Edge Matching Error Max: 9.8 meters Attribute/Label Matching: N/A Raster Sized: N/A



page 9

Adjusted Control Feature Shift: None 2.2.3 FOREST INVENTORY FOREST COVER MAPPING

Ministry of Forests Forest Cover Mapping (FC1) was provided by Ministry of Forests (Arrow Region). This information was received in as seamless Arc/Info coverage for the Arrow District and then processed to produce information on: • Alpine Forest • Alpine Ecosystems • Urban Areas • Rock For Information on the processing of this information for the generation of input layers can be found in Section 2.2.7 “Input Processing” of this document. Citation: Forest Cover/Forest Inventory Plan Data Consultant/Department: MoF, Kamloops Forest Region Publication scale: 1:20,000 Period of Compilation: 1997 Base Map Projection: UTM NAD 83 Zone 11 Quality Control: As per Table 1 and Section 4.4.1.1. in the Standards for Predictive Ecosystem Mapping in British Columbia, Inventory Standard, (RIC, 1999) the MoF FC1 data was considered to be of adequate quality to complete a Predictive Ecosystem Mapping project at a scale of 1:20,000. Edge Matching: Data was received in a seamless format. No edge-matching was required. Edge Matching Error Minimum: N/A Edge Matching Error Average: N/A Edge Matching Error Max: N/A Attribute/Label Matching: N/A Raster Sized: N/A



page 10

Adjusted Control Feature Shift: None

2.2.4 ARROW TSA PLOT DATA

This information was gathered from August 9th-October 19th 2001, for use this project by JMJ Holdings Inc. For information on the processing of this data for the generation of input layers can be found in Section 2.2.7 “Input Processing” of this document. Citation: Arrow TSA Predictive Ecosystem Mapping Project Consultant/Department: JMJ Holdings Inc. Publication scale: 1:20,000 Period of Compilation: August 9th – October 19th, 2001 Base Map Projection: UTM NAD 83 Zone 11 Quality Control: Field cards were filled in on the site and then rechecked as the data was entered into digital format within the VENUS data base. UTM coordinates were collected using a Garmin GPS 2 Plus hand held GPS with a stated accuracy of +/-15m RMS. Edge Matching: No edge matching was required Edge Matching Error Minimum: None Edge Matching Error Average: None Edge Matching Error Max: None Attribute/Label Matching: None Raster Sized: NA Adjusted Control Feature Shift: NA



page 11

2.2.5 BIOGEOCLIMATIC SUBZONE MAPPING Biogeoclimatic Subzone mapping was provided by Marvin Eng, Research Branch, Ministry of Forests, Victoria BC. This information was received in as seamless Arc/Info coverage for the study area. It was used to produce the Biogeoclimatic Subzone Input Layer. Citation: Biogeoclimatic Subzone Data, Ministry of Forests Consultant/Department: MoF; Research Branch; Arrow Forest Region; Marvin Eng Publication scale: 1:50,000 Period of Compilation: 2001 Base Map Projection: UTM NAD 83 Zone 11 Quality Control: The Biogeoclimatic Subzone mapping was developed by and approved by Tom Braumandl, former Regional Ecologist, Nelson Forest Region for use in this PEM. Edge Matching: Data was received in a seamless format. No edge-matching was required. Edge Matching Error Minimum: N/A Edge Matching Error Average: N/A Edge Matching Error Max: N/A Attribute/Label Matching: N/A Raster Sized: N/A Adjusted Control Feature Shift: None



page 12

2.2.6 JMJ HOLDINGS BIOTERRAIN MAPPING Bioterrain mapping was undertaken by JMJ Holdings Inc. for use in the Arrow TSA PEM Project. Bioterrain polygons were delineated by Jen Shypikta P.Geo and Grant Burns GIT using the classification system of Howes and Kenk (1997). Spatial data was captured by Graham Smith, Geosense, Nelson BC. using a high-resolution scanner and PCI Geomatics v8.0. This information was used to create the following thematic input layers: • Bioterrain rock • Bioterrain talus • Bioterrain vegetated rock • Bioterrain wetlands Citation: Bioterrain Mapping, JMJ Holdings Inc. Consultant/Department: JMJ Holdings Inc., Geosense Publication scale: 1:15,000 Period of Compilation: 1999/2000 Base Map Projection: UTM NAD 83 Zone 11 Quality Control: Internal quality control was accomplished by Vicky Lipinski, Donna Ross and Jen Shypitka. External quality assurance was undertaken by Deepa Filatow of the Ministry of Sustainable Resource Management, Pentiction BC. Edge Matching: Data was edge-matched by hand using ARC/Info 8.1. Edge Matching Error Minimum: N/A Edge Matching Error Average: N/A Edge Matching Error Max: N/A Attribute/Label Matching: N/A Raster Sized: N/A Adjusted Control Feature Shift: None



page 13

2.2.7 INPUT PROCESSING AND QUALITY ASSESSMENT For the purposes of this report Section 2.2.7 Input Processing will describe the processes and the rationale to convert the raw input coverages discussed in Section 2.2.2 Input Data Documentation and Quality Assessment and generate the thematic data layers used in the PEM process. Quantifying the success of the overall process is also a part of input processing and quality assessment. This section will also describe the results of the processes in terms of 2 statistical tests. These tests were deemed appropriate and approved for the assessment of spatial data by the Ministry of Sustainable Resource Management (Meidinger, 2000). The methods are: • The Confusion Matrix Description of Overall Accuracy • The Statement of Percent Correct to the Nearest +/-95% Confidence Interval To test the locational accuracy of the GIS data a confusion matrix is developed to test each class of the field data against each class of the GIS data. For instance, how many of the field plots that were classified as slope class 1 actually fall into pixels that the GIS has classified as slope class 1. The results of the matrix include a stated Error of Omission, Error of Commission, and an initial statement of Overall Accuracy. To use the slope attribute as an example, the Error of Commission represents the percentage of occurrences were the plot data (a known thing) has been classified by the GIS as being in a slope class where it does not belong. In this case we have committed the act of getting it wrong. The Error of Omission represents the percentage of occurrences where the plot data has not been classified by the GIS as being in the right slope class. In this case we have omitted the act of getting it right (Meidinger, 2000). The initial Statement of Overall Accuracy is the number of plots where the GIS and the field data agree divided by the number of plots where the GIS data and the field data disagree. The Initial statement of overall accuracy does not indicate how much overall confidence we have in the results of the Confusion Matrix Description of Overall Accuracy it’s result is limited to just the test. The Statement of Overall Accuracy defines the results of the Confusion Matrix in terms of a 95% confidence interval. For example, the confusion matrix may state an Initial Overall Accuracy of 62% but if only 8 plots were used for the test then the lower 95% confidence interval would be 25% and the upper 95% confidence interval would be 87%. In testing the same spatial data with the same result of 62% Initial Overall Accuracy but with a sample set of 100 plots the Statement of Overall Accuracy would have a lower 95% confidence interval of 56% and an upper 95% confidence interval of 69%. With more plots (a greater sample population) we can state our results with a higher degree of accuracy. The Input Data Layers and the Thematic Input Layers which were derived from them are listed in Table 1.



page 14

Table 1. Thematic Input Layers and Input Data Layers Thematic Input Layer

Data Layer PEM Data Layer

TRIM 1 TRIM1 – TNTL Mask – MASK TRIM 1 TRIM1 – TWTR Hydrography – TWTR TRIM 1 TRIM 1 – TDEM Slope – SLP

Aspect – AS Landscape Shape – SHP Slope Position – LNDPOS

Forest Cover Forest Cover Urban – UR Rock – FC_ROCK Alpine Forest – AF Alpine Ecosystems – ALP

Field Data Field Data Plot Data Biogeoclimatic Subzone Mapping

Biogeoclimatic Subzone Mapping

Biogeoclimatic Subzones – BEC

Forest Cover and TRIM 1

Forest Cover TRIM 1 – TDEM TRIM 1 – TWTR

Avalanche Paths - AV

It should be noted that an initial data set of 1200 plots were compiled and made into a spatial input layer for testing. These plots were then screened for use in tests for each GIS Input Data Layer. For each of these tests plots were excluded when they were not deemed fit for use due to discrepancies in the classes. This is the reason that sample sizes for the testing of each layer varies.



page 15

2.2.7.1 TRIM 1 INPUT LAYERS

The TRIM 1 Input layers included • TNTL • TDEM All layers were projected from their original Albers Projection into Universal Transverse Mercator Zone 11. This was done using an AML provided by the Ministry of Sustainable Resource Management and the ArcInfo 8.1 PROJECT command. The parameters of the translation are as follows. input Projection ALBERS Datum NAD83 Zunits NO Units METERS Xshift 0.0000000000 Yshift 0.0000000000 Parameters 50 0 0.000 /* 1st standard parallel 58 30 0.000 /* 2nd standard parallel -126 0 0.000 /* central meridian 45 0 0.000 /* latitude of projection's origin 1000000.00000 /* false easting (meters) 0.00000 /* false northing (meters) output Projection UTM Zone 11 Datum NAD83 Zunits NO Units METERS Xshift 0.0000000000 Yshift 0.0000000000



page 16

2.2.7.1.1 INPUT LAYERS GENERATED FROM TRIM TNTL INPUT LAYER 2.2.7.1.1.1 MASK GRID – MASK

The TRIM1 neatline coverage for the study area was converted into a 50m grid using the ArcInfo 8.1 POLYGRID command. This coverage provides the ‘template’ for all other grids. It is used in the analysis environment to ensure that all subsequent grids and their individual pixels line up so that a given 50m x 50m point in space within the study area is represented by pixels that are congruent despite the fact that they are in different input layers.

2.2.7.2. INPUT LAYERS GENERATED FROM TRIM TDEM INPUT LAYER

2.2.7.2.1 SLOPE - SLP ArcInfo Grid Slope Command was used on the Elevation Grid. The options selected were percent and a z-factor of 1. Slope identifies the maximum rate of change in value from each cell to its neighbors. An output slope grid can be calculated as percent slope or degree of slope. Conceptually, the slope function fits a plane to the z values of a 3x3 cell neighborhood around the processing or center cell. The direction the plane faces is the aspect for the processing cell. The slope for the cell is calculated from the 3x3 neighborhood using the average maximum technique. If there is a cell location in the neighborhood with a no-data z value the z value of the center cell will be assigned to the location. At the edge of the grid, at least three cells (outside the grid’s extent) will contain no-data as their z values. These cells will be assigned the center cell’s z value. The result is a flattening of the 3x3 plane that is fit to these edge cells, which thus usually leads to a reduction in the slope. The actual algorithm that is used to calculate slope is (ESRI, 2001): rise_run = SQRT(SQR(dz/dx)+SQR(dz/dy))degree_slope = ATAN(rise_run) * 57.29578 where the deltas are calculated using a 3x3 roving window.a through i represent the z_values in the window: a b cd e fg h i (dz/dx) = ((a + 2d + g) - (c + 2f + i)) / (8 * x_mesh_spacing)(dz/dy) = ((a + 2b + c) - (g + 2h + i)) / (8 * y_mesh_spacing) The resulting grid was the classified into the discreet classes shown in Table 2.



page 17

Table 2. Slope Classes

Slope Class

Description

1 0 to 5% 2 5% to 25% 3 25% to 50% 4 50% to 70% 5 70% to 100 6 >= 100%

The locational accuracy of the GIS layer was also tested against the 638 plots. The results are displayed below in Table 3. Table 3. Slope Input Layer Confusion Matrix

Field Slope GIS Slope Class 1 Class 2 Class 3 Class 4 Class 5 Class 6 Total Error of

Omission Class 1 24 6 6 0 0 0 36 33.3%Class 2 18 407 49 12 3 0 489 16.8%Class 3 4 19 409 49 15 0 496 20.6%Class 4 1 5 7 145 17 0 175 17.1%Class 5 0 0 1 1 38 0 40 5.0%Class 6 0 0 0 0 0 0 0 0.0%Total 47 437 472 207 73 0 1236 Error of Commission

48.9% 6.9% 13.3% 30.0% 47.9% 0.0%

Average Error of Commission 24.5% Average Error of Omission 15.5% Overall Error: (213/1236) 17.2% Table 4 contains the results of the locational tests. With an overall accuracy of 82.5%, this layer is suitable for use in the PEM process.



page 18

Table 4. Statement of Overall Accuracy Total Plots

1236

Number Correct 1023Percent Correct 82.7%Lower 95% Confidence Value 80%Median 95% Confidence Value

82%

Upper 95% Confidence Value 85% The locational tests used to create the confusion matrix were done using a 50m proximity analysis. Using a 50m proximity analysis is justifiable in the case of the slope layer. In looking at the algorithm used to describe how the delta values were arrived at a 3 X 3 (75m x 75m) window was used which has the effect of “smoothing” the landscape to a degree. The slope described by a given pixel in the Slope Layer is defined by the 75m x 75m area surrounding that pixel. Because of this process a certain amount of shift in classes has to be accounted for. As well, the plot locations describe areas roughly 50m x 50m. This data is stored as point data for simplicity and because no surveying is done for the outer boundaries of the plots. While the point in reality occupies no area in space it is representing an area of about the size of one pixel. The result is that plot locations can fall into the next pixel over from where they are really intended to be. Using a 50m proximity analysis has the effect of compensating for these issues. The plots were first tested using the 50m proximity analysis to see if they were ‘correct’. If the test found the proper corresponding code in the GIS layer, then the plot is ‘right’. If the 50m proximity analysis failed to locate a match, then an ‘on-point’ method is used to determine what was actually found in that plot location. The ‘on-point’ method uses the spatial location of the plot to determine what the GIS layer modeled beneath it. Because using a 50m proximity analysis may return more than one result when testing a layer, it is not possible to use it to determine exactly what is under each plot, but only if the correct values were found within a certain proximity to the plot. A source of error when comparing field determined slope to GIS calculated slope occurs when the field measured slope measurement is near the boundary between two slope classes. For example a field slope of 28% falls into slope class 3 (25-50%), however the GIS determined slope for that pixel may 24% which would put the pixel into slope class 2. That is a problem when continuous data is divided into categories. At the 95% confidence interval (CI) the stated accuracy for this layer is 82% with a lower confidence value of 80% and an upper confidence value of 85%. In examining Table 3, We can see that there is a fairly strong relationship between the field data and the GIS data for slope classes 2, 3, 4 and 6. Classes 1 and 5 display a relatively poor



page 19

relationship. The results would suggest that the GIS is under-predicting the presence of class 1, and under-predicting the presence of class 5. Overall the results of the tests suggest that the Slope Input Data Layer would be appropriate for use in the PEM but that there is room for improvement.

2.2.7.2.2 ASPECT – AS The ArcInfo Grid Aspect command was applied to the Elevation grid. There are no available options for this command. Aspect identifies the direction of maximum rate of change in z value from each cell. Aspect is expressed in positive degrees from 0 to 360, measured clockwise from the north. Aspect identifies the down-slope direction of the maximum rate of change in value from each cell to its neighbors. Aspect can be thought of as the slope direction. The Slope grid was then applied to the Aspect grid. All areas with a slope of less then 25% were given an aspect of neutral. The grid was then reclassified into discrete classes shown below in Table 5. Table 5. Aspect Classes Aspect Class

Description

1 135deg to 285deg (warm) 2 285deg to 135deg (cool) 3 Neutral (< 25% slope) Table 6. Aspect Input Layer Confusion Matrix

Field Aspect GIS Aspect Class 1 Class 2 Class 3 Total Error of

Omission Class 1 343 6 14 363 5.5% Class 2 6 323 13 342 5.6% Class 3 27 24 452 503 10.1% Total 376 353 479 1208 Error of Commission

8.8% 8.5% 5.6%

Average Error of Commission: 7.6% Average Error of Omission: 7.1% Overall error (90/1208): 7.5%



page 20

Table 7. Statement of Overall Accuracy Total Plots

1208

Number Correct 1118Percent Correct 92.6%Lower 95% Confidence Value 91%Median 95% Confidence Value

92%

Upper 95% Confidence Value 94% The aspect layer scored well in all locational tests (see Tables 6 and 7). With an overall error of 7.6%, and low errors in both omission and commission, this layer is suitable for use in the PEM process.

2.2.7.2.3 LANDSCAPE SHAPE - SHP Landscape Shape was modeled using the grid CURVATURE function. Before calculating curvature, the elevation grid was smoothed twice using a 3x3 focal filter. Curvature describes the curvature of the surface at each cell center. A positive curvature indicates that the surface is upwardly convex at that cell. A negative curvature indicates that the surface is upwardly concave at that cell. A value of zero indicates that the surface is flat. Units of the CURVATURE output grid (CURVGRID) are one over 100 z-units, or 1/100(z-units). The curvature of a surface is calculated on a cell-by-cell basis. For each cell, a fourth-order polynomial of the form Z = Ax2y2 + Bx2y + Cxy2 + Dx2 + Ey2 + Fxy + Gx + Hy + I is fit to a surface composed of a 3x3 window. The coefficients a, b, c, and so on, are calculated from this surface. The relationships between the coefficients and the nine values of elevation for every cell numbered are shown below in Figure 3 and are as follows:



page 21

Figure 3. Calculation of Curvature

A = [(Z1 + Z3 + Z7 + Z9) /4 - (Z2 + Z4 + Z6 + Z8) /2 + Z5] /L4B = [(Z1 + Z3 - Z7 - Z9) /4 - (Z2 - Z8) /2] /L3C = [(-Z1 + Z3 - Z7 + Z9) /4 + (Z4 - Z6)] /2] /L3D = [(Z4 + Z6) /2 - Z5] /L2E = [(Z2 + Z8) /2 - Z5] /L2F = (-Z1 + Z3 + Z7 - Z9) /4L2G = (-Z4 +Z6) /2LH = (Z2 - Z8) /2LI = Z5 The output of the CURVATURE function is the second derivative of the surface (i.e., the slope of the slope), such that Curvature = -2(D + E) * 100 The outputs of the initial output curvature grid were from –41.76 to +39.39. The distribution of the values was then examined in ¼ standard deviations shown in Table 8.



page 22

Table 8. Distribution of Curvature Values Max Value Min Value Standard Dev -0.67 -41.76 < -3 Std. Dev -0.61 -0.67 -3.0 to -2.75 Std. Dev. -0.56 -0.61 -2.75 to -2.5 Std. Dev. -0.5 -0.56 -2.5 to -2.25 Std. Dev. -0.45 -0.5 -2.25 to -2.0 Std. Dev. -0.39 -0.45 -2.0 to -1.75 Std. Dev. -0.34 -0.39 -1.75 to -1.5 Std. Dev. -0.28 -0.34 -1.5 to -1.25 Std. Dev. -0.22 -0.28 -1.25 to -1.0 Std. Dev. -0.17 -0.22 -1.0 to -0.75 Std. Dev. -0.11 -0.17 -0.75 to –0.5 Std. Dev. -0.06 -0.11 -0.5 to –0.25 Std. Dev. 0.0 -0.06 -0.25 to 0.0 Std. Dev. 0.0 0.0 Mean 0.06 0.0 0.0 to 0.25 Std. Dev. 0.11 0.06 0.25 to 0.5 Std. Dev. 0.17 0.11 0.5 to 0.75 Std. Dev. 0.22 0.17 0.75 to 1.0 Std. Dev. 0.28 0.22 1.0 to 1.25 Std. Dev. 0.34 0.28 1.25 to 1.5 Std. Dev. 0.39 0.34 1.5 to 1.75 Std. Dev. 0.45 0.39 1.75 to 2.0 Std. Dev. 0.5 0.45 2.0 to 2.25 Std. Dev. 0.56 0.5 2.25 to 2.5 Std. Dev. 0.61 0.56 2.5 to 2.75 Std. Dev. 0.67 0.61 2.75 to 3.0 Std. Dev. 39.39 0.67 > 3.0 Std. Dev After consulting the field staff and field ecologist values were then grouped to describe the shape of the landscape in more general terms which are found in Table 9. Table 9. Landscape Shape Classes Min Value Max Value Curvature Curvature Code -41.76 -0.9 Very Concave 1 -0.9 -0.25 Concave 2 -0.25 0.15 Straight 3 0.15 0.6 Convex 4 0.06 39.39 Very Convex 5



page 23

A Majority Filter was applied twice to the Curvature grid to reduce ‘noise’ or ‘speckling’ in the flatter areas of the landscape. This process replaces the value of each cell in grid based on the majority value of its contiguous neighboring cell. A neighborhood of the eight nearest cells was used. A clear majority needed to be obtained before the center cell value could be replaced with the surrounding value. This may have masked some of the more subtle micro features in the landscape but it was felt that at a 50m pixel size and at a 1:20,000 scale that was acceptable. This layer was not tested due to a lack of comparable field data.

