Cranbrook TSA IDF and PP Ground Call Analysis PREPARED FOR: MINISTRY OF FORESTS AND RANGE SOUTHERN INTERIOR REGION PREPARED BY: Jahraus & Associates Consulting Inc. Maple Ridge BC & Churlish Consulting Ltd. Victoria BC February 2008 Cranbrook TSA: IDF and PP Ground Call Analysis i Table of Contents 1 2 3 4 5 6 7 8 Introduction ................................................................................................................. 1 Scope and Objectives................................................................................................... 1 Methods ....................................................................................................................... 1 3.1 Ground data and forest cover attributes ............................................................... 1 3.2 Stratification ......................................................................................................... 3 Analysis Results .......................................................................................................... 4 4.1 Comparison of ground and inventory values for basic forest cover attributes..... 4 4.2 Comparison of ground and inventory volume...................................................... 4 4.3 Coefficients of variation and sample size ............................................................ 7 Discussion and Recommendations .............................................................................. 8 APPENDIX A............................................................................................................ 11 APPENDIX B ............................................................................................................ 14 APPENDIX C ............................................................................................................ 15 February 2008 Cranbrook TSA: IDF and PP Ground Call Analysis 1 1 Introduction This project was initiated by Ministry of Forests & Range (MFR) in the Southern Interior region in response to MFR and Licensee concerns in the Rocky Mountain Forest District (RMFD) regarding potential overestimation of Forest Cover Inventory volumes in the Interior Douglas Fir and Ponderosa Pine Biogeoclimatic zones of the Cranbrook Timber Supply Area. In the summer of 2007, 52 ground samples were randomly established in forested polygons (>10% crown Closure and > 30 years of age) in the IDF and PP Biogeoclimatic zones within the Cranbrook TSA portion of the RMFD. Areas in private lands, First Nations Reserves, Parks and TFL’s were been excluded. This analysis encompassed the compilation of this field data and comparison with Forest Cover attributes. 2 Scope and Objectives The terms of reference and task list for this project are provided below: 1. Compile ground data (includes: converting data from MDB format to one suitable for input into the VRI compiler; making modifications to the VRI compiler to accommodate the non-standard data collection) 2. Run VDYP7 and VDYP6 on sample inventory data to project attributes to year of ground sampling 3. Compare ground attributes and Forest Cover attributes for the established samples and compute age, height, basal area and trees/ha ratios and sampling errors (as well as coefficients of variation), as per a standard statistical adjustment. Logical leading species strata will be used in consultation with the contract coordinator. 4. Using the compiled volume data, estimate sample size requirements for phase 2 sampling using the strata defined in step 1 and the population as a whole. 5. Replicate these processes using both VDYP6 and VDYP7 and provide discussion of any noted differences. 3 Methods 3.1 Ground data and forest cover attributes Details on the project background, sample selection and data collection are provided in Appendix A. Forest cover attributes were provided by Matt Makar. All samples were from F-type (FIP) inventories and hence did not include second species heights or ages, photo-interpreted basal area/ha or trees/ha. VDYP6 v.6.6d was used to generate the VDYP6 volumes. Data was first projected to 2007. Note that this version of VDYP6 uses SINDEX version 2.3 for height projection. For the VDYP7 portion of the analysis, VDYP7 Console was used. All samples were first projected to 2007 (VDYP7 used SINDEX version 3.3 for height projection). February 2008 Cranbrook TSA: IDF and PP Ground Call Analysis 2 The ground sample data was compiled using the VRI compiler with adaptations to accommodate the data collection procedures used in this project1. Since