2.2.7.2.4 SLOPE POSITION

The Slope Position layer describes the position of a slope relative to the overall landscape shape. This process was designed to characterize the landscape into six classes: Crest (ridge), Upper Slope, Mid Slope, Lower Slope, Toe Slope, and Valley Bottom. This layer was modeled using an AML downloaded from the Swiss Federal Research Institute WSL website, and was written by Niklaus E. Zimmermann. This simple AML is used to identify topographic exposure (ridge, slope, toe slope, etc) at various spatial scales, and to hierarchically integrate these features into a single grid. Topographic position, necessary e.g. to classify soil properties over large areas, can be calculated using a hierarchically nested approach. In this AML, circular moving-windows with increasing radii are applied to a DEM, and the difference between the average elevation of the window and the center cell of the window is calculated and written to temporary output grids. The user is prompted for the number search radii, for which the temporary grids are generated. The resulting (temporary) maps are interpreted as relative topographic exposure at different spatial scales. The exposure can be interpreted as a ridge or peak if the center cell in the moving window has a higher elevation than the average elevation of the cells in the window. Contrarily if the center cell is of lower elevation than the average elevation of the window, then the center pixel can be interpreted as "toe slope" or "valley bottom". A hierarchical integration into a single map is achieved by starting with the standardized exposure values of the largest window, then adding standardized values from smaller windows where the (absolute) values of the smaller (search) scale grids exceed the values of the larger scale map. When starting the AML, the user is prompted for the name (&path, if different from the directory the AML was started) of the DEM the topographic position has to be derived from. Further, the minimum and maximum radius of the search window has to be entered, as well as the increment to enlarge the search radius stepwise. The AML does increment the search radius (and repeat the exposure calculation for each radius) until the incremented radius is bigger than the maximum radius. It then starts integrating hierarchically the generated temporary grids.



page 24

Two versions of relative topographic exposure are available as a grid, one containing the “raw” values, and a second containing a smoothed version of the topographic exposure. The smoothed version has an "s" added to the output file name. Arguments for the toposcale.aml: Name of the DEM of the study site Name of the output grid Number of pixels for smallest search radius Number of pixels for largest search radius Number of pixels to increment search radius The DEM was smoothed twice using a low pass filter using a focal mean with a 3 by 3 kernel. This generalized the overall values of the DEM, effectively reducing the elevation value range. After the smoothing process, toposcale.aml was run on the DEM. The number of pixels for the smallest search radius was set to 2, with increments of 2 up to the maximum of 10 pixels for the largest search radius. The user has the option of keeping individual exposure grids. Each exposure grids is sampled with a different radii (from the smallest to largest, following the increment path). Each exposure grid was examined for suitability, and the 4 pixel exposure grid was selected. The resulting values are shown in Table 10, stratified by standard deviation.



page 25

Table 10. Distribution of Slope Position Values Min Value Max Value Standard Dev -2814 -299 < -3 Std. Dev -299 -275 -3.00 to –2.75 Std. Dev. -274 -250 -2.75 to –2.50 Std. Dev -249 -225 -2.50 to –2.25 Std. Dev -224 -200 -2.25 to –2.00 Std. Dev -199 -175 -2.00 to –1.75 Std. Dev -174 -150 -1.75 to –1.50 Std. Dev -149 -125 -1.5 to –1.25 Std. Dev -124 -100 -1.25 to –1.00 Std. Dev -99 -75 -1.00 to –0.75 Std. Dev -74 -50 -0.75 to –0.50 Std. Dev -49 -25 -.50 to –0.25 Std. Dev -24 0 -0.25 to 0 Std. Dev 0 0 Mean 1 24 0.00 to 0.25 Std. Dev 25 49 0.25 to 0.50 Std. Dev 50 74 0.50 to 0.75 Std. Dev 75 99 0.75 to 1.00 Std. Dev 100 124 1.00 to 1.25 Std. Dev 125 149 1.25 to 1.50 Std. Dev 150 174 1.50 to 1.75 Std. Dev 175 199 1.75 to 2.00 Std. Dev 200 224 2.00 to 2.25 Std. Dev 225 249 2.25 to 2.50 Std. Dev 250 274 2.50 to 2.75 Std. Dev 275 299 2.75 to 3.00 Std. Dev 299 10894 > 3.00 Std. Dev After consultation with the field ecologists, values from -2814 to -30 were selected for lower slopes, values from –30 to 30 were classified as mid-slope, values from –30 to 10894 were delineated as upper slope. Final slope classes are found below in Table 11. The slope grid was then applied, and areas with a slope of 5% or less were classified as ‘flat’ or ‘valley bottoms’.



page 26

Table 11. Slope Position Classes Min Value Max Value Curvature Curvature Code -2814 -80 Toe 1 -80 -20 Lower 2 -20 20 Mid 3 20 80 Upper 4 80 10894 Crest 5

5% or less slope Valley Bottom 6 Table 12. Slope Position Input Layer Confusion Matrix

Field Slope Position GIS Slope Position

Class 1 Class 2 Class 3

Class 4 Class 5 Class 6 Total Error of Omission

Class 1 50 17 31 1 0 4 103 51.5%Class 2 25 125 51 4 4 4 213 41.3%Class 3 34 26 574 17 16 25 692 17.1%Class 4 3 8 60 76 10 2 159 52.2%Class 5 4 0 34 5 15 1 59 74.6%Class 6 9 3 4 0 3 66 85 22.4%Total 125 179 754 103 48 102 1311 Error of Commission

60.0% 30.2% 23.9%

26.2% 68.8% 35.3%

Average Error of Commission: 40.7% Average Error of Omission: 43.2% Overall Error (405/1311): 30.1% Table 13. Statement of Overall Accuracy of Slope Position Total Plots

1311

Number Correct 906Percent Correct 69.1%Lower 95% Confidence Value

66%

Median 95% Confidence Value

69%

Upper 95% Confidence Value

71%



page 27

The results of the confusion matrix and the overall statement of accuracy are shown in Tables 12 and 13. The relatively high errors of both omission and commission relate to the differences in scale between field data and GIS models. An ecologist in the field may see an area as concave or convex from the ground, while a GIS model may see that same area at a larger scale and call it flat. An ecologist often cannot ‘see’ all of the surrounding terrain, while the GIS model can ‘see’, regardless of trees or hills in the way.

2.2.7.3 INPUT LAYERS GENERATED FROM TRIM TWTR INPUT LAYER

2.2.7.3.1 HYDROGRAPHIC FEATURES – TWTR

A query was run on the TWTR coverage to select any double line water features by feature code (fcode). The fcodes used for the query are shown in Table 14. Table 14. TRIM 1 Feature Codes and Corresponding Hydrographic Features

Feature Line Code

TRIM 1 Description

Polygon Feature Polygon Code

GA90000110

Stream/River - Left Bank

River 2

GA90000120

Stream/River - Right Bank

River 2

GB15300000

Lake – Definite Lake >= 50 ha. 1

GB15300000

Lake – Definite Pond <= 50 ha. 9

GB15300130

Lake – Indefinite Shallow Open Water 10

GB15300140

Lake – Intermittent Shallow Open Water 10

GB24300000

Reservoir – Definite Reservoir 3

GE25850000

Sand Bar Sand Bar 5

GC17100000

Marsh Marsh 6

GC30050000

Swamp Swamp 7

GD12300000

Glacier Glacier 8

GE14850000

Island – Definite Island 4



page 28

The features were then converted to polygons using the ArcInfo 8.1 CLEAN procedure. Line work was then compared to the originals to check to insure that no large shifts had taken place. The command used to clean the coverage was CLEAN <in_cover> {out_cover} {dangle_length} {fuzzy_tolerance} {POLY | LINE} In_cover was the cover selected for cleaning, out_cover was the output coverage. The dangle length was the acceptable tolerance for dangles. This was set to 25. Fuzzy tolerance was set to 2.5m. The polygon option was selected to convert the data from line to polygon. The resulting polygons were then given a numeric value associated with the feature described. This polygon coverage was then converted to a 50m grid using the ArcInfo 8.1 GRIDPOLY command with the MASK grid defined as the Mask. Grid values were defined using the code value assigned to the polygons.

2.2.7.4 INPUT LAYERS GENERATED FROM FOREST COVER

2.2.7.4.1 ALPINE FOREST - AF Using ArcInfo 8.1, the spatial forest cover database was queried for any non-productive polygons with a NPFORESTCODE of 10. These areas represent forested alpine areas. An attribute called ‘AF’ was added to the database, and all selected polygons were coded with a ‘1’ as shown below in Table 15. Table 15. Alpine Forest Areas Classes Description Code Alpine Forest 1 Not Alpine Forest

0

The ArcInfo 8.1 DISSOLVE command was applied to the forest cover data using the UR attribute. The resulting polygonal coverage describes the presence of open range areas. The ArcInfo 8.1 POLYGRID command was used to convert the coverage to a grid. POLYGRID <in_cover> <out_grid> {value_item} {lookup_table} {weight_table} The available field data did not contain any non-productive forest information, so a statistical analysis was not carried out.



page 29

2.2.7.4.2 FOREST COVER ROCK INPUT LAYER – FC_ROCK

Using ArcInfo 8.1, the spatial forest cover database was queried for any polygon with an NPFORESTCODE of 3 (rock). An attribute called FC_ROCK was added to the forest cover database, and all selected polygons were coded with a ‘1’ as shown below in Table 16. Table 16. Forest Cover Rock Input Layer Classes Description Code Forest Cover Rock 1 No Forest Cover Rock

0

The ArcInfo 8.1 DISSOLVE command was applied to the forest cover data using the FC_ROCK attribute. The resulting polygonal coverage describes the presence of rock within the forest cover data. The ArcInfo 8.1 POLYGRID command was used to convert the coverage to a grid. POLYGRID <in_cover> <out_grid> {value_item} {lookup_table} {weight_table} The available field data did not contain any non-forest information, so a statistical analysis was not carried out.

2.2.7.4.3 URBAN AREAS – UR Using ArcInfo 8.1, the spatial forest cover database was queried for any non-productive polygons with a NPFORESTCODE of 50 or 54. These areas represent urban settlements and large road corridors. An attribute called ‘UR’ was added to the database, and all selected polygons were coded with a ‘1’ as shown below in Table 17. Table 17. Urban Areas Classes Description Code Urban Area 1 Not Urban Area

0

The ArcInfo 8.1 DISSOLVE command was applied to the forest cover data using the UR attribute. The resulting polygonal coverage describes the presence of open range areas. The ArcInfo 8.1 POLYGRID command was used to convert the coverage to a grid.



page 30

POLYGRID <in_cover> <out_grid> {value_item} {lookup_table} {weight_table} The available field data did not contain any non-forest information, so a statistical analysis was not carried out.

2.2.7.4.4 ALPINE ECOSYSTEMS - ALP Using ArcInfo 8.1, the spatial forest cover database was queried for any polygon with an NPFORESTCODE of 2 (alpine). An attribute called ALP was added to the forest cover database, and all selected polygons were coded with a ‘1’ as shown below in Table 18. Table 18. Alpine Ecosystem Input Layer Classes Description Code Alpine Ecosystem 1 Not Alpine Ecosystem

0

The ArcInfo 8.1 DISSOLVE command was applied to the forest cover data using the ALP attribute. The resulting polygonal coverage describes the presence of alpine ecosystems within the forest cover data. The ArcInfo 8.1 POLYGRID command was used to convert the coverage to a grid. POLYGRID <in_cover> <out_grid> {value_item} {lookup_table} {weight_table} The available field data did not contain any non-forest information, so a statistical analysis was not carried out.

2.2.7.5 AVALANCHE PATH AND RUNOUT ZONE INPUT LAYER - AV Using ArcInfo 8.1, the spatial forest cover database attribute NPFORESTCODE was queried for Alpine (2), Rock (3), Alpine Forest (10), and Non-Productive Brush (11). A new attribute called ‘AV_NPF’ was added to the forest cover database, and all selected polygons were coded with a ‘1’. Next, all intermediate and indefinite streams from the TWTR layer were buffered by 50 meters using the ArcInfo BUFFER command. BUFFER <in_cover> <out_cover> {buffer_item} {buffer_table} {buffer_distance} {fuzzy_tolerance} {LINE | POLY | POINT | NODE} {ROUND | FLAT} {FULL | LEFT | RIGHT} <in_cover> - the coverage containing features to be buffered.



page 31

<out_cover> - the coverage to be created. {buffer_distance} - the distance used to create buffer zones around <in_cover>

features when {buffer_item} or {buffer_item} and {buffer_table} are not specified. The default distance is 0.125 coverage units. This default buffer distance will be applied whenever this argument is skipped with a '#', omitted, or a distance of 0 is specified. The smallest buffer distance that can be computed is 0.00000005 coverage units. Specifying a buffer distance below this threshold will result in an empty output cover. For polygon features, if a negative buffer distance is used, buffers will be generated on the insides of polygons. This was set to 50.

{fuzzy_tolerance} - the minimum distance between coordinates in the output coverage.

A value of 0 will not be accepted. This was set to 2.5. The output coverage was called ‘STRMINTBUF_50’. Both coverages (STRMINTBUF_50 and AV_NPF) were converted to a GRID format using the ArcInfo 8.1 POLYGRID command, using a weed tolerance of 5. POLYGRID <in_cover> <out_grid> {value_item} {lookup_table} {weight_table} After being converted to a GRID format, a query was run (using ArcView 3.1 Spatial Analyst) on three grids: SLP, STRMINTBUF_50, and AV_NPF. The query was designed to select areas with the following combination of attributes: • The pixel was within 50 meters of an intermediate stream • The pixel had an NPFORESTCODE of 2, 3, 10, or 11 • The pixel had a SLP code of 3, 4, 5, or 6. The resulting query was saved as a temporary grid. Another query was constructed using AV_NPF and the Biogeoclimatic Subzone input layer. First, the Biogeoclimatic Subzone input layer was converted to a GRID called BEC_GRID using the ArcInfo 8.1 POLYGRID command and a weed tolerance of 5. POLYGRID <in_cover> <out_grid> {value_item} {lookup_table} {weight_table} BEC_GRID and AV_NPF and SLP were then queried for a combination of the following attributes: • A BEC Zone of:



page 32

• ESSF dc 1 • ESSF vc • ESSF wc 1 • ESSF wc 4 • ICH dw • ICH mw 2 • ICH wk 1 • ICH vk 1 • ICH xw An NPFORESTCODE of: • 2 (Alpine) A SLP of: • 3, 4, 5, 6 This query was designed to pick out the steeper, high elevation avalanche paths, which are often coded as Alpine Forest by photo-interpreters. The results of the query were then visually compared to the Landsat image for obvious coinciding avalanche chutes. While no statistical analysis is possible due to lack of appropriate field data, the vast majority of the GIS modeled avalanche zones were directly over their corresponding avalanche path on the Landsat image. The model is further refined in this case using a SLP of 3, 4, 5, or 6. This had the effect of only modeling the steeper avalanche paths/chutes, leaving the flatter portions to be modeled as runout zones. Another query was designed to model the runout zones at the end of the avalanche paths. The query was structured as follows: • A BEC Zone of: • ESSF dc 1 • ESSF vc • ESSF wc 1 • ESSF wc 4 • ICH dw • ICH mw 2 • ICH wk 1 • ICH vk 1 • ICH xw An NPFORESTCODE of: • 2 (Alpine)



page 33

A SLP of: • 1 or 2 These areas were situated at the bottom of avalanche chutes, and typically contain a much higher diversity of vegetation that supports a wide variety of wildlife, including Grizzly bears. Runout zones were modeled only within 200 meters of an avalanche chute. The results of the query were then visually compared to the Landsat image for obvious coinciding avalanche runout zones. While no statistical analysis is possible due to lack of appropriate field data, the vast majority of the GIS modeled avalanche runout zones were directly over their corresponding avalanche runout zone on the Landsat image. 3.0 KNOWLEDGE BASE DEVELOPMENT AND TESTING The PEM model uses the spatial input layers and attaches a potential site series, or combination of site series to the combination of input layer attributes. Those results are modified by secondary spatial attributes such as distance from water, depth of materials and type of materials. 3.1 FIRST RUN RASTER SITE SERIES ALLOCATION We have implemented an improved method of site series allocation within the knowledge tables that is based on GIS derived landscape position variable combinations that are key to delineation of the site series based mapping entities. It replaces the numerical scoring approach used in the original Arrow PEM.



page 34

Initially we query the input layers and derive all combinations of the following variables: Landscape position

Shape Slope class

Aspect

Toe Very concave 1 (0 to 5%) Gentle – j Lower Concave 2 (5 to 25%) Warm – w Mid Straight 3 (26-50%) Cool - k Upper Convex 4 (51-75%) Crest Very convex 5 (75 – 100% Level 6 (>100%) The area of each BEC variant that falls within each combination was determined. These tables (pivot tables) can be found in Appendix 1. The model building plot data was also summarized into the same variable combination categories, but also includes the field based BEC variant and site series call as an additional variable. The plot data is sorted by site series and the frequency of the each variable combination is noted (Appendix 4). A site series is then assigned to the variable combination based on its frequency of occurrence within that site series as indicated by the field data (Appendix 2). When more than one site series occur on a particular combination a map entity is assigned to that combination and a probability of each site series is assigned (Appendix 3). The proportion of site series within a mapping entity is based on the frequency of occurrence of that site series within the variable combination. In this way every potential landscape position is allocated a site series or a mapping entity that consists of proportions of site series observed in the field to occur on that unique combination of slope position, shape, slope class and aspect. This forms the basis of the first run of the raster PEM. The knowledge bases that assign each variable combination to a site series or mapping entity can be found in Appendix 2. The first run raster PEM is now ready for second run modifications using secondary queries using specific TRIM attributes and the vector format bioterrain and forest cover data Appendix 3). 3.2 SECOND RUN RASTER MODIFICATIONS Within many landscape shape variable combinations the data indicated that wetter site series seem to generally occur within certain distances of all water bodies except intermittent creeks, depending on the slope. A set of rules relating to slope and distance from water were derived. These rules are based on the following combination of categories.



page 35

Slope Class Distance From Water S1 (0-5%) >25 m. S1 (0-5%) 25 – 50 m. S2 (6-25%) >25 m. S2 (6-25%) 25 – 50 m. Within each BEC variant site series are allocated to the above slope class – distance from water combinations and this is used to replace the output from the landscape shape first run raster output within the distance from the water bodies on the appropriate slope classes. This step increases our ability to find wetter sites on flat and gentle terrain.

3.2.1 SECOND RUN TRIM WATER FEATURE MODIFICATIONS TRIM data contains feature codes relating to water bodies, glaciers and gravel bars. Some of these features are also useful in modification of the initial landscape shape driven first run raster model. The TRIM feature codes used in the 2003 Arrow PEM model were as follows: TRIM – swamps TRIM – marshes TRIM – glaciers TRIM – rivers TRIM – ponds TRIM – shallow open water TRIM – reservoirs TRIM – gravel bars These features are used in to identify wetlands, glaciers, ponds, lakes, reservoirs and gravel pits mapped by TRIM. The knowledge bases for each BEC variant allocate these features to the appropriate site series or approved two letter codes for non-vegetated areas. See Appendix 3 for the second run rule sets for TRIM water features and non-vegetated areas.