live/dead indicators were not collected, all trees were assumed to be live2. In addition, pathological indicators were not collected for the ground sample trees3 hence decay could not be estimated. As a result, all volume comparisons were done on a whole stem volume (WSV) basis. The analysis compared ground attributes with forest cover attributes and computed ratios and associated sampling errors. The methodology for computing the ratios was based on the standards and procedures for a VRI statistical adjustment. Height and age ratios were computed based on forest cover attributes from VDYP6 and height, age, basal area/ha & trees/ha ratios were computed based on forest cover attributes from VDYP7. These ratios would approximate the VRI Stage 1 adjustment factors. In this project, ages were collected for “representative” trees but heights were collected for all trees4. The ground ages and heights used in the ratio comparisons were based on those collected for only the “representative” trees5. Since pathological indicators were not collected and net factoring was not applied, ground volumes net decay were not available from the compiler. Hence, VRI-type adjustment factors for volume (i.e. Stage 2 adjustment factors) could not be computed. However, volume comparison ratios (similar to what are referred to as volume “impact” ratios in a VRI adjustment analysis) were developed for whole stem volumes. These volume comparison ratios were estimated by comparing the ground compiled mean whole stem volumes (WSV) with the unadjusted6 inventory mean volumes from VDYP6 and VDYP7. Although VDYP7 generates estimates of WSV at various utilization levels, VDYP6 only produces WSV at the 7.5cm+ dbh utilization level. Therefore a comparison at the 7.5cm+ dbh utilization was computed for both VDYP6 and VDYP7 and a comparison at 12.5cm+ dbh was computed for VDYP7 only. 1 Data collection followed Phase 1 ground call methodology rather than VRI Phase 2 ground sample methodology. 2 Sample #2 included a comment “all trees except 1 dead from fire approx 2yrs.” but the tree number for the single live tree was not indicated. Similarly, sample #78 indicated “dead ibm pl in plot” but the tree number for the dead tree was not specified. Since the dead trees could not be identified, all trees were assumed to be live. Broken top trees were identified, but neither broken top diameter nor projected heights were provided so the height listed became the height used for compilation. 3 The standard VRI net factoring procedure was not applied in this project. 4 The methodology was similar to that used for Phase 1 ground calls, as opposed to the VRI top height methods. 5 The heights that were collected on all trees were used in the compilation of ground volume. However, for the height ratio comparison, only trees that had ages were used to compute the average height for a sample. 6 Unadjusted volumes refer to those based on unadjusted forest cover attributes e.g. heights and ages that have not been ratio-adjusted based on ground sample information. February 2008 Cranbrook TSA: IDF and PP Ground Call Analysis 3 3.2 Stratification The distribution of the samples, by biogeoclimatic zone and leading species, is shown in Table 1. More than 60% of the samples were fir leading, based on the inventory species composition7. Table 1: Distribution of samples. Inventory leading species Biogeoclimatic Zone IDF PP Total AC EP FD LW PL PY SE 1 1 25 6 6 2 0 0 0 7 2 0 1 1 1 1 32 8 6 3 1 Total 41 11 52 Various leading species and biogeoclimatic zone-based strata were considered. However, given the distribution of samples (almost half of the samples being fir leading in IDF), the options were somewhat limited. When the number of samples within a stratum is small (as we see here for most non-fir leading species), there was a risk that any trends seen in these minor strata may be spurious. With the relatively large number of samples in the fir leading stratum, there was potential for stratification within this leading species group (by age or leading species % within the fir leading stratum). However, this more detailed level of stratification within the fir leading samples was not pursued in this analysis. It was suggested that larch leading may be a stratum of interest; this was included in one of the stratification alternatives described below. For this project, 3 stratification alternatives (referred to as “runs”) were considered: 1) No stratification (analysis based on all 52 samples as a whole) 2) Two strata: fir leading & non-fir leading 3) Three strata: fir leading, larch leading & non-fir/non-larch leading. Results for each of these three runs are provided in section 4 that follows. In addition, summary results based on a biogeoclimatic (BEC) zone stratification are provided in Appendix B. Since trends by leading species were stronger than trends by BEC, the BECbased stratification was not pursued further. 7 One sample was pine leading in the FIP file (volume-based species composition) but when the file was “completed” (run through VDYP7 where species composition is converted to a basal area basis), the leading species changed to fir. The VDYP7 leading species (i.e. Fd) was used for stratum assignment in this analysis. February 2008 Cranbrook TSA: IDF and PP Ground Call Analysis 4 4 Analysis Results 4.1 Comparison of ground and inventory values for basic forest cover attributes Ratio-of-means comparisons were computed for several VDYP6 (height and age) and VDYP7 (height, age, basal area/ha, trees/ha) forest cover attributes. Since this was a FIPbased inventory, the basal area/ha and trees/ha attributes were generated by VDYP7 (i.e. they were not available as photo-interpreted values). These comparison ratios, and their associated sampling errors, are shown in Table 2 for VDYP6 and in Tables 3, 4 & 5 for VDYP7. Note that although a standard Stage 1 VRI statistical adjustment using VDYP7 would be based on basal area/ha and tree/ha at the 7.5cm+ dbh utilization level (as in Table 4), ratio comparisons for a 12.5cm+ dbh utilization are also provided (Table 5). This sample indicates that the forest cover attributes generally overestimate8 age by about 15%. Overall, height is overestimated by about 7% or about 1.5m, on average9. Larch leading polygons show the highest degree of age and height overestimation, based on this small sample of such polygons. Basal area/ha appears to be generally underestimated by VDYP7. However, poststratification of the sample indicates that this basal area/ha underestimation may be confined primarily to fir or larch leading stands. The sample indicates that trees/ha is, on average, relatively unbiased at the 7.5cm+ dbh utilization level and slightly overestimated at the 12.5cm+ dbh utilization level. The graphical relationships between the ground values and the inventory values of these forest cover attributes are provided in Appendix C. 4.2 Comparison of ground and inventory volume The potential bias associated with VDYP6 and VDYP7 whole stem volume estimation, was also estimated from the sample by computing a ratio of mean ground volume to mean inventory volume. Sampling errors and coefficients of variation (CVs) were reported for these estimated volume ratios-of-means. These results are provided in Table 6. Overall, this sample suggests that VDYP6 whole stem volumes (at 7.5cm+ dbh utilization) are slightly overestimated (by about 2%) for this population. For a VDYP7 inventory, the sample suggests that whole stem volumes are underestimated by about 10% on average. When the sample is post-stratified by inventory leading species, the sample would suggest that both VDYP6 and VDYP7 underestimate volume in fir and larch leading stands but both yield models overestimate volume for other leading species. 8 A ratio larger than 1.0 indicates potential underestimation whereas a ratio less than 1.0 indicates potential overestimation. 9 Slight differences between inventory heights from VDYP6 and VDYP7 occur due to the fact that VDYP6 and VDYP7 use different versions of the site index function SINDEX. February 2008 Cranbrook TSA: IDF and PP Ground Call Analysis 5 Table 2: Ground and VDYP6 height and age means and ratios (with 95% sampling error %), for three alternative strata Mean age (yrs) Mean height (m) VDYP6 Age Run Stratum ratio of means n Ground Inventory n Ground Inventory 1 2 3 all samples FD leading non-FD leading FD leading LW leading non-FD/ non-LW 51 32 19 32 8 11 90 94 84 94 88 81 109 110 107 110 127 92 0.832 ±10.5% 0.858 ±12.0% 0.787 ±21.1% 0.858 ±12.0% 0.688 ±42.4% 0.888 ±17.2% 51 32 19 32 8 11 20.2 19.6 21.2 19.6 21.3 21.2 21.7 21.2 22.5 21.2 24.4 21.1 Table 3: Ground and VDYP7 height and age means and ratios (with 95% sampling error %), for three alternative strata Mean age (yrs) Mean height (m) VDYP7 Age Run Stratum ratio of means n Ground Inventory n Ground Inventory 1 2 3 all samples FD leading non-FD leading FD leading LW leading non-FD/ non-LW 51 32 19 32 8 11 90 94 84 94 88 81 109 110 107 110 127 92 0.832 ±10.5% 0.858 ±12.0% 0.787 ±21.1% 0.858 ±12.0% 0.688 ±42.3% 0.888 ±17.2% 