3.2.2 SECOND RUN VECTOR BIOTERRAIN MODIFICATIONS Bioterrain mapping was completed for the 2000 delineation of the Timber Harvesting Land Base of the Arrow TSA as part of the 2001 PEM project. The drainage portion of that data was converted to raster format and used as part of the input layers in the original PEM process. We have utilized more aspects of the Arrow 2001 bioterrain data and facilitated a vector based second run using the 2001 bioterrain polygons within the 2000 delineation of the THLB in the 2002 iteration of the Arrow PEM. This enables us to make better use of the



page 36

materials and surface expression data and increased our accuracy in certain types of sites by accounting for rock outcroppings and sites with thin materials that potentially occupy small areas within the bioterrain polygons. This greatly improves the accuracy of the output relative to rock and dry site series, which were overestimated using Landsat Imagery in the 2001 Arrow PEM. It also resulted in a deciled output of mapping entities in the TEM data base like format, which is the preferred format for QA personnel. The second run vector-based rule sets used for each BEC variant can be found in Appendix 3.

3.2.3 SECOND RUN VECTOR FOREST COVER MODIFICATIONS Information from the forest cover data bases was compiled to depict forest cover rock and urban areas. Forest cover rock designations were used in combination with the bioterrain rock delineation to further refine the allocation of rock and talus on the landscape. The original Arrow PEM model used Landsat imagery to delineate rock. However, the Landsat layer tended to overestimate rock by including dry, bare soil into its rock classification. Using bioterrain and forest cover rock designations improved the PEM rock category in the Timber Harvesting Landbase where bioterrain and accurate depiction of forest cover rock is available to the model. 3.3 PEM MODEL OVERVIEW An overview of the PEM model used for the Arrow 2003 is depicted in Figure 4. It basically consists of four stages where the landscape shape derived raster result is modified by spatial attributes from the bioterrain data which is in a deciled, vector (polygon) format. This approach was very successful in the Okanagan PEM (Ketcheson et al 2002).



page 37

Figure 4. Arrow 2003 PEM Model Overview.



page 38

4.0 RESULTS 4.1 PEM MODEL MAP ENTITY ALLOCATION BY BEC VARIANT The 2003 Arrow PEM model allocated both the Timber Harvesting Landbase and the Non-Timber Harvesting Landbase to a site series or map entity. The map entities, their area and proportional representation within the BEC variant are reported in Table 19. These map entities are derived from the site series classification of Braumandl and Curran (1992), Lloyd et al (1990), Marcoux et al (1996, 1997). The site series classification of BEC variants not described in either of the above publications were derived from the original Arrow PEM. These units are found mostly in parkland and upper BEC variants not routinely described to the site series level by the published Land Management Handbooks in either the Nelson or Kamloops Forest Regions. There is a degree of overlap between site series on some landscape positions, especially in the ICHmw2 and ICHdw, that makes it difficult to accurately predict which site series will occur. Combination units (complex mapping entities consisting of two or more site series) were initially assigned to polygons on these landscape positions. Based on sampling data and information from Braumandl and Curran (1992), rule sets were developed to allocate the proportional areas of individual site series within the polygons labeled with combination units. These rules can be found in Appendix 3. The individual map entities (site series), their areas and proportional representation within the BEC subzone/variant in which they occur are reported in Table 19.



page 39

Table 19. Arrow TSA 2003 PEM Results Subzone Site

Series #

Map Symbol

Site Series Name Area (hectares)

Percentage of BEC variant

ATun 00 AH Alpine heath 331.83 16.1% ATun GL Glacier/ Permanent

snowfield 27.07 1.3%

ATun KH Krumholtz 14.40 0.7% ATun PD Pond 0.06 <0.01% ATun RO Rock outcrop (including

Talus) 1693.51 81.9%

TOTAL AT 2066.89 100.00% ESSFdc1 00 AR Avalanche runout zone 47.83 1.1% ESSFdc1 00 AC Avalanche chute 186.33 4.4% ESSFdc1 02 EP PlSe – Pinegrass 385.82 9.1% ESSFdc1 03 FG Bl – Grouseberry - Cladonia 1568.34 37.1% ESSFdc1 06 FH Bl – Horsetail – Glow moss 34.00 1.0% ESSFdc1 01 FR Bl – Rhododendron -

Grouseberry 1532.97 36.3%

ESSFdc1 05 FT Bl – Trapper’s tea 230.86 5.5% ESSFdc1 00 PD Pond 0.80 <0.01% ESSFdc1 RO Rock outcrop 48.00 1.1% ESSFdc1 04 RV Bl – Rhododendron –

Valerian 185.32 4.4%

ESSFdc1 07 SS Sedge Sphagnum 5.57 0.1% TOTAL ESSFdc1

4225.84 100.00%

ESSFdcp1 02 FH Bl –Heath 4.13 4.0% ESSFdcp1 03 JM Juniper – Mountain

hairgrass 17.06 16.4%

ESSFdcp1 01 MH Mountain-heather 5.25 5.0% ESSFdcp1 RO Rock outcrop (including

Talus) 77.69 74.6%

TOTAL ESSFdcp1 104.13 100.00% ESSFwc1 00 AR Avalanche runout zone 249.08 0.4% ESSFwc1 00 AC Avalanche chute 2442.73 3.6% ESSFwc1 03 FD Bl - Devil's club - Lady fern 6697.45 10.0% ESSFwc1 02 FF Bl – Falsebox - Grouseberry 22851.35 34.1% ESSFwc1 04 FH Bl – Horsetail -

Brachythecium 338.82 0.5%

ESSFwc1 01 FR Bl - Rhododendron - Oak fern

33406.32 49.8%

ESSFwc1 GB Gravel/ sand bar 0.19 <0.0% ESSFwc1 OW Shallow open water 0.91 <0.0% ESSFwc1 PD Pond 32.83 0.5% ESSFwc1 RO Rock outcrop 929.75 1.4% ESSFwc1 05 SS Sedge - Sphagnum 33.67 0.1% ESSFwc1 TA Talus 4.44 <0.0% ESSFwc1 UR Urban/ suburban

Includes transportation corridors 116.81 0.2%



page 40

TOTAL ESSFwc1

67104.33 100.00%

ESSFwc4 00 AR Avalanche runout zone 0.0 0.0% ESSFwc4 00 AC Avalanche chute 7789.4 5.62% ESSFwc4 02 FF Bl - Rhododendron -

Falsebox 33159.0 23.92

ESSFwc4 06 FH Bl - Horsetail - Brachythecium

43.5 0.03%

ESSFwc4 05 FL Bl - Rhododendron - Lady fern

3950.3 2.85%

ESSFwc4 01 FR Bl - Rhododendron - Oak fern

55142.2 39.77%

ESSFwc4 03 FW Bl - Rhododendron - Woodrush

10811.8 7.8%

ESSFwc4 04 RF Bl - Rhododendron - Foamflower

27702.4 19.98%

ESSFwc4 00 TA Talus 32.62 <0.0% ESSFwc4 08 WS Willow Sedge 0.00 <0.0% ESSFwc4 RO Rock outcrop 15.30 <0.0% TOTAL ESSFwc4

138,646.59 100.00%

ESSFwcp4 02 FH Bl - Heath 1119.00 5.0% ESSFwcp4 03 JM Juniper – Mountain

hairgrass 2216.85 10.0%

ESSFwcp4 00 KR Krummholz 4222.98 18.8% ESSFwcp4 01 MH Mountain-heather 2018.09 9.0% ESSFwcp4 OW Shallow open water 0.25 0.0% ESSFwcp4 PD Pond 32.61 0.2% ESSFwcp4 RO Rock outcrop (including

Talus) 12576.71 56.0%

ESSFwcp4 04 SW Sedge – Western pasqueflower

241.07 1.1%

ESSFwc4 00 TA Talus 12.71 0.01% ESSFwcp4 00 WL Wetland 2.50 0.0% TOTAL ESSFwcp 22443.09 100.00% ICHdw 02 DO FdPy – Oregon-grape –

Parsley fern 12503.36 9.8%

ICHdw GB Gravel/ sand bar 247.05 0.2% ICHdw 03 HD CwHw – White pine – Devil’s

club 5855.93 4.6%

ICHdw LA Lake 15515.18 12.2% ICHdw OW Shallow open water 8.62 <0.0% ICHdw PD Pond 82.68 0.1% ICHdw 04 RD CwHw – Devil’s club – Lady

fern 1716.21 1.3%

ICHdw RI River 1017.38 0.8% ICHdw RE Reservoir 3.48 0.0% ICHdw RO Rock outcrop 6329.13 4.9% ICHdw UR Urban/ suburban 1448.98 1.1% ICHdw 00 WL Wetland 125.26 0.1% ICHdw 01a XF

RFa CwFd – Falsebox (sx-sm phase)

44719.13 34.94%



page 41

ICHdw 01b XG RFb

CwFd – Falsebox (m-shg phase)

38418.03 30.0%

TOTAL ICHdw 127995.85 100.00% ICHmw2 00 AR Avalanche runout zone 342.79 0.2% ICHmw2 00 AC Avalanche chute 2920.25 1.3% ICHmw2 09 BS Bluejoint – Sedge 172.06 0.1% ICHmw2 GB Gravel bar 196.45 0.1% ICHmw2 03 DF FdCw – Falsebox – Prince’s

pine 55112.59 23.6%

ICHmw2 01 HF HwCw – Falsebox – Feathermoss

61253.14 26.2%

ICHmw2 05 HO CwHw – Oakfern – Foamflower

12189.71 5.2%

ICHmw2 OW Shallow open water 0.38 0.0% ICHmw2 00 LA Lake 34660.35 14.8% ICHmw2 00 PD Pond 100.66 <0.0% ICHmw2 02 RC Rhacomitrium Cladonia 209.17 0.1% ICHmw2 06 RD CwHw – Devil’s club – Lady

fern 7478.93 3.2%

ICHmw2 04 RF CwFd – Falsebox 51304.66 22.0% ICHmw2 07 RH CwHw – Horsetail 1118.69 0.5% ICHmw2 RI River 110.27 0.1% ICHmw2 RO Rock outcrop 2916.79 1.3% ICHmw2 00 TA Talus 14.29 <0.0% ICHmw2 08 RS CwSxw – Skunk cabbage 2602.62 1.1% ICHmw2 UR Urban/ suburban 856.94 0.4% TOTAL ICHmw2 233560.73 100.00% ICHvk1 00 AR Avalanche runout zone 20.08 11.4% ICHvk1 00 AC Avalanche chute 55.66 31.7% ICHvk1 03 HF HwCw – Falsebox –

Feathermoss 9.71 0.5%

ICHvk1 04 HS CwHw – Oak fern Spiny woodfern

24.63 14.0%

ICHvk1 06 RC CwSxw – Skunk cabbage 5.09 2.9% ICHvk1 01 RD CwHw – Devil’s club – Lady

fern 60.32 34.37%

ICHvk1 05 RH CwSxw – Devil’s club Horsetail

62.62 0.6%

TOTAL ICHvk1 238.12 100.00% ICHwk1 00 AR Avalanche runout zone 54.57 0.5% ICHwk1 00 AC Avalanche chute 1011.11 9.9% ICHwk1 07 CD Act – Dogwood – Twinberry 254.56 2.5% ICHwk1 04 HF HwCw – Falsebox –

Feathermoss 1672.04 16.4%

ICHwk1 01 HO CwHw – Oak fern 4131.18 40.4% ICHwk1 LA Lake 652.80 6.4% ICHwk1 PD Pond 10.20 0.1% ICHwk1 08 RC CwSxw – Skunk cabbage 106.03 1.0% ICHwk1 05 RD CwHw – Devil’s club –

Lady’s fern 1928.81 18.9%

ICHwk1 06 RH CwSxw – Devil’s club – Horsetail

139.57 1.4%

ICHwk1 RI River 14.17 0.1%



page 42

ICHwk1 00 GB Gravel bar 7.18 0.1% ICHwk1 00 TA Talus 1.90 <0.0 ICHwk1 RO Rock outcrop 181.24 1.8% TOTAL ICHwk1 10165.36 100.00% ICHxw #Site Series Classification based on Marcoux et al 1996

02 DN# FdPy – Mallow ninebark – Bluebunch wheatgrass

40.38 <0.0%

ICHxw 03 DO FdPy – Oregon grape – Birch-leaced spirea

785.70 21.9%

ICHxw 01 DP Fd – Prince’s pine 1384.95 38.62% ICHxw GB Gravel/ sand bar 1.40 <0.0% ICHxw PD Pond 0.32 <0.0% ICHxw 05 HS CwHw – Prince’s pine – Wild

sarsaparilla 805.56 22.5%

ICHxw OW Shallow open water 0.56 <0.1% ICHxw 06 RD CwHw – Devil’s club – Oak

fern 329.33 9.2%

ICHxw RI River 18.81 0.5% ICHxw RO Rock outcrop 11.12 0.3% ICHxw UR Urban/ suburban 207.23 5.8% ICHxw 00 WL Wetland 0.35 <0.0% TOTAL ICHxw 3585.70 100.00% IDFun &Site series classification based Marcoux et al 1997

02 BJ& Bluebunch wheatgrass - Junegrass

25.80 1.2%

IDFun 05 FS Fd - Snowberry 345.20 16.3% IDFun 03 FW FdPy - Bluebunch

wheatgrass - Junegrass 39.25 1.9%

IDFun GB Gravel bar 2.41 0.1% IDFun LA Lake 1409.01 66.5% IDFun 04 PS PyFd - Snowberry 153.93 7.27% IDFun RO Rock outcrop 0.21 <0.0% IDFun 06 RH Cw – Hooker’s fairybells 137.26 6.5% IDFun UR Urban/ suburban 5.66 0.3% TOTAL IDFun 2118.74 100.00%



page 43

4.2 GOODNESS OF FIT AND ACCURACY ASSESSMENT A set of 200 randomly collected plots was sampled within a 1 km buffer of the TRIM roads within the Arrow TSA. This plot data represented 90 polygons within the 1km buffer. The PEM output was also buffered to only include all the results within one kilometer of the TRIM roads. These results were compared to the randomly collected plot data using the Meidinger (2000) protocol for the assessment of accuracy of ecosystem maps.

4.2.1 CHI-SQUARED TEST OF ECOSYSTEM UNIT PROPORTIONS

4.2.1.1 ESSF wc1 TEST OF ECOSYSTEM UNIT PROPORTIONS

Of the plots used for this accuracy assessment, 29 fell within the ESSF wc1. The distribution of those plots is described below in Table 20.

Table 20. Distribution of Site Series in ESSF wc1 Plots – Accuracy Data Map Entities Site Series PlotsESSF wc1 FR ESSF wc1 01 17ESSF wc1 FF ESSF wc1 02 8ESSF wc1 FD ESSF wc1 03 4Grand Total 29 Within the sample area there were 48703.88 ha of ESSF wc1 within a 1km buffer of TRIM roads. The distribution of site series within the buffer is described below in Table 21. Table 21. Distribution of Map Entities and Site Series in the ESSF wc1 PEM Polygons Map Entity Site Series Area (ha) AC ESSF wc1 00 1425.27AR ESSF wc1 00 160.09GB ESSF wc1 00 0.19OW ESSF wc1 00 0.70PD ESSF wc1 00 13.44RO ESSF wc1 00 541.48TA ESSF wc1 00 3.47UR ESSF wc1 00 111.73FR ESSF wc1 01 23877.39FF ESSF wc1 02 16813.93FD ESSF wc1 03 5424.63FH ESSF wc1 04 301.17SS ESSF wc1 05 30.39Grand Total 48703.88 In order to compare the distributions of the two data sets, distinct classes had to be generated. In the ESSF wc1 01 there were enough samples in both the plot and PEM



page 44

data sets to warrant a single class. The ESSF wc1 02 is similar. The 03,04 and 05 site series are all grouped together for several reasons. Within the plot data there were no 04 or 05 site series. The PEM data for the sample area shows that these two site series make up less than 1% of the subzone. With the PEM showing so little of these site series it suggests a similar distribution to the plot data. However, in order to show this similarity, these site series had to be grouped with another class in an attempt to bring the Expected Frequency to over 5. The total Expected Frequency for the ESSF wc1 03/04/05 class is still only 3.6 but this should be adequate for the purposes of this test. The edatopic grid for the ESSF wc1 also suggests that this grouping is appropriate. All of the subhygric to subhydric site series (03,04,05) are grouped together. The results of the Chi Squared test are shown below in Table 22. Table 22. Chi Squared Test of Ecosystem Proportions in the ESSFwc1 – Accuracy Data

Site Series Observed Expected (Observed-Expected)2

Expected Freq Prop Area Prop Freq

ESSF wc1 01 17 58.6% 23877.39 53% 14.9 0.293534ESSF wc1 02 8 27.6% 16813.93 36% 10.5 0.594382ESSF wc1 03/04/05 4 13.8% 5756.19 12% 3.6 0.045879Total 29 100.0% 46447.51 100% 29.0 0.933795 Chi-Square = .9337 Degrees of Freedom = 2 P-Value = .6269 With a P-value of .6269 this test passes, exceeding the alpha value of .1. A similar test using all of the model building plots for the ESSF wc1 produces a less positive result but is still considered a pass. Table 23 below contains the distribution of site series within the entire sample data set. Table 23. Distribution of Site Series in ESSF wc1 Plots – Model Building Data SITEMC_S1 Num Total ESSFwc1 FR ESSFwc1 01 84ESSFwc1 FF ESSFwc1 02 71ESSFwc1 FD ESSFwc1 03 23ESSFwc1 SS ESSFwc1 05 2Total 180



page 45

Table 24. Chi Squared Test of Ecosystem Proportions in the ESSFwc1 – Model Building Data

Site Series Observed Expected (Observed-Expected)2 Expected

Freq Prop Area Prop Freq ESSF wc1 01 84 46.7% 23877.39 51% 92.5 0.786886ESSF wc1 02 71 39.4% 16813.93 36% 65.2 0.523462ESSF wc1 03/04/05 25 13.9% 5756.19 12% 22.3 0.325058Total 180 100.0% 46447.51 100% 180.0 1.635406 Chi-Square = 1.635 Degrees of Freedom = 2 P-Value = .4414 With a P-value of .4414 this test passes, exceeding the alpha value of .1.

4.2.1.2 ESSF wc4 TEST OF ECOSYSTEM UNIT PROPORTIONS

Of the plots used for this accuracy assessment, 27 fell within the ESSF wc4. The distribution of those plots is described below in Table 25.

Table 25. Distribution of Site Series in ESSF wc4 Plots – Accuracy Data Map Entities Site Series Count FR ESSF wc4 01 21 FF ESSF wc4 02 1 FW ESSF wc4 03 1 RF ESSF wc4 04 2 FL ESSF wc4 05 1 SS ESSF wc4 07 1 Grand Total 27 Within the sample area there were 71,886 ha of ESSF wc4 within a 1km buffer of TRIM roads. The distribution of map entities and their corresponding site series within the buffer is described below in Table 26.



page 46

Table 26. Distribution of Site Series in the ESSF wc4 PEM Polygons

Map Entity Site Series Area (ha) ESSFwc4 AC 00 2836.17ESSFwc4 AR 00 0.0007ESSFwc4 FF 02 16837.06ESSFwc4 FH 06 22.76ESSFwc4 FL 05 1683.21ESSFwc4 FR 01 30178.91ESSFwc4 FW 03 5572.85ESSFwc4 RF 04 14737.15ESSFwc4 RO 00 5.86ESSFwc4 TA 00 12.81

Grand Total 71,886.82 In order to compare the distribution of the plots (observed data) to the PEM (expected) data, some grouping of classes had to be done. There were no plots that were 06 or 08. This meant that these classes had to be merged with others. Site series 02 and 03 were grouped because due to size and because these are the drier ecosystems on the edaptopic grid. The 05,06,07, and 08 were all summed together to produce a group representing the wetter site series on the edaptopic grid. Because the 01 and 04 site series occupy the same location on the edatopic grid these two site series were grouped. The results of the Chi Squared test are shown below in Table 27.