51 32 19 32 8 11 20.2 19.6 21.2 19.6 21.3 21.2 21.6 21.2 22.3 21.2 23.9 21.2 VDYP6 Height ratio of means 0.932 ±5.7% 0.925 ±6.6% 0.943 ±11.2% 0.925 ±6.6% 0.870 ±16.2% 1.004 ±16.7% VDYP7 Height ratio of means 0.935 ±5.7% 0.926 ±6.6% 0.950 ±11.0% 0.926 ±6.6% 0.888 ±15.6% 1.000 ±17.0% Table 4: Ground and VDYP7 basal area/ha and trees/ha at 7.5cm+dbh utilization means and ratios (with 95% sampling error %), for three alternative strata Mean basal area/ha @7.5cm+ (BA) Mean trees/ha @7.5cm+ (TPH) VDYP7 BA VDYP7 TPH Run Stratum ratio of means ratio of means n Ground Inventory n Ground Inventory 1 2 3 all samples FD leading non-FD leading FD leading LW leading non-FD/ non-LW 52 32 20 32 8 12 26.1 26.9 24.7 26.9 34.0 18.4 22.6 22.0 23.4 22.0 25.5 22.1 1.154 ±12.5% 1.223 ±14.9% 1.052 ±23.7% 1.223 ±14.9% 1.333 ±30.2% 0.835 ±33.8% 52 32 20 32 8 12 758 786 712 786 942 559 758 738 792 738 842 758 0.999 ±20.8% 1.065 ±28.3% 0.900 ±31.7% 1.065 ±28.3% 1.119 ±39.9% 0.738 ±56.0% February 2008 Cranbrook TSA: IDF and PP Ground Call Analysis 6 Table 5: Ground and VDYP7 basal area/ha and trees/ha at 12.5cm+dbh utilization means and ratios (with 95% sampling error %), for three alternative strata Mean basal area/ha @12.5cm+ (BA) Mean trees/ha @12.5cm+ (TPH) VDYP7 BA VDYP7 TPH Run Stratum ratio of means ratio of means n Ground Inventory n Ground Inventory 1 2 3 all samples FD leading non-FD leading FD leading LW leading non-FD/ non-LW 52 32 20 32 8 12 24.5 25.3 23.2 25.3 32.3 17.2 20.7 20.0 21.7 20.0 23.8 20.3 1.186 ±12.9% 1.264±14.1% 1.070 ± 26.2% 1.264±14.1% 1.356 ± 33.8% 0.738 ± 56.0% 52 32 20 32 8 12 549 565 524 565 722 393 519 481 578 481 638 539 1.059 ± 18.8% 1.173 ± 23.0% 0.907 ± 33.2% 1.173 ± 23.0% 1.131 ± 41.9% 0.729 ± 57.4% Table 6: Ground, VDYP6 & VDYP7 whole stem volume at 7.5cm+ dbh utilization means and ratios (with 95% sampling error %), for three alternative strata (“volume impact”) Mean vol/ha (WSV @7.5cm+ dbh) VDYP6 WSV @7.5cm+ volume VDYP7 WSV @7.5cm+ volume (inventory volumes based on unadjusted attributes) ratio of means ratio of means Run Stratum n Ground VDYP6 VDYP7 Ratio & SE% CV of ratio Ratio & SE% CV of ratio 1 2 3 all samples FD leading non-FD leading FD leading LW leading non-FD/ non-LW 52 32 20 32 8 12 185.6 184.7 187.1 184.7 249.5 145.5 189.5 172.6 216.5 172.6 229.0 208.2 168.4 153.5 192.3 153.5 196.8 189.3 0.980 ± 13.1% 1.070 ± 15.9% 0.864 ± 22.9% 1.070 ± 15.9% 1.090 ± 32.5% 0.699 ± 29.6% 47% 44% 49% 44% 39% 47% 51% 47% 54% 47% 39% 53% 1.102 ± 14.1% 1.203 ± 17.1% 0.973 ± 25.1% 1.203 ± 17.1% 1.268 ± 32.7% 0.769 ± 33.7% Table 7: Ground & VDYP7 whole stem volume at 12.5cm+ dbh utilization means and ratios (with 95% sampling error %), for three alternative strata (“volume impact”) Mean vol/ha (WSV @12.5cm+ dbh) VDYP7 WSV @12.5cm+ volume ratio of means Run Stratum n Ground Unadjusted Inventory (VDYP7) Ratio & SE % CV of ratio 1 2 3 all samples FD leading non-FD leading FD leading LW leading non-FD/ non-LW leading 52 32 20 32 8 12 177.9 177.0 179.3 177.0 241.2 138.0 160.3 145.7 183.7 145.7 188.9 180.2 1.110 ± 14.5% 1.215 ± 16.7% 0.976 ± 26.9% 1.215 ± 16.7% 1.277 ± 35.2% 0.765 ± 36.0% 52% 46% 57% 46% 42% 57% February 2008 Cranbrook TSA: IDF and PP Ground Call Analysis 7 Note that VDYP7 whole stem volumes were consistently less than VDYP6 volumes, by roughly 10% on average. The consistency of this trend is illustrated in Figure 1 below. VDYP7 WSV @7.5cm+ dbh 400 300 200 100 0 0 100 200 300 400 VDYP6 WSV @7.5cm+ dbh Figure 1: Comparison of VDYP6 and VDYP7 volumes (whole stem at 7.5 cm+ dbh utilization). The diagonal line represents a 1:1 correspondence. 4.3 Coefficients of variation and sample size The sampling error for the overall estimated ratio of ground to VDYP7 whole stem volume at 12.5cm+ dbh utilization is about 14.5% at a 95% confidence level. This is based on the current sample size of 52. Using the CV of the ratio, the estimated sample size required to achieve a desired precision can be calculated. The CV of the ratio R is computed as: where is the variance of the ratio. The sample size, for a given CV and target percentage error (PE) is then computed as: where t is the value from the Student t distribution for a given significance level (α) and degrees of freedom (n-1). Note that since n appears on both sides of the equation (it is required for the t value on the right side of the equation), the result must be found through iteration. February 2008 Cranbrook TSA: IDF and PP Ground Call Analysis 8 The CVs for the volume ratios in Table 7 were used to estimate the sample size required to achieve a 10% and a 15% sampling error for total adjusted volume. These results are shown in Table 8 below. Table 8: Estimated sample sizes to achieve ±10% and ±15% percentage error for total adjusted volume, at a 5% significance level (95% confidence level), based on whole stem volume ratio CVs from the current sample. Run 1 2 3 Stratum all samples FD leading non-FD leading FD leading LW leading non-FD/ non-LW leading Sample size at 5% significance level 10% PE 15% PE n Ratio CV 52 32 20 32 8 1.110 1.215 0.976 1.215 1.277 52% 46% 57% 46% 42% 106 84 127 84 70 49 39 58 39 33 12 0.765 57% 127 58 These sample size estimates are based on whole stem volume ratios. In a Phase 2 VRI ground sampling project implementation plan, sampling error targets are generally specified in terms of volume estimates at a standard utilization (typically 12.5cm+ dbh) net decay, waste & breakage (dwb). It is not known how CVs for a volume ratio net dwb would compare with the whole stem volume ratio CVs shown in Table 7. However it is suspected that the CVs for volumes net dwb would be higher. If the CVs shown in Table 8 are inflated (by adding 5 percentage points), the following sample sizes would result: Table 9: Estimated sample sizes to achieve ±10% and ±15% percentage error for total adjusted volume, at a 5% significance level10, based on inflated whole stem volume ratio CVs from the current sample (inflated CV% = original CV% + 5%). Run 1 2 3 Stratum all samples FD leading non-FD leading FD leading LW leading non-FD/ non-LW leading Sample size at 5% significance level 10% PE 15% PE n Ratio Inflated CV 52 32 20 32 8 1.110 1.215 0.976 1.215 1.277 57% 51% 62% 51% 47% 127 102 150 102 87 58 47 68 47 40 12 0.765 62% 150 68 5 Discussion and Recommendations The forest cover attributes used in this analysis were from a FIP-based inventory. As a result, there were no photo-interpreted basal areas/ha or trees/ha and hence these attributes had to be generated by VDYP7. The analysis of the sample data indicates that 10 Or 95% confidence level February 2008 Cranbrook TSA: IDF and PP Ground Call Analysis 9 inventory ages and, to a lesser degree, inventory heights are generally overestimated in this population. Basal area/ha appears to be generally underestimated by VDYP7. However, stratification by leading species suggests that the basal area underestimation is more closely related to Fd and Lw leading polygons. Although there is little overall average bias associated with the VDYP7-generated trees/ha attribute, the scatterplots for the trees/ha attribute show poor correlation with the ground compiled trees/ha. The ground compiled volume estimates from this sample suggest that VDYP7 underestimates volume in Fd and Lw leading polygons, following the trend observed for basal area/ha. For non-FD and non-Lw leading polygons, VDYP7 volume appears to be overestimated. Note that these results are based on whole stem volumes, since decay information for the ground sample was not available. Once net factoring and a net volume adjustment factor (NVAF) are applied to the ground volumes, as in a VRI Phase 2 sample, the ratio of ground volume to VDYP7 inventory volume may change. Although VDYP6 volume bias trends by leading species followed a pattern similar to that observed for VDYP7 (i.e. volumes underestimated for Fd and Lw leading polygons whereas volumes were overestimated for non-FD and non-Lw polygons), the magnitude of the bias differed considerably. For this sample, the volumes estimated by VDYP6 were consistently larger than the volumes estimated by VDYP7. The overall ratio of mean VDYP6 to mean VDYP7 whole stem volume was 1.12, indicating that on average VDYP6 volume was 12% higher than VDYP7 volume. Volumes net dwb could not be compiled for the ground sample but the two yield models could still be compared at this utilization. Average volumes from VDYP6 were about 16% higher than average volumes from VDYP7 at 12.5cm+ utilization net dwb. In the absence of a VRI Phase 2 statistical adjustment, this analysis suggests that moving from VDYP6 to VDYP7 will lower the overall inventory volume. As part of this analysis, coefficients of variation (CVs) were computed so that potential sample size for VRI Phase 2 ground sampling could be estimated. Since the plot configuration in this project (3 point plots) differed from a standard VRI ground sample (5 point plots), the observed variability reflected by the CV may not be the same as what would be observed in a VRI Phase 2. In addition, the CV in this analysis was