Table 27. Chi Squared Test of Ecosystem Proportions in the ESSFwc4 – Accuracy Data

Site Series Observed Expected (Observed-Expected)2 Expected

Frequency Proportion Area Proportion Frequency ESSF wc4 01,04 23 82.1% 44916.06 65% 18.2 1.25ESSF wc4 02,03 3 10.7% 22409.92 32% 9.1 4.08ESSF wc4 05,06,07,08 2 7.1% 1705.98 2% 0.7 2.47Total 28 100.0% 69031.96 100% 28.0 7.81 Chi-Square =7.81 Degrees Freedom = 2 P-Value = .0201 The p-value of .0201 is lower than the alpha value of .1; this test should be regarded as a failure. A similar test using all of the model building plots for the ESSF wc4. Table 28 below contains the distribution of site series within the entire sample data set.



page 47

Table 28. Distribution of Site Series in ESSF wc4 Plots – Model Building Data Plots Num Count ESSFwc4 FR 01 36ESSFwc4 FF 02 15ESSFwc4 FW 03 15ESSFwc4 RF 04 23ESSFwc4 FL 05 1ESSFwc4 SS 07 2Total 92 Table 29. Chi Squared Test of Ecosystem Proportions in the ESSFwc4 – Model Building Data

Site Series Observed Expected (Observed-Expected)2

Expected Freq Prop Area Proportion Frequency

ESSF wc4 01,04 59 64% 44916.06 65.1% 59.9 0.012ESSF wc4 02, 03 30 33% 22409.92 32.5% 29.9 0.001ESSF wc4 05,06,07,08 3 3% 1705.98 2.5% 2.3 0.232Total 92 100.0% 69031.96 100% 92.0 0.245

Chi-Square =.245 Degrees Freedom = 2 P-Value = .8846 The p-value of .8846 is greater than the alpha value of .1; this test should be regarded as a success and a near perfect fit.

4.2.1.3 ICH dw TEST OF ECOSYSTEM UNIT PROPORTIONS Of the plots used for this accuracy assessment, 60 fell within the ICH dw. The distribution of those plots is described below in Table 30. Table 30. Distribution of Site Series in ICH dw Plots – Accuracy Data Map Enitities Site Series TotalRFa ICH dw 01a 31RFb ICH dw 01b 15DO ICH dw 02 13HD ICH dw 03 1Grand Total 60 Within the sample area there were 97,897 ha of ICH dw within a 1km buffer of TRIM roads. The distribution of site series within the buffer is described below in Table 31.



page 48

Table 31. Distribution of Site Series in the ICH dw PEM Polygons Map Entity Site Series Total GB ICH dw 00 245.96LA ICH dw 00 7421.98OW ICH dw 00 2.50PD ICH dw 00 68.79RE ICH dw 00 3.48RI ICH dw 00 1017.38RO ICH dw 00 5443.4TA ICH dw 00 5.17UR ICH dw 00 1419.43WL ICH dw 00 117.51XF ICH dw 01A 36207.25XG ICH dw 01B 31178.38DO ICH dw 02 9758.39HD ICH dw 03 4868.05RD ICH dw 04 1588.04Grand Total 97897.79 In order to compare the plot (observed) data to the PEM (expected) data several classes had to be grouped. The 02 and RO were grouped because they represent the drier site series on the edaptopic grid. In the ICH dw the 02 site series is sparsely vegetated and is of interspersed with rocky outcrops. The 03 and 04, site series were grouped in order to form another class representing the wetter site series on the edatopic grid. . Table 32 shows the results of the Chi Squared test.

Table 32. Chi Squared Test of Ecosystem Proportions in the ICH dw – Accuracy Data

Observed Expected Site Series Frequency Proportion Area Proportion Frequency

(Observed-Expected)2 Expected

ICH dw 01a 31 51.67% 36207.25 41.33% 24.8 1.55ICH dw 01b 15 25.00% 31178.38 35.59% 21.4 1.89ICH dw 02/RO 13 21.67% 13753.86 15.70% 9.4 1.36ICH dw 03,04 1 1.67% 6456.09 7.37% 4.4 2.65Total 60 100.00% 87595.58 100.00% 60.0 7.45 Chi-Square = 7.45 Degrees Freedom = 3 P- Value = .0588 The P-Value of .0588 is lower than the alpha value of .1. This test should be considered a failure.



page 49

A similar test using all of the model building plots for the ICH dw. Table 33 below contains the distribution of site series within the entire sample data set.

Table 33. Distribution of Site Series in ICH dw Plots – Model Building Data Map Entity Site Series Count ICHdw RFa ICH dw 01a 68ICHdw RFb ICH dw 01b 44ICHdw DO ICH dw 02 32ICHdw HD ICH dw 03 6ICHdw RD ICH dw 04 2

Total 152

Table 34. Chi Squared Test of Ecosystem Proportions in the ICH dw – Model Building Data



ICH dw 01a 68 44.74% 36207.25 41.33% 62.8 0.43 ICH dw 01b 44 28.95% 31178.38 35.59% 54.1 1.89 ICH dw 02/RO 32 21.05% 13753.86 15.70% 23.9 2.77 ICH dw 03,04 8 5.26% 6456.09 7.37% 11.2 0.92 Total 152 100.00% 100062.2 100.00% 152.0 5.97 Chi-Square = 5.97 Degrees Freedom = 3 P- Value = .1128 The P-Value of .1128 is higher than the alpha value of .1. This test should be considered a success.

4.2.1.4 ICH mw2 TEST OF ECOSYSTEM UNIT PROPORTIONS Of the plots used for this accuracy assessment, 78 fell within the ICH mw2. The distribution of those plots is described below in Table 35.



page 50

Table 35. Distribution of Site Series in ICH mw2 Plots – Accuracy Data Map Entities Site Series CountRO ICH mw2 00 3HF ICH mw2 01 29DF ICH mw2 03 5RF ICH mw2 04 30HO ICH mw2 05 7RS ICH mw2 08 2BS ICH mw2 09 2Total 78 Within the sample area there were 172,718 ha of ICH mw2 within a 1km buffer of the TRIM road system. The distribution of those site series mapped in the buffer is described below in Table 36.

Table 36. Distribution of Site Series In ICH mw2 PEM Polygons Map Entity Site Series Total AC ICHmw2 00 2104.03AR ICHmw2 00 289.71GB ICHmw2 00 195.07LA ICHmw2 00 14514.75OW ICHmw2 00 0.38PD ICHmw2 00 99.88RI ICHmw2 00 110.16RO ICHmw2 00 1814.48TA ICHmw2 00 11.37UR ICHmw2 00 849.88HF ICHmw2 01 49641.17RC ICHmw2 02 200.76DF ICHmw2 03 42890.90RF ICHmw2 04 40015.77HO ICHmw2 05 10009.16RD ICHmw2 06 6419.99RH ICHmw2 07 1010.67RS ICHmw2 08 2372.98BS ICHmw2 09 167.89

Grand Total 172718.99 The results of the comparison of the distribution of site series in the plot data to the field data are in Table 37. Site series 03 and 04 (the mesic to xeric) occupy the same locations on the edatopic grid so they were grouped for the purposes of this test. There are no plots for the 02 so it had to be grouped into a larger class. It was grouped with the Rock and Talus to create a class representing the drier site series on the edaptopic grid. The 03 and 04 were grouped because they occupy the same locations on the edatopic grid, as well there vegetation lists were very similar. The 05 remained it’s own class because there were enough plots and enough area in the data to support it. The



page 51

06, 07, 08, 09, and Avalanche Chutes were grouped in order to form a class large enough for the test. With the exception of the 00 these site series encompass the mesic to subhydric, making them a good logical grouping.

Table 37. Chi Squared Test of Ecosystem Proportions in the ICH mw2 – Accuracy Data


(Observed-Expected)2

Expected ICH mw2 01 29 37.2% 49641.17 31.69% 24.7 0.74ICH mw2 02/RO/TA 0 0.0% 2026.61 1.29% 1.0 3.93ICH mw2 03 / 04 35 44.9% 82906.67 52.92% 41.3 0.96ICH mw2 05 7 9.0% 10009.16 6.39% 5.0 0.82ICH mw2 06/07/08/09/AC 7 9.0% 12075.56 7.71% 6.0 0.67Total 78 100.00% 156659.2 100.00% 78.0 7.12 Chi-Square = 7.12 Degrees Freedom = 4 P-Value = .1299 With an alpha value of .1 and a P-value of .1299 this test should be considered a success. A similar test using all of the model building plots for the ICH dw. Table 38 below contains the distribution of site series within the entire sample data set. Table 38. Distribution of Site Series in ICH mw2 Plots – Model Building Data Map Entity Site Series Count ICHmw2 AC ICHmw2 00 3ICHmw2 RO ICHmw2 00 1ICHmw2 HF ICHmw2 01 76ICHmw2 DF ICHmw2 03 60ICHmw2 RF ICHmw2 04 43ICHmw2 HO ICHmw2 05 25ICHmw2 RD ICHmw2 06 6ICHmw2 RH ICHmw2 07 1ICHmw2 RS ICHmw2 08 6ICHmw2 BS ICHmw2 09 3Total 224



page 52

Table 39. Chi Squared Test of Ecosystem Proportions in the ICH mw2 – Model Building Data



ICH mw2 01 76 33.9% 49641.17 31.69% 71.0 0.36 ICH mw2 02/RO/TA 1 0.4% 2026.60 1.29% 2.9 1.24 ICH mw2 03/ 04 103 46.0% 82906.67 52.92% 118.5 2.04 ICH mw2 05 25 11.2% 10009.16 6.39% 14.3 7.98 ICH mw2 06/07/08/09/AC 19 8.5% 12075.56 7.71% 17.3 0.17 Total 224 100.00% 156659.17 100.00% 224.0 11.79 Chi-Square = 11.79 Degrees Freedom = 4 P-Value = .018 With an alpha value of .1 and a P-value of .018 this test should be considered a failure. Table 40 defines the overall result of the chi-square test of proportions for the 4 sub-zones tested using the accuracy assessment data set.

Table 40. Summary of Test of Ecosystem Proportions by Subzone – Accuracy Data Subzone P-Value ESSF wc1 0.6269 ESSFwc4 0.0201 ICHmw2 0.1299 ICHdw 0.058 Table 41 defines the overall result of the chi-square test of proportions for the 4 sub-zones tested using the model building data set. Table 41. Summary of Test of Ecosystem Proportions by Subzone – Model Building Data Subzone P-Value ESSF wc1 0.4414 ESSFwc4 0.8846 ICHmw2 0.0186 ICHdw 0.1128

4.2.1.5 TEST OF ECOSYSTEM UNIT PROPORTIONS FOR ALL SUBZONES In an effort to get a more comprehensive idea of how the proportions of the entire sampled PEM data set compare with the proportions of the plot data a chi-square test was run for all plots and all sampled PEM data. In order to ensure that that an expected frequency of at least 5 was present for each data class several different site series were grouped.



page 53

The 00 Class includes all non-vegetated (excluding water features) and anthropogenic plots. The logic behind this grouping is apparent, criteria for non-vegetated and anthropogenic sites are the same regardless of subzone (see Table 42).

Table 42. Anthropogenic and Non-Vegetated Plots and PEM Area Class 00 Plots PEM Area (ha)ESSF wc1 00 0 2242.2ESSF wc4 00 0 2854.9ICH mw2 00 0 5264.5ICH dw 00 3 1670.6Total 3 12032.2 The Mesic 01/04 Class (see Table 43) is composed of the zonal site series and the ESSF wc4 01. The ESSF wc4 04 occupies the same position on the edatopic grid as the ESSF wc4 01; it also occupies a similar location on the landscape making it a good match with this class. These sites all occupy the center of the edatopic grid. Their moisture regime ranges range from subhygric to submesic but for the most part the occupy the mesic.

Table 43. Site Series in the Mesic 01/04 Class Mesic 01/04 Plots PEM Area (ha)ESSF wc1 01 17 23877.4ESSF wc4 01 21 30178.9ESSF wc4 04 2 14737.15ICH dw 01b 15 31178.38ICH mw2 01 29 49641.17Total 84 149613.00 The Xeric 02 class (see Table 44) is composed of the driest site series in each subzone. Typically these site series are sub-xeric to very xeric. On the landscape they usually occupy upper to crest positions. Table 44. Site Series in the Xeric 02 Class Xeric 02 Plots PEM Area (ha)ESSF wc1 02 8 16813.9ESSF wc4 02 1 16837.1ICH dw 02 13 9758.4ICH mw2 02 0 200.8Total 22 43610.2



page 54

The subhygric 03 class (see Table 45) consists of the ICH dw 03 and ESSF wc1 03. Both of these site series occupy the subhygric portion of the of the edatopic grid. They both occupy lower slope positions and occur on variable aspects.

Table 45. Site Series in the Subhygric 03 Class Subhygric 03 Plots PEM Area(ha) ESSF wc1 03 4 5424.6ICH dw 03 1 4868.1Total 5 10292.7 The Subxeric/Submesic class (see Table 46) consists of the ICHmw2 03 and 04, ICH dw 01a, and ESSFwc4 03. The ICH mw2 03 and 04 both occupy the same locations on the edatopic grid. Their species lists are the same with the exception of soopolallie and saskatoon, which are assumed to grow only on the warm aspects. Both site series can occupy variable positions and variable slopes. The ICH dw 01a occupies a similar but more limited location on the edatopic grid ranging from submesic to subxeric. Like the ICHmw2 03 and 04 the ICH dw 01a can occur on variable slopes and in variable positions. The ESSFwc4 03 occupies a similar position on the edatopic grid to the ICH dw01a. While it does not have as much in common as the other members of this group it contribution to the class is minimal with only 1 plot and 5344 ha. Table 46. Site Series in the Submesic/Subxeric Class Submesic/Subxeric Plots PEM Area(ha) ICHmw2 03 5 42890.9ICHmw2 04 30 40015.8ICH dw 01a 31 36207.2ESSF wc4 03 1 5572.9Total 67 124686.8 The Hygric 04 / 05 Class consists of the site series listed in Table 47. These site series all occupy the mesic to hygric positions on the edatopic grid. These site series predominantly occupy the lower slopes, toes, and flat areas on cool or neutral aspects.

Table 47. Site Series in the Hygric 04 / 05 Class Hygric 04 and 05 Plots PEM Area(ha)ICHdw 04 0 301.2ESSFwc1 04 0 30.3ESSFwc1 05 0 1683.2ICHmw2 05 7 1588.0ESSFwc4 05 1 10009.1Total 8 13612.0 The site series in the subhygric-subhydric class are listed below in Table 48. All of these site series are located in the subhygric to subhydric portion of the edatopic grid but they



page 55

are predominantly found in the hygric and subhydric ranges. All of these site series are found on flats, toes, depressions, and lower slope. They all occupy neutral aspect locations with the exception of the 06’s, which occupy variable aspects.

Table 48. Site Series in the Subhygric-Subhydric Class Subhygric/Subhydric Plots PEM Area

(ha) ESSFwc4 06 0 22.8ESSFwc4 07 1 0.0ESSFwc4 08 0 0.0ICHmw2 06 0 6420.0ICHmw2 07 0 1010.7ICHmw2 08 2 2373.0ICHmw2 09 2 167.9Total 5 9994.3 Using the previously mentioned classes a Chi-Squared test was done on the entire data set. The results are shown in Table 49.

Table 49. Pooled Chi Square Test Observed Data Expected Data Site Series Classes

Plots Percentage Area Proportion Frequency (Observed-Expected)2

Expected 00 3 2% 12032.2 3% 6.4 1.82Mesic 01/04 84 43% 149613.0 41% 79.8 0.22Xeric 02 22 11% 43610.2 12% 23.3 0.07Subhygric 03 5 3% 10292.7 3% 5.5 0.04Submesic/Subxeric 67 35% 124686.8 34% 66.5 0.00Hydric 04/05 8 4% 13612.0 4% 7.3 0.08Subhygric/Subhydric 5 3% 9994.3 3% 5.3 0.02Total 194 100% 363841.1 100% 194.0 2.25 Chi-Squared = 2.25 Degrees of Freedom = 6 P-Value = .8949 With a P-Value of .8949 and an alpha value of .1 this test should be considered a success. In another interpretation of the data using the Chi-Squared test, the distribution of site series in the plot data is compared to the distribution of site series in the PEM data using both grouped and ungrouped data classes. The mesic site series are ungrouped with the exception of the ESSF wc4 01 and ESSF wc4 04. These plots are grouped because there are only two plots in the 04 and it occupies the same location as the 01 on the edatopic grid. The results of this test are shown in the Table 50.



page 56

Table 50. Partially Grouped Chi-Square Test Observed Data Expected Data Site Series

Plots % Area Proportion Frequency (Observed-Expected)2

Expected ESSF wc1 01 17 9% 23877.39 6.6% 12.7 1.43ESSF wc4 01 / 04 23 12% 44916.06 12.3% 23.9 0.04ICH dw 01b 15 8% 31178.38 8.6% 16.6 0.16ICH mw2 01 29 15% 49641.17 13.6% 26.5 0.24Class 00 3 2% 12032.17 3.3% 6.4 1.82Xeric 02 22 11% 43610.15 12.0% 23.3 0.07Subhygric 03 5 3% 10292.68 2.8% 5.5 0.04Submesic/Subxeric 67 35% 124686.77 34.3% 66.5 0.00Hygric 04/05 8 4% 13611.98 3.7% 7.3 0.08Subhygric/Subhydric 5 3% 9994.30 2.7% 5.3 0.02Total 194 100% 363841.05 100.0% 194.0 3.90 Chi Square = 3.9 P-Value = .9179 Degrees Freedom = 9 With an alpha value of 0.1 and a P-Value of .9179 this test should be considered a success.

4.2.2 PROPORTION OF DOMINANT ENTITY CORRECT This test describes whether or not the dominant entity from the PEM produced polygon corresponds with the dominant entity from the plot data. The plot data was grouped by polygon. On average there were 3 plots per polygon but in some cases there were fewer.

4.2.2.1 ESSF wc1 PROPORTION OF DOMINANT ENTITY CORRECT

The Plot and Polygon data for the ESSF wc1 is displayed below in Table 51. The shading on the table is an effect used to identify plots from the same polygon. The Dominant column is used to score polygon as to whether or not the dominant site series in the plot data is the dominant site series in the polygon. The Count column is used to record the number of polygons being tested.



page 57

Table 51. ESSF wc1 Dominant Entity Correct Calculation

PLOT DATA POLYGON DATA DOMINANT ENTITY PLOT# SITECALL POLY# SS1 D1 SS2 D2 SS3 D3 SS4 D4 SS5 D5 DOMINANT COUNT 419-1 FR 335555 FR 63% FF 25% FD 12% 0% 0% 1 1419-2 FR 338998 FR 67% FF 29% FD 4% 0% 0% 1 1419-3 FR 338998 FR 67% FF 29% FD 4% 0% 0% 423-1 FR 343377 FR 60% FF 23% FD 17% 0% 0% 1 1423-2 FR 343377 FR 60% FF 23% FD 17% 0% 0% 423-3 FR 343377 FR 60% FF 23% FD 17% 0% 0% 440-1 FD 374069 FF 62% FR 29% FD 9% 0% 0% 0 1440-2 FD 374069 FF 62% FR 29% FD 9% 0% 0% 440-3 FD 374069 FF 62% FR 29% FD 9% 0% 0% 523-1 FR 491366 FF 83% FR 16% FD 1% 0% 0% 0 1523-3 FR 491366 FF 83% FR 16% FD 1% 0% 0% 537-3 FD 519265 FR 53% FF 29% FD 16% FH 2% 0% 1 1537-1 FR 519265 FR 53% FF 29% FD 16% FH 2% 0% 537-2 FR 519265 FR 53% FF 29% FD 16% FH 2% 0% 550-1 FF 538463 FF 43% FD 22% FR 19% FH 15% PD 1% 1 1550-2 FR 539124 FR 43% FD 30% FF 27% 0% 0% 1 1550-3 FR 539124 FR 43% FD 30% FF 27% 0% 0% 664-1 FR 797680 FR 69% FF 22% FD 9% 0% 0% 1 1664-2 FR 797680 FR 69% FF 22% FD 9% 0% 0% 664-3 FR 797680 FR 69% FF 22% FD 9% 0% 0% 716-1 FF 841140 FF 78% FR 22% 0% 0% 0% 1 1716-2 FR 841832 FR 61% FF 27% FD 12% 0% 0% 1 1716-3 FR 841832 FR 61% FF 27% FD 12% 0% 0% 725-1 FF 847502 FF 67% FR 24% FD 9% 0% 0% 1 1725-2 FF 847502 FF 67% FR 24% FD 9% 0% 0% 725-3 FF 847502 FF 67% FR 24% FD 9% 0% 0% 843-1 FF 962950 FR 91% FF 9% 0% 0% 0% 1 1843-2 FF 962950 FR 91% FF 9% 0% 0% 0% 843-3 FF 962950 FR 91% FF 9% 0% 0% 0%

TOTAL 11 13 The results of the Dominant Entity Correct Test for the ESSF wc1 are displayed below in Table 52.

Table 52. ESSF wc1 Dominant Entity Correct Total Polygons 13Score 11Percent Correct 85%Lower +/- 95% Confidence Value 8Mid +/- 95% Confidence Value 11Upper +/- 95% Confidence Value 13Lower +/- 95% Confidence Interval 61.5%Mid +/- 95% Confidence Interval 84.6%Upper +/- 95% Confidence Interval 100.0%



page 58

4.2.2.2 ESSF wc4 PROPORTION OF DOMINANT ENTITY CORRECT The Plot and Polygon data for the ESSF wc4 is displayed below in Table 53. The shading on the table is an effect used to identify plots from the same polygon. The Dominant column is used to score polygon as to whether or not the dominant site series in the plot data is the dominant site series in the polygon. The Count column is used to record the number of polygons being tested. Table 53. ESSF wc4 Dominant Entity Correct Calculation

PLOT DATA POLYGON DATA DOMINANT ENTITYPLOT# SITECALL POLY# SS1 D1 SS2 D2 SS3 D3 SS4 D4 SS5 D5 DOMINANT COUNT504-3 FR 462832 FR 88% RF 11% FF 1% 0% 1 1153-1 FR 379378 RF 44% FR 28% FF 24% FW 4% 0 1153-2 FR 379378 RF 44% FR 28% FF 24% FW 4% 153-3 FR 379378 RF 44% FR 28% FF 24% FW 4% 504-1 FR 461795 FR 87% RF 12% FF 1% 0% 1 1504-2 FR 461795 FR 87% RF 12% FF 1% 0% 526-3 FR 495067 FR 96% RF 3% FF 1% 0% 1 1533-3 FR 507021 FR 100% 0% 0% 0% 1 1526-1 FR 497989 FR 93% RF 7% 0% 0% 1 1526-2 FR 497989 FR 93% RF 7% 0% 0% 528-1 FR 500427 FR 89% RF 11% 0% 0% 1 1528-2 FR 500427 FR 89% RF 11% 0% 0% 533-1 FR 502465 FR 88% RF 12% 0% 0% 1 1533-2 FR 502465 FR 88% RF 12% 0% 0% 573-3 FW 588303 FR 100% 0% 0% 0% 0 1573-1 RF 588303 FR 100% 0% 0% 0% 573-2 RF 588303 FR 100% 0% 0% 0% 630-1 FF 762778 FR 88% RF 12% 0% 0% 0 1630-3 FR 762778 FR 88% RF 12% 0% 0% 630-2 SS 762778 FR 88% RF 12% 0% 0% 633-2 FR 766928 FR 88% RF 12% 0% 0% 1 1633-3 FR 766928 FR 88% RF 12% 0% 0% 901-1 FR 766928 FR 88% RF 12% 0% 0% 1 1901-2 FR 766928 FR 88% RF 12% 0% 0% 901-3 FR 766928 FR 88% RF 12% 0% 0% 528-3 FR 502144 FR 88% RF 12% 0% 0% 1 1633-1 FL 765706 FR 89% RF 11% 0% 0% 0 1TOTAL 10 14 The results of the ESSF wc4 Dominant Entity Correct Test are shown below in Table 54.



page 59

Table 54. ESSF wc4 Dominant Entity Correct Total Polygons 14Score 10Percent Correct 71%Lower +/- 95% Confidence Value 7Mid +/- 95% Confidence Value 10Upper +/- 95% Confidence Value 13Lower +/- 95% Confidence Interval 54.8%Mid +/- 95% Confidence Interval 71%Upper +/- 95% Confidence Interval 87.1%

4.2.2.3 ICH dw PROPORTION OF DOMINANT ENTITY CORRECT The Plot and Polygon data for the ICH dw is displayed below in Table 55. The shading on the table is an effect used to identify plots from the same polygon. The Dominant column is used to score polygon as to whether or not the dominant site series in the plot data is the dominant site series in the polygon. The Count column is used to record the number of polygons being tested. Most of the polygons scores are obvious cases of there being a dominant site series in the polygon data and a single dominant site series in the plot data. There are however a few polygons where the scoring requires some explanation. Polygon 21913 has 51% DO, 34% RFa, 9% RO, and 6% RFb. There are 2 plots in this polygon; plots 602-1 and 602-2. These plots recorded 100% DO and 100% RFa. The DO is the dominant site series in the polygon but dominance is shared between the 2 site series called in the plot data. As opposed to saying that the data is tied and allocating or not allocating points on that basis the second or third site site series have been used to break the tie. If there was no RFa in the PEM polygon then no points would be allocated because of the tie in the plot data. This is the case with polygon 75895. The PEM has called the Polygon 100% RFa. There are 2 plots in the polygon, one being RFa, and the other being DO. Because they are tied for dominance and the PEM does not support the presence of DO anywhere in the polygon no points are allotted for this polygon.



page 60

Table 55. ICH dw Dominant Entity Correct Calculation. Polygon Data Plot Data Dominant Entity

POLY-ID S1 DEC1 S2 DEC2 S3 DEC3 S4 DEC4 S5 DEC5 PLOT S1 DEC1 S2 DEC2 DOMINANT COUNT5110 DO 0.61 RFa 0.39 0 0 0 545-1 DO 100 0 1 1 5110 DO 0.61 RFa 0.39 0 0 0 545-2 DO 100 0 5110 DO 0.61 RFa 0.39 0 0 0 545-3 DO 100 0

10266 RFb 0.58 RFa 0.21 HD 0.15 RD 0.05 DO 0.01 601-1 RFa 100 0 0 1 10266 RFb 0.58 RFa 0.21 HD 0.15 RD 0.05 DO 0.01 601-2 RFa 100 0 10266 RFb 0.58 RFa 0.21 HD 0.15 RD 0.05 DO 0.01 601-3 RFa 100 0 15158 RFa 0.7 DO 0.3 0 0 0 JD1-1 RFa 100 0 1 1 15158 RFa 0.7 DO 0.3 0 0 0 JD1-2 RFa 100 0 15158 RFa 0.7 DO 0.3 0 0 0 JD1-3 RFa 100 0 21913 DO 0.51 RFa 0.34 RO 0.09 RFb 0.06 0 602-1 DO 100 0 1 1 21913 DO 0.51 RFa 0.34 RO 0.09 RFb 0.06 0 602-2 RFa 100 0 21971 RFa 0.76 DO 0.24 0 0 0 602-3 RFa 100 0 1 1 21990 RFa 0.51 RFb 0.3 RO 0.17 DO 0.02 HD 0 603-2 RFb 100 0 0 1 21990 RFa 0.51 RFb 0.3 RO 0.17 DO 0.02 HD 0 603-3 RFb 100 0 22531 RFb 0.45 HD 0.37 RFa 0.14 DO 0.03 RD 0.01 603-1 RFb 100 0 1 1 59380 RFb 0.38 HD 0.29 RFa 0.22 RD 0.1 DO 0.01 660-1 RFb 100 0 1 1 59380 RFb 0.38 HD 0.29 RFa 0.22 RD 0.1 DO 0.01 660-2 RFb 100 0 59380 RFb 0.38 HD 0.29 RFa 0.22 RD 0.1 DO 0.01 660-3 RFb 100 0 60249 RFb 0.7 HD 0.2 RD 0.1 0 0 VR1-2 RFa 100 0 0 1 60293 RFa 0.6 RFb 0.4 0 0 0 VR1-3 RFa 100 0 1 1 60413 RFa 0.6 RFb 0.4 0 0 0 VR1-1 RFb 100 0 0 1 61949 RFa 0.87 RFb 0.07 DO 0.05 HD 0.01 0 666-3 RFb 70 HD 30 0 1 61949 RFa 0.87 RFb 0.07 DO 0.05 HD 0.01 0 666-1 RFb 100 0 61949 RFa 0.87 RFb 0.07 DO 0.05 HD 0.01 0 666-2 RFb 100 0 63625 RFa 0.9 RFb 0.08 DO 0.02 0 0 670-3 RFa 100 0 1 1 64318 RFa 0.73 RFb 0.18 DO 0.05 HD 0.02 UR 0.02 673-1 RFa 100 0 1 1 64318 RFa 0.73 RFb 0.18 DO 0.05 HD 0.02 UR 0.02 673-2 RFa 100 0 64318 RFa 0.73 RFb 0.18 DO 0.05 HD 0.02 UR 0.02 673-3 RFb 100 0 64654 RFa 1 0 0 0 0 670-1 DO 100 0 0 1 65261 DO 1 0 0 0 0 670-2 RFa 100 0 0 1 68943 RFa 0.84 DO 0.12 RFb 0.04 HD 0 0 694-1 RFa 100 0 1 1 68943 RFa 0.84 DO 0.12 RFb 0.04 HD 0 0 694-2 RFa 100 0 68943 RFa 0.84 DO 0.12 RFb 0.04 HD 0 0 694-3 RFa 100 0 70806 RFa 0.75 RFb 0.15 RO 0.1 0 0 705-3 HD 100 0 0 1 70939 RFa 0.53 RFb 0.47 RO 0 0 0 705-1 DO 100 0 0 1 70939 RFa 0.53 RFb 0.47 RO 0 0 0 705-2 RFa 100 0 75895 RFa 1 0 0 0 0 731-1 RFa 100 0 0 1 75895 RFa 1 0 0 0 0 731-3 DO 100 0 76365 DO 1 0 0 0 0 731-2 DO 100 0 1 1 77287 RFa 1 0 0 0 0 765-2 RFa 100 0 1 1 77287 RFa 1 0 0 0 0 765-3 RFa 100 0 79594 RFb 0.66 RFa 0.22 HD 0.09 RD 0.02 UR 0.01 772-2 RFb 100 0 1 1 79823 DO 1 0 0 0 0 765-1 DO 100 0 1 1 80063 RFb 0.67 RFa 0.12 UR 0.12 HD 0.06 RD 0.03 772-3 RFb 60 DO 40 1 1 80063 RFb 0.67 RFa 0.12 UR 0.12 HD 0.06 RD 0.03 772-1 RFb 100 0 86847 DO 1 0 0 0 0 799-1 DO 100 0 1 1 86847 DO 1 0 0 0 0 799-2 DO 100 0 86847 DO 1 0 0 0 0 799-3 DO 100 0 98110 RFa 0.48 RFb 0.25 DO 0.24 HD 0.03 0 1260-1 RFa 100 0 1 1 98110 RFa 0.48 RFb 0.25 DO 0.24 HD 0.03 0 1260-2 RFa 100 0 98110 RFa 0.48 RFb 0.25 DO 0.24 HD 0.03 0 1260-3 RFa 100 0

129666 RFa 0.62 DO 0.27 RFb 0.11 0 0 1353-1 RFa 100 0 1 1 129666 RFa 0.62 DO 0.27 RFb 0.11 0 0 1353-2 RFa 100 0 129666 RFa 0.62 DO 0.27 RFb 0.11 0 0 1353-3 RFa 100 0 130795 RFa 0.53 RFb 0.38 HD 0.05 DO 0.03 RD 0.01 886-1 RFb 100 0 1 1 130795 RFa 0.53 RFb 0.38 HD 0.05 DO 0.03 RD 0.01 886-2 RFa 100 0 130795 RFa 0.53 RFb 0.38 HD 0.05 DO 0.03 RD 0.01 886-3 RFa 100 0 132361 RFa 0.44 RFb 0.41 DO 0.06 UR 0.06 HD 0.03 894-3 RFa 100 0 1 1 133147 RFa 0.78 RFb 0.12 DO 0.1 HD 0 0 894-1 RFa 100 0 1 1 133147 RFa 0.78 RFb 0.12 DO 0.1 HD 0 0 894-2 DO 100 0

21 31



page 61

The results of the ICH dw Dominant Entity Correct test are described below in Table 56.

Table 56. ICH dw Dominant Entity Correct Total Polygons 31Score 21Percent Correct 68%Lower +/- 95% Confidence Value 16Mid +/- 95% Confidence Value 21Upper +/- 95% Confidence Value 26Lower +/- 95% Confidence Interval

52%

Mid +/- 95% Confidence Interval 68%Upper +/- 95% Confidence Interval

84%

4.2.2.4 ICH mw2 PROPORTION OF DOMINANT ENTITY CORRECT

The Plot and Polygon data for the ICH mw2 is displayed below in Table 57. The shading on the table is an effect used to identify plots from the same polygon. The Dominant column is used to score polygon as to whether or not the dominant site series in the plot data is the dominant site series in the polygon. The count column is used to record the number of polygons being tested. The results of this test are presented in Table 57.



page 62

Table 57. ICH mw2 Dominant Entity Correct Calculation. Polygon Data Plot

Data Dominant

POLY-ID S1 DEC1 S2 DEC2 S3 DEC3 S4 DEC4 S5 DEC5 S6 DEC6 S7 DEC7 PLOT SS1 DEC1 SS2 DEC2 Dom COUNT 125799 RF 0.44 HF 0.3 DF 0.13 HO 0.09 RD 0.04 0 0 417-1 RF 100 0 1 1126104 RF 1 0 0 0 0 0 0 417-2 RF 100 0 1 1126584 DF 0.44 HF 0.41 HO 0.11 RF 0.04 0 0 0 417-3 RF 100 0 0 1148403 HF 0.76 RF 0.24 0 0 0 0 0 424-3 RF 100 0 0 1149581 RF 0.56 HF 0.38 DF 0.06 0 0 0 0 424-1 RF 100 0 1 1149581 RF 0.56 HF 0.38 DF 0.06 0 0 0 0 424-2 RF 100 0173679 HF 0.29 RF 0.19 RD 0.18 DF 0.13 HO 0.12 RI 0.09 RS 0 456-1 HF 100 0 1 1178497 RF 0.47 HF 0.3 DF 0.21 HO 0.01 RD 0.01 0 0 456-2 HF 100 0 0 1178497 RF 0.47 HF 0.3 DF 0.21 HO 0.01 RD 0.01 0 0 456-3 HF 100 0191022 AC 1 0 0 0 0 0 0 475-1 RO 100 0 0 1191022 AC 1 0 0 0 0 0 0 475-2 RO 100 0191022 AC 1 0 0 0 0 0 0 475-3 RO 100 0230227 HF 0.54 RF 0.3 DF 0.09 HO 0.04 RD 0.01 RS 0.02 0 512-1 HF 100 0 1 1230227 HF 0.54 RF 0.3 DF 0.09 HO 0.04 RD 0.01 RS 0.02 0 512-2 HF 100 0230227 HF 0.54 RF 0.3 DF 0.09 HO 0.04 RD 0.01 RS 0.02 0 512-3 HF 100 0238249 HF 0.45 DF 0.17 HO 0.15 RD 0.1 RF 0.09 RS 0.02 RH 0.02 520-1 HF 100 0 1 1238249 HF 0.45 DF 0.17 HO 0.15 RD 0.1 RF 0.09 RS 0.02 RH 0.02 520-2 HF 100 0238249 HF 0.45 DF 0.17 HO 0.15 RD 0.1 RF 0.09 RS 0.02 RH 0.02 520-3 HF 100 0268593 DF 0.4 HF 0.32 RF 0.17 HO 0.1 RD 0.01 RS 0 0 556-1 HF 100 0 0 1268593 DF 0.4 HF 0.32 RF 0.17 HO 0.1 RD 0.01 RS 0 0 556-2 HO 100 0268593 DF 0.4 HF 0.32 RF 0.17 HO 0.1 RD 0.01 RS 0 0 556-3 HO 100 0276262 HF 0.42 DF 0.19 RF 0.16 HO 0.13 RD 0.06 RH 0.02 RS 0.02 562-1 HO 100 0 0 1276262 HF 0.42 DF 0.19 RF 0.16 HO 0.13 RD 0.06 RH 0.02 RS 0.02 562-2 HF 100 0276262 HF 0.42 DF 0.19 RF 0.16 HO 0.13 RD 0.06 RH 0.02 RS 0.02 562-3 RF 100 0278042 HF 0.45 DF 0.18 HO 0.14 RF 0.08 RD 0.06 RS 0.04 RH 0.05 558-1 HF 100 0 1 1278042 HF 0.45 DF 0.18 HO 0.14 RF 0.08 RD 0.06 RS 0.04 RH 0.05 558-2 HF 100 0278042 HF 0.45 DF 0.18 HO 0.14 RF 0.08 RD 0.06 RS 0.04 RH 0.05 558-3 HF 100 0298220 RF 0.4 HF 0.38 DF 0.08 HO 0.07 RD 0.07 RS 0 0 589-1 HF 100 0 0 1298220 RF 0.4 HF 0.38 DF 0.08 HO 0.07 RD 0.07 RS 0 0 589-2 HF 100 0298220 RF 0.4 HF 0.38 DF 0.08 HO 0.07 RD 0.07 RS 0 0 589-3 HF 100 0301216 RD 0.35 RS 0.23 HO 0.18 HF 0.11 RH 0.09 RF 0.02 DF 0.02 593-1 RS 100 0 0 1301216 RD 0.35 RS 0.23 HO 0.18 HF 0.11 RH 0.09 RF 0.02 DF 0.02 593-2 HO 100 0301216 RD 0.35 RS 0.23 HO 0.18 HF 0.11 RH 0.09 RF 0.02 DF 0.02 593-3 HO 100 0305641 RF 0.8 HF 0.11 DF 0.09 HO 0 RD 0 0 0 JD2-1 RF 100 0 1 1305641 RF 0.8 HF 0.11 DF 0.09 HO 0 RD 0 0 0 JD2-2 RF 100 0305641 RF 0.8 HF 0.11 DF 0.09 HO 0 RD 0 0 0 JD2-3 RF 100 0306692 RF 0.65 HF 0.24 DF 0.11 HO 0 RD 0 0 0 CD1-1 RF 100 0 1 1306692 RF 0.65 HF 0.24 DF 0.11 HO 0 RD 0 0 0 CD1-2 RF 100 0306692 RF 0.65 HF 0.24 DF 0.11 HO 0 RD 0 0 0 CD1-3 RF 100 0307910 RF 0.56 HF 0.28 RD 0.08 HO 0.04 DF 0.04 RS 0 0 CD2-1 RF 100 0 1 1307910 RF 0.56 HF 0.28 RD 0.08 HO 0.04 DF 0.04 RS 0 0 CD2-2 RF 100 0307910 RF 0.56 HF 0.28 RD 0.08 HO 0.04 DF 0.04 RS 0 0 CD2-3 RF 100 0308081 RF 0.75 HF 0.15 DF 0.1 HO 0 RD 0 0 0 595-1 RF 100 0 1 1308081 RF 0.75 HF 0.15 DF 0.1 HO 0 RD 0 0 0 595-2 RF 100 0308081 RF 0.75 HF 0.15 DF 0.1 HO 0 RD 0 0 0 595-3 RF 100 0367651 HF 0.51 DF 0.22 HO 0.1 RF 0.09 RD 0.05 RS 0.02 RH 0.01 661-3 RS 80 HF 20 1 1367651 HF 0.51 DF 0.22 HO 0.1 RF 0.09 RD 0.05 RS 0.02 RH 0.01 661-1 HF 100 0367651 HF 0.51 DF 0.22 HO 0.1 RF 0.09 RD 0.05 RS 0.02 RH 0.01 661-2 HF 100 0390010 HF 0.5 DF 0.23 RF 0.11 HO 0.1 RD 0.04 RS 0.01 RH 0 702-1 HF 100 0 0 1390010 HF 0.5 DF 0.23 RF 0.11 HO 0.1 RD 0.04 RS 0.01 RH 0 702-2 DF 100 0390010 HF 0.5 DF 0.23 RF 0.11 HO 0.1 RD 0.04 RS 0.01 RH 0 702-3 DF 100 0398085 HF 0.48 DF 0.17 HO 0.14 RF 0.11 RD 0.07 RS 0.02 RH 0.01 733-1 HF 100 0 1 1398085 HF 0.48 DF 0.17 HO 0.14 RF 0.11 RD 0.07 RS 0.02 RH 0.01 733-2 HF 100 0398923 HF 0.33 RH 0.2 DF 0.14 RS 0.12 HO 0.12 RF 0.05 RD 0.04 733-3 HF 100 0 1 1399809 RS 0.28 HF 0.26 HO 0.19 DF 0.1 RD 0.08 RF 0.05 RH 0.04 739-1 BS 100 0 0 1399809 RS 0.28 HF 0.26 HO 0.19 DF 0.1 RD 0.08 RF 0.05 RH 0.04 739-2 BS 100 0399809 RS 0.28 HF 0.26 HO 0.19 DF 0.1 RD 0.08 RF 0.05 RH 0.04 739-3 HO 100 0