based on ratios of whole stem volumes, not ratios of volumes net dwb (VRI Phase 2 project implementation plans typically specify sampling error targets based on volumes net dwb). For the purpose of VRI Phase 2 sample size estimation, it is suggested that the CVs from this sample data be inflated to account for these differences. The most appropriate magnitude of the CV inflation needs to be determined but a minimum increment of 5% is suggested. It is recommended that the level of CV inflation be discussed with MFR Forest Analysis and Inventory Branch (FAIB) staff prior to any final decisions on sample size. Risk and budgetary factors will also come into play in the final decision, however, a minimum sample size of 130 is recommended for this population in order to meet a 10% sampling error target. February 2008 Cranbrook TSA: IDF and PP Ground Call Analysis 10 Key points from the analysis of the sample data in this project are summarized below: VDYP6 volumes are consistently higher than VDYP7 volumes. Volume trends in fir and larch leading polygons differ from volume trends in other species (i.e. non-fir and non-larch) leading polygons. For fir and larch leading polygons both yield models underestimate whole stem volume, on average; for non-fir & non-larch leading polygons both yield models overestimate whole stem volume. Based on attribute bias differences observed among the leading species strata, it is suggested that non-fir and non-larch leading polygons be kept in a separate stratum. Prior to any VRI Phase 2 sample selection, the FIP-based inventory should be “completed” (i.e. the data should have its initial run through VDYP7), since leading species (and hence stratum assignment) may change after “completion”. The CVs estimated from this project may need to be inflated prior to sample size estimation to account for differences in plot design and volume utilization. This should be discussed with FAIB staff. Based on the project CV% plus 5% to account for the differences above, it is estimated that a VRI Phase 2 ground sample of 130 can achieve a 10% sampling error target for total volume at a 95% confidence level. This data should be considered for further exploration in the upcoming VDYP7 sensitivity analysis. February 2008 Cranbrook TSA: IDF and PP Ground Call Analysis 11 6 APPENDIX A Project Background and Ministry Support Cranbrook TSA Ground Call Analysis A.1 Project Overview Ministry of Forests Staff and Licensees in the Rocky Mountain Forest District are concerned that the Forest Cover Inventory volumes in the Interior Douglas Fir and Ponderosa Pine Biogeoclimatic zones of the Cranbrook Timber Supply Area are overestimated. This project will undertake the volume compilation 52 ground samples established in the summer of 2007 and analysis and comparison of field data and Forest Cover attributes, as described in the contract Schedule A. A.2 Project Background Problem statement Licensee and Ministry of Forests personnel in the Rocky Mountain Forest District (RMFD) have expressed concern that forest inventory volumes in the Interior Douglas Fir (IDF) and Ponderosa Pine (PP) Biogeoclimatic zones are overestimated. The issue was raised in the VRI Strategic Inventory Plan (VSIP) meeting that recently took place for the Cranbrook TSA. VRI ground sampling is proposed in the VSIP, and inventory adjustment will correct the inventory volumes that are found during sampling. In order to determine the existence and extent of the problem an Inventory Assessment of the volumes in these zones is proposed. This will allow the more costly phase 2 and NVAF sampling to be more efficiently applied. The assessment will allow the phase 2 sampling to target strata where problems are actually observed, and provide baseline coefficients of variation to help accurately define phase 2 sample sizes, thus reducing phase 2 sampling costs. A preliminary analysis of data collected on a different small project in the Wasa area seems to suggest that overestimation of heights in the inventory may be a potential cause of volume overestimation in these Biogeoclimatic zones. Sample Selection Methodology All forested polygons (>10% crown Closure and > 30 years of age) in the IDF and PP Biogeoclimatic zones within the RMFD were extracted from the provincial forest cover February 2008 Cranbrook TSA: IDF and PP Ground Call Analysis 12 inventory. From this overall population, areas that are in private lands, First Nations Reserves, Parks and TFL’s have been excluded. A 500m buffer was applied to the TRIM roads coverage