page 63

Polygon Data

Plot Data

Dominant

POLY-ID S1 DEC1 S2 DEC2 S3 DEC3 S4 DEC4 S5 DEC5 S6 DEC6 S7 DEC7 PLOT SS1 DEC1 SS2 DEC2 Dom COUNT 400660 HF 0.55 RF 0.3 HO 0.07 DF 0.05 RD 0.03 RS 0 0 736-1 RF 100 0 0 1400660 HF 0.55 RF 0.3 HO 0.07 DF 0.05 RD 0.03 RS 0 0 736-2 RF 100 0400660 HF 0.55 RF 0.3 HO 0.07 DF 0.05 RD 0.03 RS 0 0 736-3 RF 100 0402332 RD 0.66 HO 0.19 HF 0.07 RF 0.05 DF 0.03 RS 0 0 740-1 HF 100 0 0 1402332 RD 0.66 HO 0.19 HF 0.07 RF 0.05 DF 0.03 RS 0 0 740-3 HF 100 0403149 HF 0.37 RF 0.29 RD 0.16 HO 0.11 DF 0.07 RS 0 0 740-2 HF 100 0 1 1403149 HF 0.37 RF 0.29 RD 0.16 HO 0.11 DF 0.07 RS 0 0 742-1 HF 100 0403149 HF 0.37 RF 0.29 RD 0.16 HO 0.11 DF 0.07 RS 0 0 742-2 RF 100 0403149 HF 0.37 RF 0.29 RD 0.16 HO 0.11 DF 0.07 RS 0 0 742-3 HF 100 0406204 DF 1 0 0 0 0 0 0 745-1 DF 100 0 1 1406204 DF 1 0 0 0 0 0 0 745-2 DF 100 0406204 DF 1 0 0 0 0 0 0 745-3 DF 100 0422339 HF 0.45 RF 0.25 DF 0.18 HO 0.09 RD 0.03 RS 0 0 790-1 RF 100 0 0 1422339 HF 0.45 RF 0.25 DF 0.18 HO 0.09 RD 0.03 RS 0 0 790-2 HO 100 0422339 HF 0.45 RF 0.25 DF 0.18 HO 0.09 RD 0.03 RS 0 0 790-3 HF 100 0440797 RF 0.54 HF 0.39 DF 0.06 HO 0 RD 0.01 0 0 833-1 RF 100 0 1 1440797 RF 0.54 HF 0.39 DF 0.06 HO 0 RD 0.01 0 0 833-2 RF 100 0440867 HF 0.49 RF 0.43 DF 0.04 HO 0.02 RD 0.02 RS 0 0 833-3 RF 100 0 0 1459024 RF 0.53 HF 0.36 RD 0.04 HO 0.04 DF 0.03 0 0 867-1 RF 100 0 1 1459024 RF 0.53 HF 0.36 RD 0.04 HO 0.04 DF 0.03 0 0 867-2 RF 100 0459024 RF 0.53 HF 0.36 RD 0.04 HO 0.04 DF 0.03 0 0 867-3 RF 100 0

18 32

Table 58. ICH mw2 Dominant Entity Correct Total Polygons 32Score 18Percent Correct 56%Lower +/- 95% Confidence Value 12Mid +/- 95% Confidence Value 18Upper +/- 95% Confidence Value 23Lower +/- 95% Confidence Interval

37.5%

Mid +/- 95% Confidence Interval 56.3%Upper +/- 95% Confidence Interval

71.9%

4.2.2.5 OVERALL PROPORTION OF DOMINANT ENTITY CORRECT

The overall proportion of Dominant Entity Correct describes the abilities of the PEM to predict the correct dominant entity in terms of all 4 subzones. In Table 59, mid +/- 95% confidence interval was used in a weighted average calculation. The rational behind using a weighted average calculation is that the 4 tested subzones all had different amounts of area available for sampling.



page 64

Table 59. Summary of Dominant Entity Correct Tests Weighted Average Subzone Number of Polygons Mid +/- 95%

Confidence Interval ESSF wc1 13 85% ESSF wc4 14 71% ICH dw 31 68% ICH mw2 32 56% Total 90 66.7% The weighted average for the mid +/- 95% confidence interval of the dominant entity correct test is 66.7%. In Table 60 the results are presented with no weighting applied for the number of polygons from each subzone.

Table 60. Summary of Dominant Entity Correct Tests Unweighted Average Subzone Number of Polygons Mid +/- 95%

Confidence Interval ESSF wc1 13 85% ESSF wc4 14 71% ICH dw 31 68% ICH mw2 32 56% Total 90 70%

4.2.3 THE PERCENT OVERLAP TEST The percent overlap test compares the map entity proportions in the plot data to the map entity proportions the PEM polygon data.

4.2.3.1 ESSF wc1 PERCENT OVERLAP TEST The calculations for determining the percent overlap for this subzone are found below in Table 61. When taking this result into account the fact that the best possible score that this data set could achieve regardless of whether or not the primary polygon call matches the plot call and had the same distribution is 8.4/13 or 64.6%. The percent overlap for the ESSF wc1 is 6.26/13 or 48%.



page 65

Table 61. ESSF wc1 Percent Overlap Calculations PLOT# SITECALL POLY# SS1 D1 SS2 D2 SS3 D3 SS4 D4 SS5 D5 Overlap COUNT419-1 FR 335555 FR 63% FF 25% FD 12% 0% 0% 63% 1419-2 FR 338998 FR 67% FF 29% FD 4% 0% 0% 67% 1419-3 FR 338998 FR 67% FF 29% FD 4% 0% 0% 423-1 FR 343377 FR 60% FF 23% FD 17% 0% 0% 60% 1423-2 FR 343377 FR 60% FF 23% FD 17% 0% 0% 423-3 FR 343377 FR 60% FF 23% FD 17% 0% 0% 440-1 FD 374069 FF 62% FR 29% FD 9% 0% 0% 9% 1440-2 FD 374069 FF 62% FR 29% FD 9% 0% 0% 440-3 FD 374069 FF 62% FR 29% FD 9% 0% 0% 523-1 FR 491366 FF 83% FR 16% FD 1% 0% 0% 16% 1523-3 FR 491366 FF 83% FR 16% FD 1% 0% 0% 537-3 FD 519265 FR 53% FF 29% FD 16% FH 2% 0% 41% 1537-1 FR 519265 FR 53% FF 29% FD 16% FH 2% 0% 537-2 FR 519265 FR 53% FF 29% FD 16% FH 2% 0% 550-1 FF 538463 FF 43% FD 22% FR 19% FH 15% PD 1% 43% 1550-2 FR 539124 FR 43% FD 30% FF 27% 0% 0% 43% 1550-3 FR 539124 FR 43% FD 30% FF 27% 0% 0% 664-1 FR 797680 FR 69% FF 22% FD 9% 0% 0% 69% 1664-2 FR 797680 FR 69% FF 22% FD 9% 0% 0% 664-3 FR 797680 FR 69% FF 22% FD 9% 0% 0% 716-1 FF 841140 FF 78% FR 22% 0% 0% 0% 78% 1716-2 FR 841832 FR 61% FF 27% FD 12% 0% 0% 61% 1716-3 FR 841832 FR 61% FF 27% FD 12% 0% 0% 725-1 FF 847502 FF 67% FR 24% FD 9% 0% 0% 67% 1725-2 FF 847502 FF 67% FR 24% FD 9% 0% 0% 725-3 FF 847502 FF 67% FR 24% FD 9% 0% 0% 843-1 FF 962950 FR 91% FF 9% 0% 0% 0% 9% 1843-2 FF 962950 FR 91% FF 9% 0% 0% 0% 843-3 FF 962950 FR 91% FF 9% 0% 0% 0%

TOTAL 6.26 13

4.2.3.2 ESSF wc4 PERCENT OVERLAP TEST The calculations for determining the percent overlap for this subzone are found below in Table 62. When taking this result into account the fact that the best possible score that this data set could achieve regardless of whether or not the primary polygon call matches the plot call and had the same distribution is 12.26/14 or 87.5%. The percent overlap for the ESSF wc4 is 9.62/14 or 68.7%



page 66

Table 62. ESSF wc4 Percent Overlap Calculations PLOT DATA POLYGON DATA DOMINANT ENTITY

PLOT# SITECALL POLY# SS1 D1 SS2 D2 SS3 D3 SS4 D4 SS5 D5 DOMINANT COUNT

528-3 FR 502144 FR 88% RF 12% 0% 0% 88% 1504-3 FR 462832 FR 88% RF 11% FF 1% 0% 88% 1153-1 FR 379378 RF 44% FR 28% FF 24% FW 4% 28% 1153-2 FR 379378 RF 44% FR 28% FF 24% FW 4% 153-3 FR 379378 RF 44% FR 28% FF 24% FW 4% 504-1 FR 461795 FR 87% RF 12% FF 1% 0% 87% 1504-2 FR 461795 FR 87% RF 12% FF 1% 0% 526-1 FR 497989 FR 93% RF 7% 0% 0% 93% 1526-2 FR 497989 FR 93% RF 7% 0% 0% 528-1 FR 500427 FR 89% RF 11% 0% 0% 89% 1528-2 FR 500427 FR 89% RF 11% 0% 0% 533-1 FR 502465 FR 88% RF 12% 0% 0% 88% 1533-2 FR 502465 FR 88% RF 12% 0% 0% 573-3 FW 588303 FR 100% 0% 0% 0% 0% 1573-1 RF 588303 FR 100% 0% 0% 0% 573-2 RF 588303 FR 100% 0% 0% 0% 630-1 FF 762778 FR 88% RF 12% 0% 0% 29% 1630-3 FR 762778 FR 88% RF 12% 0% 0% 630-2 SS 762778 FR 88% RF 12% 0% 0% 633-2 FR 766928 FR 88% RF 12% 0% 0% 88% 1633-3 FR 766928 FR 88% RF 12% 0% 0% 901-1 FR 766928 FR 88% RF 12% 0% 0% 88% 1901-2 FR 766928 FR 88% RF 12% 0% 0% 901-3 FR 766928 FR 88% RF 12% 0% 0% 526-3 FR 495067 FR 96% RF 3% FF 1% 0% 96% 1533-3 FR 507021 FR 100% 0% 0% 0% 100% 1633-1 FL 765706 FR 89% RF 11% 0% 0% 0.00 1TOTAL 9.62 14

4.2.3.3 ICH dw PERCENT OVERLAP TEST The calculations for determining the percent overlap for this subzone are found below in Table 63. When taking this result into account the fact that the best possible score that this data set could achieve regardless of whether or not the primary polygon call matches the plot call and had the same distribution is 16.09/31 or 51.9%. The reason this test cannot succeed is that the first decile in many of the polygons sampled is relatively low. The percent overlap for the ICH dw is (15.74/31) 50.7%.



page 67

Table 63. ICH dw Percent Overlap Calculations

Polygon Data Plot Data Overlap Poly-ID S1 Dec 1 S2 Dec 2 S3 Dec 3 S4 Dec 4 S5 Dec 5 S6 Dec 6 PLOT SS1 1-Dec Overlap COUNT

5110 DO 0.61 Rfa 0.39 0 0 0 0 545-1 DO 100 0.61 15110 DO 0.61 Rfa 0.39 0 0 0 0 545-2 DO 1005110 DO 0.61 Rfa 0.39 0 0 0 0 545-3 DO 100

10266 RFb 0.58 Rfa 0.21 HD 0.15 RD 0.05 DO 0.01 GB 0 601-1 RFa 100 0.21 110266 RFb 0.58 Rfa 0.21 HD 0.15 RD 0.05 DO 0.01 GB 0 601-2 RFa 10010266 RFb 0.58 Rfa 0.21 HD 0.15 RD 0.05 DO 0.01 GB 0 601-3 RFa 10015158 RFa 0.7 DO 0.3 0 0 0 0 JD1-1 RFa 100 0.7 115158 RFa 0.7 DO 0.3 0 0 0 0 JD1-2 RFa 10015158 RFa 0.7 DO 0.3 0 0 0 0 JD1-3 RFa 10021913 DO 0.51 Rfa 0.34 RO 0.09 RFb 0.06 0 0 602-1 DO 100 0.42 121913 DO 0.51 Rfa 0.34 RO 0.09 RFb 0.06 0 0 602-2 RFa 10021971 RFa 0.76 DO 0.24 0 0 0 0 602-3 RFa 100 0.76 121990 RFa 0.51 RFb 0.3 RO 0.17 DO 0.02 HD 0 0 603-2 RFb 100 0.3 121990 RFa 0.51 RFb 0.3 RO 0.17 DO 0.02 HD 0 0 603-3 RFb 10022531 RFb 0.45 HD 0.37 RFa 0.14 DO 0.03 RD 0.01 0 603-1 RFb 100 0.45 159380 RFb 0.38 HD 0.29 RFa 0.22 RD 0.1 DO 0.01 0 660-1 RFb 100 0.38 159380 RFb 0.38 HD 0.29 RFa 0.22 RD 0.1 DO 0.01 0 660-2 RFb 10059380 RFb 0.38 HD 0.29 RFa 0.22 RD 0.1 DO 0.01 0 660-3 RFb 10060249 RFb 0.7 HD 0.2 RD 0.1 0 0 0 VR1-2 RFa 100 0 160293 RFa 0.6 RFb 0.4 0 0 0 0 VR1-3 RFa 100 0.6 160413 RFa 0.6 RFb 0.4 0 0 0 0 VR1-1 RFb 100 0.4 161949 RFa 0.87 RFb 0.07 DO 0.05 HD 0.01 0 0 666-3 RFb 70 0.08 161949 RFa 0.87 RFb 0.07 DO 0.05 HD 0.01 0 0 666-1 RFb 10061949 RFa 0.87 RFb 0.07 DO 0.05 HD 0.01 0 0 666-2 RFb 10063625 RFa 0.9 RFb 0.08 DO 0.02 0 0 0 670-3 RFa 100 0.9 164318 RFa 0.73 RFb 0.18 DO 0.05 HD 0.02 UR 0.02 0 673-1 RFa 100 0.54 164318 RFa 0.73 RFb 0.18 DO 0.05 HD 0.02 UR 0.02 0 673-2 RFa 10064318 RFa 0.73 RFb 0.18 DO 0.05 HD 0.02 UR 0.02 0 673-3 RFb 10064654 RFa 1 0 0 0 0 0 670-1 DO 100 0 165261 DO 1 0 0 0 0 0 670-2 RFa 100 0 168943 RFa 0.84 DO 0.12 RFb 0.04 HD 0 0 0 694-1 RFa 100 0.84 168943 RFa 0.84 DO 0.12 RFb 0.04 HD 0 0 0 694-2 RFa 10068943 RFa 0.84 DO 0.12 RFb 0.04 HD 0 0 0 694-3 RFa 10070806 RFa 0.75 RFb 0.15 RO 0.1 0 0 0 705-3 HD 100 0 170809 RFa 0.53 RFb 0.47 RO 0 0 0 0 705-1 DO 100 0.26 170939 RFa 0.53 RFb 0.47 RO 0 0 0 0 705-2 RFa 10075895 RFa 1 0 0 0 0 0 731-1 RFa 100 0.5 175895 RFa 1 0 0 0 0 0 731-3 DO 10076365 DO 1 0 0 0 0 0 731-2 DO 100 1 177287 RFa 1 0 0 0 0 0 765-2 RFa 100 1 177287 RFa 1 0 0 0 0 0 765-3 RFa 10079594 RFb 0.66 Rfa 0.22 HD 0.09 RD 0.02 UR 0.01 DO 0 772-2 RFb 100 0.66 179823 DO 1 0 0 0 0 0 765-1 DO 100 1 180063 RFb 0.67 Rfa 0.12 UR 0.12 HD 0.06 RD 0.03 0 772-3 RFb 60 0.67 180063 RFb 0.67 Rfa 0.12 UR 0.12 HD 0.06 RD 0.03 0 772-1 RFb 10086847 DO 1 0 0 0 0 0 799-1 DO 100 1 186847 DO 1 0 0 0 0 0 799-2 DO 10086847 DO 1 0 0 0 0 0 799-3 DO 10098110 RFa 0.48 RFb 0.25 DO 0.24 HD 0.03 0 0 1260-1 RFa 100 0.48 198110 RFa 0.48 RFb 0.25 DO 0.24 HD 0.03 0 0 1260-2 RFa 10098110 RFa 0.48 RFb 0.25 DO 0.24 HD 0.03 0 0 1260-3 RFa 100

129666 RFa 0.62 DO 0.27 RFb 0.11 0 0 0 1353-1 RFa 100 0.62 1129666 RFa 0.62 DO 0.27 RFb 0.11 0 0 0 1353-2 RFa 100129666 RFa 0.62 DO 0.27 RFb 0.11 0 0 0 1353-3 RFa 100



page 68

130795 RFa 0.53 RFb 0.38 HD 0.05 DO 0.03 RD 0.01 0 886-1 RFb 100 0.48 1130795 RFa 0.53 RFb 0.38 HD 0.05 DO 0.03 RD 0.01 0 886-2 RFa 100130795 RFa 0.53 RFb 0.38 HD 0.05 DO 0.03 RD 0.01 0 886-3 RFa 100132361 RFa 0.44 RFb 0.41 DO 0.06 UR 0.06 HD 0.03 0 894-3 RFa 100 0.44 1133147 RFa 0.78 RFb 0.12 DO 0.1 HD 0 0 0 894-1 RFa 100 0.44 1133147 RFa 0.78 RFb 0.12 DO 0.1 HD 0 0 0 894-2 DO 100

15.74 31

4.2.3.4 ICH mw2 PERCENT OVERLAP TEST The calculations for determining the percent overlap for this subzone are found below in Table 64. The percent overlap for the ICH mw2 is (13.1/32) 40.9%. When taking this result into account the fact that 16 out of the 32 polygons sampled had a primary decile of less than 50% should be considered. Regardless of whether or not the primary polygon call matches the plot call, if the primary call makes up less than 50% of the polygon then that sample will fail. Because half of the polygons in the sample set have first deciles of less than 50% this test cannot result in a successful score greater than >55.7%. Table 64. ICH mw2 Percent Overlap Calculations.

Polygons Plots Score Poly-ID S1 1-Dec S2 2-Dec S3 3-

DecS4 4-Dec S5 5-Dec S6 6-Dec S7 7-

Dec PLOT S1 D 1

Olap CNT

125799 RF 0.44 HF 0.3 DF 0.13 HO 0.09 RD 0.04 0 0 417-1 RF 100 0.44 1126104 RF 1 0 0 0 0 0 0 417-2 RF 100 1 1126584 DF 0.44 HF 0.41 HO 0.11 RF 0.04 0 0 0 417-3 RF 100 0.04 1148403 HF 0.76 RF 0.24 0 0 0 0 0 424-3 RF 100 0.24 1149581 RF 0.56 HF 0.38 DF 0.06 0 0 0 0 424-1 RF 100 0.56 1149581 RF 0.56 HF 0.38 DF 0.06 0 0 0 0 424-2 RF 100173679 HF 0.29 RF 0.19 RD 0.18 DF 0.13 HO 0.12 RI 0.09 RS 0 456-1 HF 100 0.29 1178497 RF 0.47 HF 0.3 DF 0.21 HO 0.01 RD 0.01 0 0 456-2 HF 100 0.3 1178497 RF 0.47 HF 0.3 DF 0.21 HO 0.01 RD 0.01 0 0 456-3 HF 100191022 AC 1 0 0 0 0 0 0 475-1 RO 100 0 1191022 AC 1 0 0 0 0 0 0 475-2 RO 100191022 AC 1 0 0 0 0 0 0 475-3 RO 100230227 HF 0.54 RF 0.3 DF 0.09 HO 0.04 RD 0.01 RS 0.02 0 512-1 HF 100 0.54 1230227 HF 0.54 RF 0.3 DF 0.09 HO 0.04 RD 0.01 RS 0.02 0 512-2 HF 100230227 HF 0.54 RF 0.3 DF 0.09 HO 0.04 RD 0.01 RS 0.02 0 512-3 HF 100238249 HF 0.45 DF 0.17 HO 0.15 RD 0.1 RF 0.09 RS 0.02 RH 0.02 520-1 HF 100 0.45 1238249 HF 0.45 DF 0.17 HO 0.15 RD 0.1 RF 0.09 RS 0.02 RH 0.02 520-2 HF 100238249 HF 0.45 DF 0.17 HO 0.15 RD 0.1 RF 0.09 RS 0.02 RH 0.02 520-3 HF 100268593 DF 0.4 HF 0.32 RF 0.17 HO 0.1 RD 0.01 RS 0 0 556-1 HF 100 0.17 1268593 DF 0.4 HF 0.32 RF 0.17 HO 0.1 RD 0.01 RS 0 0 556-2 HO 100268593 DF 0.4 HF 0.32 RF 0.17 HO 0.1 RD 0.01 RS 0 0 556-3 HO 100276262 HF 0.42 DF 0.19 RF 0.16 HO 0.13 RD 0.06 RH 0.02 RS 0.02 562-1 HO 100 0.23 1276262 HF 0.42 DF 0.19 RF 0.16 HO 0.13 RD 0.06 RH 0.02 RS 0.02 562-2 HF 100276262 HF 0.42 DF 0.19 RF 0.16 HO 0.13 RD 0.06 RH 0.02 RS 0.02 562-3 RF 100278042 HF 0.45 DF 0.18 HO 0.14 RF 0.08 RD 0.06 RS 0.04 RH 0.05 558-1 HF 100 0.45 1278042 HF 0.45 DF 0.18 HO 0.14 RF 0.08 RD 0.06 RS 0.04 RH 0.05 558-2 HF 100278042 HF 0.45 DF 0.18 HO 0.14 RF 0.08 RD 0.06 RS 0.04 RH 0.05 558-3 HF 100298220 RF 0.4 HF 0.38 DF 0.08 HO 0.07 RD 0.07 RS 0 0 589-1 HF 100 0.38 1



page 69

298220 RF 0.4 HF 0.38 DF 0.08 HO 0.07 RD 0.07 RS 0 0 589-2 HF 100298220 RF 0.4 HF 0.38 DF 0.08 HO 0.07 RD 0.07 RS 0 0 589-3 HF 100301216 RD 0.35 RS 0.23 HO 0.18 HF 0.11 RH 0.09 RF 0.02 DF 0.02 593-1 RS 100 0.19 1301216 RD 0.35 RS 0.23 HO 0.18 HF 0.11 RH 0.09 RF 0.02 DF 0.02 593-2 HO 100301216 RD 0.35 RS 0.23 HO 0.18 HF 0.11 RH 0.09 RF 0.02 DF 0.02 593-3 HO 100305641 RF 0.8 HF 0.11 DF 0.09 HO 0 RD 0 0 0 JD2-1 RF 100 0.8 1305641 RF 0.8 HF 0.11 DF 0.09 HO 0 RD 0 0 0 JD2-2 RF 100305641 RF 0.8 HF 0.11 DF 0.09 HO 0 RD 0 0 0 JD2-3 RF 100306692 RF 0.65 HF 0.24 DF 0.11 HO 0 RD 0 0 0 CD1-1 RF 100 0.65 1306692 RF 0.65 HF 0.24 DF 0.11 HO 0 RD 0 0 0 CD1-2 RF 100306692 RF 0.65 HF 0.24 DF 0.11 HO 0 RD 0 0 0 CD1-3 RF 100307910 RF 0.56 HF 0.28 RD 0.08 HO 0.04 DF 0.04 RS 0 0 CD2-1 RF 100 0.56 1307910 RF 0.56 HF 0.28 RD 0.08 HO 0.04 DF 0.04 RS 0 0 CD2-2 RF 100307910 RF 0.56 HF 0.28 RD 0.08 HO 0.04 DF 0.04 RS 0 0 CD2-3 RF 100308081 RF 0.75 HF 0.15 DF 0.1 HO 0 RD 0 0 0 595-1 RF 100 0.75 1308081 RF 0.75 HF 0.15 DF 0.1 HO 0 RD 0 0 0 595-2 RF 100308081 RF 0.75 HF 0.15 DF 0.1 HO 0 RD 0 0 0 595-3 RF 100367651 HF 0.51 DF 0.22 HO 0.1 RF 0.09 RD 0.05 RS 0.02 RH 0.01 661-3 RS 80 0.34 1367651 HF 0.51 DF 0.22 HO 0.1 RF 0.09 RD 0.05 RS 0.02 RH 0.01 661-1 HF 100367651 HF 0.51 DF 0.22 HO 0.1 RF 0.09 RD 0.05 RS 0.02 RH 0.01 661-2 HF 100390010 HF 0.5 DF 0.23 RF 0.11 HO 0.1 RD 0.04 RS 0.01 RH 0 702-1 HF 100 0.32 1390010 HF 0.5 DF 0.23 RF 0.11 HO 0.1 RD 0.04 RS 0.01 RH 0 702-2 DF 100390010 HF 0.5 DF 0.23 RF 0.11 HO 0.1 RD 0.04 RS 0.01 RH 0 702-3 DF 100398085 HF 0.48 DF 0.17 HO 0.14 RF 0.11 RD 0.07 RS 0.02 RH 0.01 733-1 HF 100 0.48 1398085 HF 0.48 DF 0.17 HO 0.14 RF 0.11 RD 0.07 RS 0.02 RH 0.01 733-2 HF 100398923 HF 0.33 RH 0.2 DF 0.14 RS 0.12 HO 0.12 RF 0.05 RD 0.04 733-3 HF 100 0.33 1399809 RS 0.28 HF 0.26 HO 0.19 DF 0.1 RD 0.08 RF 0.05 RH 0.04 739-1 BS 100 0.06 1399809 RS 0.28 HF 0.26 HO 0.19 DF 0.1 RD 0.08 RF 0.05 RH 0.04 739-2 BS 100399809 RS 0.28 HF 0.26 HO 0.19 DF 0.1 RD 0.08 RF 0.05 RH 0.04 739-3 HO 100400660 HF 0.55 RF 0.3 HO 0.07 DF 0.05 RD 0.03 RS 0 0 736-1 RF 100 0.3 1400660 HF 0.55 RF 0.3 HO 0.07 DF 0.05 RD 0.03 RS 0 0 736-2 RF 100400660 HF 0.55 RF 0.3 HO 0.07 DF 0.05 RD 0.03 RS 0 0 736-3 RF 100402332 RD 0.66 HO 0.19 HF 0.07 RF 0.05 DF 0.03 RS 0 0 740-1 HF 100 0.07 1402332 RD 0.66 HO 0.19 HF 0.07 RF 0.05 DF 0.03 RS 0 0 740-3 HF 100403149 HF 0.37 RF 0.29 RD 0.16 HO 0.11 DF 0.07 RS 0 0 740-2 HF 100 0.35 1403149 HF 0.37 RF 0.29 RD 0.16 HO 0.11 DF 0.07 RS 0 0 742-1 HF 100403149 HF 0.37 RF 0.29 RD 0.16 HO 0.11 DF 0.07 RS 0 0 742-2 RF 100403149 HF 0.37 RF 0.29 RD 0.16 HO 0.11 DF 0.07 RS 0 0 742-3 HF 100406204 DF 1 0 0 0 0 0 0 745-1 DF 100 1 1406204 DF 1 0 0 0 0 0 0 745-2 DF 100406204 DF 1 0 0 0 0 0 0 745-3 DF 100422339 HF 0.45 RF 0.25 DF 0.18 HO 0.09 RD 0.03 RS 0 0 790-1 RF 100 0.26 1422339 HF 0.45 RF 0.25 DF 0.18 HO 0.09 RD 0.03 RS 0 0 790-2 HO 100422339 HF 0.45 RF 0.25 DF 0.18 HO 0.09 RD 0.03 RS 0 0 790-3 HF 100440797 RF 0.54 HF 0.39 DF 0.06 HO 0 RD 0.01 0 0 833-1 RF 100 0.54 1440797 RF 0.54 HF 0.39 DF 0.06 HO 0 RD 0.01 0 0 833-2 RF 100440867 HF 0.49 RF 0.43 DF 0.04 HO 0.02 RD 0.02 RS 0 0 833-3 RF 100 0.43 1459024 RF 0.53 HF 0.36 RD 0.04 HO 0.04 DF 0.03 0 0 867-1 RF 100 0.53 1459024 RF 0.53 HF 0.36 RD 0.04 HO 0.04 DF 0.03 0 0 867-2 RF 100459024 RF 0.53 HF 0.36 RD 0.04 HO 0.04 DF 0.03 0 0 867-3 RF 100

13.1 32



page 70

Table 65. Summary of Percent Overlap Test Subzone Total Overlap Total Polygons Score ESSF wc1 6.26 13 48.2% ESSF wc4 9.62 14 68.7% ICH dw 15.70 31 50.7% ICH mw2 13.10 32 40.9% Total 90 52.1% Taking into account the distribution of samples in the 4 BEC zones sampled the weighted average Percent overlap is 50%.

4.2.4 ON POINT ACCURACY ASSESSMENT There are 612,284 ha within the Arrow TSA study area. This includes land inside the TSA that is not within Parks and Reserves, Tree Farm Licences, or Private Lands. This 612,284 ha has been mapped using a PEM system that may produce up to 10 deciles within a polygon. This has created some concern because it is possible to decile a polygon to the point where it includes virtually every site series within the subzone. If you were to place a plot within the polygon to check the accuracy of that polygon it wouldn’t matter what the observed site series was because with 10 different site series PEM’d the argument could always be made that the polygon was correct. The reality is that the PEM may have predicted 9 out of 10 wrong. While the potential for this to happen is real it is not the case within the Arrow TSA PEM. A vast majority of the polygons within the PEM are single decile polygons. The 612,284 ha within the study area were mapped with 341,687 polygons. Of these polygons 181,854 (53%) were single decile polygons. In comparison to most TEM’s this is a very high proportion. Table 66 below compares distribution of deciled polygons for 3 different TEM’s as compared to the Arrow TSA PEM. Table 66. Distribution of Polygon Deciles

Project SingleDecile

Single % Double Decile

Double % TripleDecile

Triple % Total Polygons

Arrow PEM 181854 53% 68391 20% 38562 11% 341687 Slocan TEM 423 7% 2016 34% 3544 59% 5983 TFL14 TEM 1568 17% 3894 43% 3527 39% 8989 Steamboat TEM 1026 19% 2136 39% 2358 43% 5520 Compared to similar mapping products PEM is much more spatially discrete relying less on deciles to adequately describe the site series within a polygon. It is more likely for the PEM to make a single site series call within a polygon than it is for TEM. There is no accuracy data available for any of these TEM’s so we can not compare the spatial accuracy of the products. What table 66 does suggest is that PEM product is much more likely to provide data that can be judged on a right or wrong basis with a single plot call than a TEM product, which is the most comparable type of mapping available.



page 71

Table 67 describes 194 accuracy assessment observations made in the field for 4 different subzones compared to the PEM predicted site series for that location. This is an independent data set that was not used in the model building process.

Table 67. Plot and Polygon Comparison PLOT PlotCall POLY# SS1 D1 SS2 D2 SS3 D3 440-1 ESSFwc1 FD 374069 ESSFwc1 FF 62% ESSFwc1 FR 29% ESSFwc1 FD 9%440-2 ESSFwc1 FD 374069 ESSFwc1 FF 62% ESSFwc1 FR 29% ESSFwc1 FD 9%440-3 ESSFwc1 FD 374069 ESSFwc1 FF 62% ESSFwc1 FR 29% ESSFwc1 FD 9%537-3 ESSFwc1 FD 519265 ESSFwc1 FR 53% ESSFwc1 FF 29% ESSFwc1 FD 16%550-1 ESSFwc1 FF 538463 ESSFwc1 FF 43% ESSFwc1 FD 22% ESSFwc1 FR 19%716-1 ESSFwc1 FF 841140 ESSFwc1 FF 78% ESSFwc1 FR 22% 0%725-1 ESSFwc1 FF 847502 ESSFwc1 FF 67% ESSFwc1 FR 24% ESSFwc1 FD 9%725-2 ESSFwc1 FF 847502 ESSFwc1 FF 67% ESSFwc1 FR 24% ESSFwc1 FD 9%725-3 ESSFwc1 FF 847502 ESSFwc1 FF 67% ESSFwc1 FR 24% ESSFwc1 FD 9%843-1 ESSFwc1 FF 962950 ESSFwc1 FR 91% ESSFwc1 FF 9% 0%843-2 ESSFwc1 FF 962950 ESSFwc1 FR 91% ESSFwc1 FF 9% 0%843-3 ESSFwc1 FF 962950 ESSFwc1 FR 91% ESSFwc1 FF 9% 0%419-1 ESSFwc1 FR 335555 ESSFwc1 FR 63% ESSFwc1 FF 25% ESSFwc1 FD 12%419-2 ESSFwc1 FR 338998 ESSFwc1 FR 67% ESSFwc1 FF 29% ESSFwc1 FD 4%419-3 ESSFwc1 FR 338998 ESSFwc1 FR 67% ESSFwc1 FF 29% ESSFwc1 FD 4%423-1 ESSFwc1 FR 343377 ESSFwc1 FR 60% ESSFwc1 FF 23% ESSFwc1 FD 17%423-2 ESSFwc1 FR 343377 ESSFwc1 FR 60% ESSFwc1 FF 23% ESSFwc1 FD 17%423-3 ESSFwc1 FR 343377 ESSFwc1 FR 60% ESSFwc1 FF 23% ESSFwc1 FD 17%523-1 ESSFwc1 FR 491366 ESSFwc1 FF 83% ESSFwc1 FR 16% ESSFwc1 FD 1%523-3 ESSFwc1 FR 491366 ESSFwc1 FF 83% ESSFwc1 FR 16% ESSFwc1 FD 1%537-1 ESSFwc1 FR 519265 ESSFwc1 FR 53% ESSFwc1 FF 29% ESSFwc1 FD 16%537-2 ESSFwc1 FR 519265 ESSFwc1 FR 53% ESSFwc1 FF 29% ESSFwc1 FD 16%550-2 ESSFwc1 FR 539124 ESSFwc1 FR 43% ESSFwc1 FD 30% ESSFwc1 FF 27%550-3 ESSFwc1 FR 539124 ESSFwc1 FR 43% ESSFwc1 FD 30% ESSFwc1 FF 27%664-1 ESSFwc1 FR 797680 ESSFwc1 FR 69% ESSFwc1 FF 22% ESSFwc1 FD 9%664-2 ESSFwc1 FR 797680 ESSFwc1 FR 69% ESSFwc1 FF 22% ESSFwc1 FD 9%664-3 ESSFwc1 FR 797680 ESSFwc1 FR 69% ESSFwc1 FF 22% ESSFwc1 FD 9%716-2 ESSFwc1 FR 841832 ESSFwc1 FR 61% ESSFwc1 FF 27% ESSFwc1 FD 12%716-3 ESSFwc1 FR 841832 ESSFwc1 FR 61% ESSFwc1 FF 27% ESSFwc1 FD 12%630-1 ESSFwc4 FF 762778 ESSFwc4 FR 88% ESSFWC4 RF 12% 0%633-1 ESSFwc4 FL 765706 ESSFwc4 FR 89% ESSFWC4 RF 11% 0%153-1 ESSFwc4 FR 379378 ESSFwc4 RF 44% ESSFWC4 FR 28% ESSFWC4 FF 24%153-2 ESSFwc4 FR 379378 ESSFwc4 RF 44% ESSFWC4 FR 28% ESSFWC4 FF 24%153-3 ESSFwc4 FR 379378 ESSFwc4 RF 44% ESSFWC4 FR 28% ESSFWC4 FF 24%504-1 ESSFwc4 FR 461795 ESSFwc4 FR 87% ESSFWC4 RF 12% ESSFWC4 FF 1%504-2 ESSFwc4 FR 461795 ESSFwc4 FR 87% ESSFWC4 RF 12% ESSFWC4 FF 1%504-3 ESSFwc4 FR 462832 ESSFwc4 FR 88% ESSFWC4 RF 11% ESSFWC4 FF 1%526-1 ESSFwc4 FR 497989 ESSFwc4 FR 93% ESSFWC4 RF 7% 0%526-2 ESSFwc4 FR 497989 ESSFwc4 FR 93% ESSFWC4 RF 7% 0%526-3 ESSFwc4 FR 495067 ESSFwc4 FR 96% ESSFWC4 RF 3% ESSFWC4 FF 1%528-1 ESSFwc4 FR 500427 ESSFwc4 FR 89% ESSFWC4 RF 11% 0%528-2 ESSFwc4 FR 500427 ESSFwc4 FR 89% ESSFWC4 RF 11% 0%528-3 ESSFwc4 FR 502144 ESSFwc4 FR 88% ESSFWC4 RF 12% 0%533-1 ESSFwc4 FR 502465 ESSFwc4 FR 88% ESSFWC4 RF 12% 0%533-2 ESSFwc4 FR 502465 ESSFwc4 FR 88% ESSFWC4 RF 12% 0%