and this was overlaid with the target population described above. Any polygons that have a portion that falls within the buffer were considered for sampling. Proximity to roads was used primarily to reduce access costs (helicopter will not be required). There will be an unknown bias with this method as some polygons within the larger population will not have a chance of being sampled, but due to extensive road networks in the area, the risk of obtaining results that do not reflect the population will be low. As the results of the project will be used to refine further unbiased sampling (phase 2), and will not be used to adjust the inventory itself, the risks posed are minimal. Samples were selected using a process similar to the VRI Probability Proportional to Size With Replacement (PPSWR) process. The total area of all polygons in the resultant database was totalled and random numbers were generated between 0 and the total area of all polygons. A polygon was selected if a generated random number was larger than the accumulated total area corresponding to the polygon immediately preceding it, and the random number was smaller than or equal to its accumulated area. Initially the Cranbrook TSA portion of the district was sampled. A total of 52 samples were established in the Cranbrook TSA. In order to provide an unbiased sample location within the polygon, a random coordinate start point within the buffer zone was selected at the sample selection stage. This may result in high variability between the photo interpreted inventory and the ground data on a polygon basis, but when summarized for the project will produce an unbiased and statistically valid result. Sample Establishment Methodology The field sample design consisted of 3 point variable radius plots in each sample location where the following data was collected: Sample, plot number Basal Area Factor Plot Type (Full/Half/Quarter) Tree number Species Beetle codes Diameter (on live trees only) Heights (on live trees only) Random Bearing 1 Representative age of the leading and second species from the cluster The field crew navigated to the coordinate given using GPS and establish the 3 point transect on a random bearing determined from the VRI random number table. February 2008 Cranbrook TSA: IDF and PP Ground Call Analysis A.3 13 Ministry Support The Ministry will provide ground call data, Forest Cover population data, and forest cover data for the established samples in a MS Access database. A description of the fields and tables included in the access database follows. IDFPP3 is the ground call database. Below are the column headings. “Sample” is the sample number. It will correspond with the "sample number" column in the "sample list" database. “BRG” is the bearing from the first plot in the sample to the next plot “PLT” is the plot number. Note that there may not always be three plots. “FHQ” is full, half or quarter plot. Quite often only one tree in each plot is labelled with a F, H, or Q. The value applies to all trees in the plot. In plots where none of the trees has been labelled as F, H, or Q, it is to be considered a full plot. “BAF” is the basal area factor. Again there may be only one tree in each plot that has a BAF. “Bands” shows the number of relascope bands used in the plot. The number of bands has been converted into a BAF where a relascope was used, so the Bands column can be ignored. “Tree” is the tree number. If the value here is "0" then there was no trees in the plot, so make sure these trees are not included in the basal area calc.. There is one tree 99 in the database, which would also not be included in the basal area. “DBH”, and Height contain the diameter at breast height and .tree height respectively. “Age” is the age at breast height of the representative tree. The presence of an age can be used as a flag for a representative tree. For almost all samples there will be a leading and second species, but occasionally there may only be a leading. “LEN” is the length in CM for age prorate. There is only one prorate length in the database, the age is the counted age only, and does not take into account the prorate, so an age corrected for breast height and prorate where applicable needs to be calculated. “Beetle” is the MPB codes. The “Sample List” table is the PPSWR sample list that was selected for the project. The "sample number" is the last column. This sample list includes a lot of extras, and covers the entire Rocky Mountain Forest District. Only samples in the Cranbrook TSA portion have been established. The "orgunit_cd" field can be used to tell which polygons are in which TSA (DIN is Invermere, and DCB is Cranbrook). This field actually is the old Forest District codes, but in the case of Cranbrook it corresponds with the TSA boundary as well. The first 52 samples in Cranbrook that could be accessed were established, so there will be some skipped samples. The veg_Clip_cc10_age30 table is the entire population for the project, again the Cranbrook TSA portion will have to be extracted. February 2008 Cranbrook TSA: IDF and PP Ground Call Analysis 14 7 APPENDIX B Ground and forest cover attribute ratios, for biogeoclimatic zone stratification Table B-1: Ground and VDYP7 height and age means and ratios, overall and for biogeoclimatic zone strata Mean age (yrs) Mean height (m) VDYP7 Age Run Stratum ratio of means n Ground Inventory n Ground Inventory 1 4 all samples IDF PP 51 40 11 90 90 92 109 110 103 0.832 ±10.5% 0.816 ± 11.8% 0.894 ± 25.8% 51 40 11 20.2 20.0 20.8 21.6 21.6 21.7 VDYP7 Height ratio of means 0.935 ±5.7% 0.929 ± 6.5% 0.958 ± 13.9% Table B-2: Ground and VDYP7 basal area/ha and trees/ha at 7.5cm+dbh utilization means and ratios, overall and for biogeoclimatic zone strata Mean basal area/ha @7.5cm+ (BA) Mean trees/ha @7.5cm+ (TPH) VDYP7 BA VDYP7 TPH Run Stratum ratio of means ratio of means n Ground Inventory n Ground Inventory 1 4 all samples IDF PP 52 41 11 26.1 25.5 28.1 22.6 22.3 23.5 1.154 ±12.5% 1.143 ± 14.5% 1.196 ± 29.3% 52 41 11 758 755 767 758 751 789 0.999 ±20.8% 1.006 ± 23.2% 0.972 ± 55.5% Table B-3: Ground & VDYP7 whole stem volume at 12.5cm+ dbh utilization means and ratios, overall and for biogeoclimatic zone strata Mean vol/ha (WSV @12.5cm+ dbh) VDYP7 WSV @12.5cm+ volume impact ratio of means Run Stratum Unadjusted Ratio & sampling error n Ground CV of ratio Inventory (VDYP7) % at 95% confidence 1 2 all samples IDF PP 52 41 11 177.9 173.1 195.7 160.3 160.2 160.6 1.110 ± 14.5% 1.080 ± 17.5% 1.218 ± 27.0% 52% 56% 40% February 2008 Cranbrook TSA: IDF and PP Ground Call Analysis 15 8 APPENDIX C Graphs of ground vs. VDYP7 forest cover attributes (diagonal line represents perfect correlation). 200 100 0 0 100 200 200 100 0 300 300 Ground age (yrs 300 Ground age (yrs Ground age (yrs 300 0 100 VDYP7 inventory age (yrs) 100 0 300 Fig C-2: Lw leading age relationship 30 30 20 15 10 5 Ground height (m) 30 Ground height (m) 35 25 25 20 15 10 10 15 20 25 30 VDYP7 inventory height (m) Fig C-4: Fd leading height relationship. 35 0 300 25 20 15 10 5 5 5 200 Fig C-3: Non-Fd, non-Lw leading age relationship 35 0 100 VDYP7 inventory age (yrs) 35 0 0 VDYP7 inventory age (yrs) Fig C-1: Fd leading age relationship. Ground height (m) 200 200 0 5 10 15 20 25 30 VDYP7 inventory height (m) Fig C-5: Lw leading height relationship 35 0 0 5 10 15 20 25 30 35 VDYP7 inventory height (m) Fig C-6: Non-Fd, non-Lw leading height relationship February 2008 40 20 0 20 40 60 VDYP7 basal area/ha @7.5cm+ dbh 0 0 20 40 60 Ground trees/ha @7.5cm+ dbh 1,500 1,000 500 0 500 1,000 1,500 2,000 VDYP7 trees/ha @7.5cm+ dbh Fig C-10: Fd leading trees/ha at 7.5cm+ dbh relationship. 2,500 20 0 0 20 40 60 2,500 2,000 1,500 1,000 500 0 40 Fig C-9: Non-Fd, non-Lw leading basal area/ha at 7.5cm+ dbh relationship 2,500 2,000 60 VDYP7 basal area/ha @7.5cm+ dbh Fig C-8: Lw leading basal area/ha at 7.5cm+ dbh relationship 2,500 Ground trees/ha @7.5cm+ dbh 20 VDYP7 basal area/ha @7.5cm+ dbh Fig C-7: Fd leading basal area/ha at 7.5cm+ dbh relationship. 0 40 Ground trees/ha @7.5cm+ dbh 0 60 Ground basal area/ha @7.5cm+ dbh 60 16 Ground basal area/ha @7.5cm+ dbh Ground basal area/ha @7.5cm+ dbh Cranbrook TSA: IDF and PP Ground Call Analysis 0 500 1,000 1,500 2,000 VDYP7 trees/ha @7.5cm+ dbh Fig C-11: Lw leading trees/ha at 7.5cm+ dbh relationship 2,500 2,000 1,500 1,000 500 0 0 500 1,000 1,500 2,000 2,500 VDYP7 trees/ha @7.5cm+ dbh Fig C-12: Non-Fd, non-Lw leading trees/ha at 7.5cm+ dbh relationship February 2008 Cranbrook TSA: IDF and PP Ground Call Analysis 17 400 300 200 100 0 0 100 200 300 400 500 VDYP7 WSV @12.5cm+ dbh Fig C-13: Fd leading whole stem volume at 12.5cm+ dbh relationship. 500 Ground WSV @12.5cm+ dbh 500 Ground WSV @12.5cm+ dbh Ground WSV @12.5cm+ dbh 500 400 300 200 100 0 0 100 200 300 400 500 VDYP7 WSV @12.5cm+ dbh Fig C-14: Lw leading whole stem volume at 12.5cm+ dbh relationship 400 300 200 100 0 0 100 200 300 400 500 VDYP7 WSV @12.5cm+ dbh Fig C-15: Non-Fd, non-Lw leading whole stem volume at 12.5cm+ dbh relationship February 2008 Cranbrook TSA: IDF and PP Ground Call Analysis 18 February 2008