page 72

533-3 ESSFwc4 FR 507021 ESSFwc4 FR 100% 0% 0%630-3 ESSFwc4 FR 762778 ESSFwc4 FR 88% ESSFWC4 RF 12% 0%633-2 ESSFwc4 FR 766928 ESSFwc4 FR 88% ESSFWC4 RF 12% 0%633-3 ESSFwc4 FR 766928 ESSFwc4 FR 88% ESSFWC4 RF 12% 0%901-1 ESSFwc4 FR 766928 ESSFwc4 FR 88% ESSFWC4 RF 12% 0%901-2 ESSFwc4 FR 766928 ESSFwc4 FR 88% ESSFWC4 RF 12% 0%901-3 ESSFwc4 FR 766928 ESSFwc4 FR 88% ESSFWC4 RF 12% 0%573-3 ESSFwc4 FW 588303 ESSFwc4 FR 100% 0% 0%573-1 ESSFwc4 RF 588303 ESSFwc4 FR 100% 0% 0%573-2 ESSFwc4 RF 588303 ESSFwc4 FR 100% 0% 0%630-2 ESSFwc4 SS 762778 ESSFwc4 FR 88% ESSFWC4 RF 12% 0%545-1 ICHdw DO 521614 ICHdw DO 61% ICHdw RFa 39% 0%545-2 ICHdw DO 521614 ICHdw DO 61% ICHdw RFa 39% 0%545-3 ICHdw DO 521614 ICHdw DO 61% ICHdw RFa 39% 0%602-1 ICHdw DO 669813 ICHdw DO 51% ICHdw RFa 34% ICHdw RO 9%670-1 ICHdw DO 812830 ICHdw DO 100% 0% 0%705-1 ICHdw DO 832670 ICHdw Rfa 53% ICHdw RFb 47% 0%731-2 ICHdw DO 864161 ICHdw DO 100% 0% 0%731-3 ICHdw DO 861549 ICHdw Rfa 100% 0% 0%765-1 ICHdw DO 880783 ICHdw DO 100% 0% 0%799-1 ICHdw DO 900077 ICHdw DO 100% 0% 0%799-2 ICHdw DO 900077 ICHdw DO 100% 0% 0%799-3 ICHdw DO 900077 ICHdw DO 100% 0% 0%894-2 ICHdw DO 1011659 ICHdw Rfa 78% ICHdw RFb 12% ICHdw DO 10%705-3 ICHdw HD 832031 ICHdw Rfa 75% ICHdw RFb 15% ICHdw RO 10%1260-1 ICHdw RFa 924965 ICHdw Rfa 48% ICHdw RFb 25% ICHdw DO 24%1260-2 ICHdw RFa 924965 ICHdw Rfa 48% ICHdw RFb 25% ICHdw DO 24%1260-3 ICHdw RFa 924965 ICHdw Rfa 48% ICHdw RFb 25% ICHdw DO 24%1353-1 ICHdw RFa 1004382 ICHdw Rfa 62% ICHdw DO 27% ICHdw RFb 11%1353-2 ICHdw RFa 1004382 ICHdw Rfa 62% ICHdw DO 27% ICHdw RFb 11%1353-3 ICHdw RFa 1004382 ICHdw Rfa 62% ICHdw DO 27% ICHdw RFb 11%601-1 ICHdw RFa 585453 ICHdw RFb 58% ICHdw RFa 21% ICHdw HD 15%601-2 ICHdw RFa 585453 ICHdw RFb 58% ICHdw RFa 21% ICHdw HD 15%601-3 ICHdw RFa 585453 ICHdw RFb 58% ICHdw RFa 21% ICHdw HD 15%602-2 ICHdw RFa 669813 ICHdw DO 51% ICHdw RFa 34% ICHdw RO 9%602-3 ICHdw RFa 670241 ICHdw Rfa 76% ICHdw DO 24% 0%670-2 ICHdw RFa 812830 ICHdw DO 100% 0% 0%670-3 ICHdw RFa 807608 ICHdw Rfa 90% ICHdw RFb 8% ICHdw DO 2%673-1 ICHdw RFa 809611 ICHdw Rfa 73% ICHdw RFb 18% ICHdw DO 5%673-2 ICHdw RFa 809611 ICHdw Rfa 73% ICHdw RFb 18% ICHdw DO 5%694-1 ICHdw RFa 825699 ICHdw Rfa 84% ICHdw DO 12% ICHdw RFb 4%694-2 ICHdw RFa 825699 ICHdw Rfa 84% ICHdw DO 12% ICHdw RFb 4%694-3 ICHdw RFa 825699 ICHdw Rfa 84% ICHdw DO 12% ICHdw RFb 4%705-2 ICHdw RFa 832670 ICHdw Rfa 53% ICHdw RFb 47% 0%731-1 ICHdw RFa 861549 ICHdw Rfa 100% 0% 0%765-2 ICHdw RFa 870631 ICHdw Rfa 100% 0% 0%765-3 ICHdw RFa 870631 ICHdw Rfa 100% 0% 0%886-2 ICHdw RFa 1006985 ICHdw Rfa 53% ICHdw RFb 38% ICHdw HD 5%886-3 ICHdw RFa 1006985 ICHdw Rfa 53% ICHdw RFb 38% ICHdw HD 5%894-1 ICHdw RFa 1011659 ICHdw Rfa 78% ICHdw RFb 12% ICHdw DO 10%894-3 ICHdw RFa 1010227 ICHdw Rfa 44% ICHdw RFb 41% ICHdw DO 6%JD1-1 ICHdw RFa 635974 ICHdw Rfa 70% ICHdw DO 30% 0%



page 73

JD1-2 ICHdw RFa 635974 ICHdw Rfa 70% ICHdw DO 30% 0%JD1-3 ICHdw RFa 635974 ICHdw Rfa 70% ICHdw DO 30% 0%VR1-2 ICHdw RFa 798074 ICHdw RFb 70% ICHdw HD 20% ICHdw RD 10%VR1-3 ICHdw RFa 798210 ICHdw Rfa 60% ICHdw RFb 40% 0%603-1 ICHdw RFb 673420 ICHdw RFb 45% ICHdw HD 37% ICHdw XF 14%603-2 ICHdw RFb 670384 ICHdw Rfa 51% ICHdw RFb 30% ICHdw RO 17%603-3 ICHdw RFb 670384 ICHdw Rfa 51% ICHdw RFb 30% ICHdw RO 17%660-1 ICHdw RFb 795627 ICHdw RFb 38% ICHdw HD 29% ICHdw XF 22%660-2 ICHdw RFb 795627 ICHdw RFb 38% ICHdw HD 29% ICHdw XF 22%660-3 ICHdw RFb 795627 ICHdw RFb 38% ICHdw HD 29% ICHdw XF 22%666-1 ICHdw RFb 803207 ICHdw Rfa 87% ICHdw RFb 7% ICHdw DO 5%666-2 ICHdw RFb 803207 ICHdw Rfa 87% ICHdw RFb 7% ICHdw DO 5%666-3 ICHdw RFb 803207 ICHdw Rfa 87% ICHdw RFb 7% ICHdw DO 5%673-3 ICHdw RFb 809611 ICHdw Rfa 73% ICHdw RFb 18% ICHdw DO 5%772-1 ICHdw RFb 881570 ICHdw RFb 67% ICHdw RFa 12% ICHdw UR 12%772-2 ICHdw RFb 880054 ICHdw RFb 66% ICHdw RFa 22% ICHdw HD 9%772-3 ICHdw RFb 881570 ICHdw RFb 67% ICHdw RFa 12% ICHdw UR 12%886-1 ICHdw RFb 1006985 ICHdw Rfa 53% ICHdw RFb 38% ICHdw HD 5%VR1-1 ICHdw RFb 798686 ICHdw Rfa 60% ICHdw RFb 40% 0%739-1 ICHmw2 BS 856711 ICHmw2 RS 28% ICHmw2 HF 26% ICHmw2 HO 19%739-2 ICHmw2 BS 856711 ICHmw2 RS 28% ICHmw2 HF 26% ICHmw2 HO 19%702-2 ICHmw2 DF 839886 ICHmw2 HF 50% ICHmw2 DF 23% ICHmw2 RF 11%702-3 ICHmw2 DF 839886 ICHmw2 HF 50% ICHmw2 DF 23% ICHmw2 RF 11%745-1 ICHmw2 DF 868430 ICHmw2 DF 100% 0% 0%745-2 ICHmw2 DF 868430 ICHmw2 DF 100% 0% 0%745-3 ICHmw2 DF 868430 ICHmw2 DF 100% 0% 0%456-1 ICHmw2 HF 381326 ICHmw2 HF 29% ICHmw2 RF 19% ICHmw2 RD 18%456-2 ICHmw2 HF 390209 ICHmw2 RF 47% ICHmw2 HF 30% ICHmw2 DF 21%456-3 ICHmw2 HF 390209 ICHmw2 RF 47% ICHmw2 HF 30% ICHmw2 DF 21%512-1 ICHmw2 HF 476127 ICHmw2 HF 54% ICHmw2 RF 30% ICHmw2 DF 9%512-2 ICHmw2 HF 476127 ICHmw2 HF 54% ICHmw2 RF 30% ICHmw2 DF 9%512-3 ICHmw2 HF 476127 ICHmw2 HF 54% ICHmw2 RF 30% ICHmw2 DF 9%520-1 ICHmw2 HF 489691 ICHmw2 HF 45% ICHmw2 DF 17% ICHmw2 HO 15%520-2 ICHmw2 HF 489691 ICHmw2 HF 45% ICHmw2 DF 17% ICHmw2 HO 15%520-3 ICHmw2 HF 489691 ICHmw2 HF 45% ICHmw2 DF 17% ICHmw2 HO 15%556-1 ICHmw2 HF 552427 ICHmw2 DF 40% ICHmw2 HF 32% ICHmw2 RF 17%558-1 ICHmw2 HF 574517 ICHmw2 HF 45% ICHmw2 DF 18% ICHmw2 HO 14%558-2 ICHmw2 HF 574517 ICHmw2 HF 45% ICHmw2 DF 18% ICHmw2 HO 14%558-3 ICHmw2 HF 574517 ICHmw2 HF 45% ICHmw2 DF 18% ICHmw2 HO 14%562-2 ICHmw2 HF 570229 ICHmw2 HF 42% ICHmw2 DF 19% ICHmw2 RF 16%589-1 ICHmw2 HF 625949 ICHmw2 RF 40% ICHmw2 HF 38% ICHmw2 DF 8%589-2 ICHmw2 HF 625949 ICHmw2 RF 40% ICHmw2 HF 38% ICHmw2 DF 8%589-3 ICHmw2 HF 625949 ICHmw2 RF 40% ICHmw2 HF 38% ICHmw2 DF 8%661-1 ICHmw2 HF 790804 ICHmw2 HF 51% ICHmw2 DF 22% ICHmw2 HO 10%661-2 ICHmw2 HF 790804 ICHmw2 HF 51% ICHmw2 DF 22% ICHmw2 HO 10%702-1 ICHmw2 HF 839886 ICHmw2 HF 50% ICHmw2 DF 23% ICHmw2 RF 11%733-1 ICHmw2 HF 853285 ICHmw2 HF 48% ICHmw2 DF 17% ICHmw2 HO 14%733-2 ICHmw2 HF 853285 ICHmw2 HF 48% ICHmw2 DF 17% ICHmw2 HO 14%733-3 ICHmw2 HF 854977 ICHmw2 HF 33% ICHmw2 RH 20% ICHmw2 DF 14%740-1 ICHmw2 HF 861520 ICHmw2 RD 66% ICHmw2 HO 19% ICHmw2 HF 7%740-2 ICHmw2 HF 863001 ICHmw2 HF 37% ICHmw2 RF 29% ICHmw2 RD 16%740-3 ICHmw2 HF 861520 ICHmw2 RD 66% ICHmw2 HO 19% ICHmw2 HF 7%



page 74

742-1 ICHmw2 HF 863001 ICHmw2 HF 37% ICHmw2 RF 29% ICHmw2 RD 16%742-3 ICHmw2 HF 863001 ICHmw2 HF 37% ICHmw2 RF 29% ICHmw2 RD 16%790-3 ICHmw2 HF 897478 ICHmw2 HF 45% ICHmw2 RF 25% ICHmw2 DF 18%556-2 ICHmw2 HO 552427 ICHmw2 DF 40% ICHmw2 HF 32% ICHmw2 RF 17%556-3 ICHmw2 HO 552427 ICHmw2 DF 40% ICHmw2 HF 32% ICHmw2 RF 17%562-1 ICHmw2 HO 570229 ICHmw2 HF 42% ICHmw2 DF 19% ICHmw2 RF 16%593-2 ICHmw2 HO 630959 ICHmw2 RD 35% ICHmw2 RS 23% ICHmw2 HO 18%593-3 ICHmw2 HO 630959 ICHmw2 RD 35% ICHmw2 RS 23% ICHmw2 HO 18%739-3 ICHmw2 HO 856711 ICHmw2 RS 28% ICHmw2 HF 26% ICHmw2 HO 19%790-2 ICHmw2 HO 897478 ICHmw2 HF 45% ICHmw2 RF 25% ICHmw2 DF 18%417-1 ICHmw2 RF 313177 ICHmw2 RF 44% ICHmw2 HF 30% ICHmw2 DF 13%417-2 ICHmw2 RF 313621 ICHmw2 RF 100% 0% 0%417-3 ICHmw2 RF 314328 ICHmw2 DF 44% ICHmw2 HF 41% ICHmw2 HO 11%424-1 ICHmw2 RF 344866 ICHmw2 RF 56% ICHmw2 HF 38% ICHmw2 DF 6%424-2 ICHmw2 RF 344866 ICHmw2 RF 56% ICHmw2 HF 38% ICHmw2 DF 6%424-3 ICHmw2 RF 343305 ICHmw2 HF 76% ICHmw2 RF 24% 0%562-3 ICHmw2 RF 570229 ICHmw2 HF 42% ICHmw2 DF 19% ICHmw2 RF 16%595-1 ICHmw2 RF 645001 ICHmw2 RF 75% ICHmw2 HF 15% ICHmw2 DF 10%595-2 ICHmw2 RF 645001 ICHmw2 RF 75% ICHmw2 HF 15% ICHmw2 DF 10%595-3 ICHmw2 RF 645001 ICHmw2 RF 75% ICHmw2 HF 15% ICHmw2 DF 10%736-1 ICHmw2 RF 858548 ICHmw2 HF 55% ICHmw2 RF 30% ICHmw2 HO 7%736-2 ICHmw2 RF 858548 ICHmw2 HF 55% ICHmw2 RF 30% ICHmw2 HO 7%736-3 ICHmw2 RF 858548 ICHmw2 HF 55% ICHmw2 RF 30% ICHmw2 HO 7%742-2 ICHmw2 RF 863001 ICHmw2 HF 37% ICHmw2 RF 29% ICHmw2 RD 16%790-1 ICHmw2 RF 897478 ICHmw2 HF 45% ICHmw2 RF 25% ICHmw2 DF 18%833-1 ICHmw2 RF 941287 ICHmw2 RF 54% ICHmw2 HF 39% ICHmw2 DF 6%833-2 ICHmw2 RF 941287 ICHmw2 RF 54% ICHmw2 HF 39% ICHmw2 DF 6%833-3 ICHmw2 RF 941443 ICHmw2 HF 49% ICHmw2 RF 43% ICHmw2 DF 4%867-1 ICHmw2 RF 985384 ICHmw2 RF 53% ICHmw2 HF 36% ICHmw2 RD 4%867-2 ICHmw2 RF 985384 ICHmw2 RF 53% ICHmw2 HF 36% ICHmw2 RD 4%867-3 ICHmw2 RF 985384 ICHmw2 RF 53% ICHmw2 HF 36% ICHmw2 RD 4%CD1-1 ICHmw2 RF 642230 ICHmw2 RF 65% ICHmw2 HF 24% ICHmw2 DF 11%CD1-2 ICHmw2 RF 642230 ICHmw2 RF 65% ICHmw2 HF 24% ICHmw2 DF 11%CD1-3 ICHmw2 RF 642230 ICHmw2 RF 65% ICHmw2 HF 24% ICHmw2 DF 11%CD2-1 ICHmw2 RF 644692 ICHmw2 RF 56% ICHmw2 HF 28% ICHmw2 RD 8%CD2-2 ICHmw2 RF 644692 ICHmw2 RF 56% ICHmw2 HF 28% ICHmw2 RD 8%CD2-3 ICHmw2 RF 644692 ICHmw2 RF 56% ICHmw2 HF 28% ICHmw2 RD 8%JD2-1 ICHmw2 RF 639974 ICHmw2 RF 80% ICHmw2 HF 11% ICHmw2 DF 9%JD2-2 ICHmw2 RF 639974 ICHmw2 RF 80% ICHmw2 HF 11% ICHmw2 DF 9%JD2-3 ICHmw2 RF 639974 ICHmw2 RF 80% ICHmw2 HF 11% ICHmw2 DF 9%475-1 ICHmw2 RO 412573 ICHmw2 AC 100% 0% 0%475-2 ICHmw2 RO 412573 ICHmw2 AC 100% 0% 0%475-3 ICHmw2 RO 412573 ICHmw2 AC 100% 0% 0%593-1 ICHmw2 RS 630959 ICHmw2 RD 35% ICHmw2 RS 23% ICHmw2 HO 18%661-3 ICHmw2 RS 790804 ICHmw2 HF 51% ICHmw2 DF 22% ICHmw2 HO 10% 125 of these field calls match the first decile call within the PEM. That means that 64.4% of the plot observations are in agreement with primary PEM call. If the second decile is taken into account 155 or 79.8%of the plot calls are in agreement with the first or second deciles of the PEM. Taking into account the third decile in the PEM 166 or 85.5% of the field calls are in agreement with the PEM . Taking into account the fourth



page 75

decile 171 or 88% of the plot field calls are in agreement with the PEM. None of the plot calls match 5th,6th,7th,8th,9th,or 10th decile. This should not be surprising though as these deciles generally describe very limited areas that are not representative of the overall polygon and not likely to be sampled. Some might argue that this approach to assessing the overall accuracy of the PEM product is opportunistic. It would make sense that the more deciles found within a polygon the more likely the plot would record a match with the site series predicted within that polygon. This is certainly the case if the selection of site series within the polygon were random. It would be expected that plots with a single call would be unlikely predict the right ecosystem while plots with 7 deciles would be much more likely to predict the right ecosystem. Table 68 describes the results of comparing plot calls to PEM calls based on the number of deciles in each polygon.

Table 68. Relationship of Correct Calls to Deciles Number of

Deciles Number of

Plots Number Correct

Average

1 22 14 64%2 32 27 84%3 47 45 96%4 19 19 100%5 46 41 89%6 4 4 100%7 24 21 88%

To clarify this table, there are 22 plots that fall within polygons containing only 1 decile. 64% of these plots had the same field call as the PEM. You would expect that the more opportunity (more deciles) the plot had to be correct the better but this is not the case with the data set. In fact, plots with 7 deciles do not score as high as those with 3 deciles. This table suggests that while there may be some relationship between the number of deciles and the ability of the PEM to predict the correct site series it is not a very significant one and is unlikely to skew the results of an on point accuracy assessment. Comparing the observed site series in the field to the site series found within the PEM polygons results in an accuracy score of 171 out of 194 or 88.1%.



page 76

5.0 REFERENCES CITED BC Forest Productivity Council, Site Productivity Working Group June 2001. SIBEC Sampling and Data Standards, Version 5.1, Victoria, BC Braumandl, T. and M. Curran. 1992.A Field Guide for Site Identification and Interpretation for the Nelson Forest Region. Land Management Handbook Number 20. BC Ministry of Forests. ESRI, 2001. Personal Communication. Howes D.E. and E. Kenk (eds.) 1997. Terrain Classification System for British Columbia. Ketcheson, M., T. Dool, L. Bradley, G. Kernaghan, J. Shypitka, T. Robertson, K. Misurak, V. Lipinski and D. Ross. 2002. Okanagan TSA Dry Belt Predictive Ecosystem Mapping (PEM) and Site Index Adjustment. An unpublished report to Glen Dick, Okanagan Innovative Forest Practices Association, Lumby BC. Ketcheson, M., T. Dool and S. Wilson. 2001. Arrow TSA Site Index Adjustment Project, Predictive Ecosystem Mapping (PEM) Year Two Final Project Report. A report and maps for Paul Jeakins, Arrow IFPA, Slocan BC. Lloyd, D. 1990. A Guide to Site Identification and Interpretation for the Kamloops Forest Region. Land Management Handbook Number 23. BC Ministry of Forests. Lloyd, D. 2002a. Revised Site Series Classification and Localized BEC Lines for the IDFdk1, IDFdm1, IDFdm2, MSdm1, MSdm2, MSxk, ESSFdc1, ESSFdc2, and ESSFxc. Unpublished Report Tables and Maps to Glen Dick, Okanagan Innovative Forest Practices Association. Lloyd, D. 2002b. Review of Dry Belt Okanagan PEM Map Reliability. Unpublished report to Glen Dick, Okanagan Innovative Forest Practices Association. Marcoux, D., M. Mathers, J. Riddell and M.V. Ketcheson. 1996. Terrestrial Ecosystem Mapping with Wildlife Interpretations for the Wilson-Stagleap area (near Nelson, B.C.). 1 map sheet. Unpublished Report to John Gwilliam, Columbia Basin Fish & Wildlife Compensation Program, Nelson, B.C. Marcoux, D., J. Riddell and M.V. Ketcheson. 1997. Terrestrial Ecosystem Mapping with Wildlife Interpretations for the Broadwater area (near Castlegar, B.C.). 1 map sheet. Unpublished Report to John Gwilliam, Columbia Basin Fish & Wildlife Compensation Program, Nelson, B.C. Meidinger, D. 2000. Protocol for Quality Assurance and Accuracy Assessment of Ecosystem Maps. Research Branch, BC Ministry of Forests, Victoria.



page 77

Meidinger, D. 2001. Ecosystem mapping Accuracy and Timber Supply Applications. BC Ministry of Forests Research Branch, Victoria BC. Resources Inventory Committee prepared by Terrestrial Ecosystem Mapping Alternatives Task Force. 1999. Standards for Predicitive Ecosystem Mapping - Inventory Standard. Version1.0. Serrouya, Rob. Personal communication March 2002. Arrow Innovative Forest Practices Association, Kokanee Forest Consultants, Nelson BC. Zimmermann, Niklaus E. 2000. toposcale.aml. www.wsl.ch/staff/niklaus.zimmerman/programs.html#4

ARROW TSA PREDICTIVE ECOSYSTEM MAPPING (PEM) 2003 RE...

Documents

Transcript of ARROW TSA PREDICTIVE ECOSYSTEM MAPPING (PEM) 2003 RE...