Forest biomass carbon inventory for the Russas and Valparaiso Properties, Acre State, Brazil

Transcription

1 Forest biomass carbon inventory for the Russas and Valparaiso Properties, Acre State, Brazil Prepared for CarbonCo by TerraCarbon CarbonCo 3 Bethesda Metro Center, Suite 700 Bethesda, MD U.S.A. TerraCarbon LLC 5901 N. Sheridan Rd. Peoria, Illinois U.S.A. Author: James M. Eaton, TerraCarbon Igor Agapejev de Andrade, TECMAN

2 Acknowledgements We would like to thank the field crew including: Francisco Bezerra Fernandes, Manoel Valmir Pereira da Silva, Rizomar Lopes de Araújo, and Valdimar de Souza Azevedo for all their hard work and long days in the forest. We also appreciate the efforts of Fabio Thaines of TECMAN for his assistance in managing the forest inventory. Further, we thank Pedro Freitas for his help in all facets of this forest inventory including help with translation. Thanks also to Ilderlei Souza Rodrigues Cordeiro and Manoel Batista Lopes for their assistance in Cruzeiro do Sul and for hosting inventory personnel. Finally, we are grateful to CarbonCo for providing us the opportunity to help lead this forest inventory. TerraCarbon, Version 1.1 / February 19,

3 Table of Contents 1.0 Introduction Project property Inventory Design Objectives and approach Field measurement protocols Quality control and data archiving Field Implementation List of forest inventory plots Data analysis Results and Discussion Dead wood density results Descriptive statistics and precision Validation of allometric equations Appendices Appendix A. Standard Operating Procedures TerraCarbon, Version 1.1 / February 19,

4 1.0 Introduction The purpose of this forest inventory was to produce an estimate of forest biomass carbon stocks in the year 2013, including live aboveground, belowground biomass, and dead wood in the project areas, located in the municipalities of Cruzeiro do Sul and Porto Walter in Acre, Brazil. This inventory is being developed under the Verified Carbon Standard (VCS), VCS guidance on Agriculture, Forestry and Other Land Use Requirements, and the modular AD Partners REDD methodology VM Project property The inventoried area covers 72,032 ha in Acre, Brazil. The area is located 40 km south of the town Cruzeiro do Sul along the Juruá River. The forest inventory area (Figure 2.1a) is composed of two adjacent properties, Seringal Russas and Seringal Valparaiso, separated by the Valparaiso River. Seringal Russas is south of the Valparaiso river while Seringal Valparaiso is north of the river. The northeast parcel, at the end of the secondary road, Ramal 3, is part of Seringal Valparaiso. Figure 2.1b is a map of the finalized project boundaries. Figure 2.1a. Map of Seringal Russas and Seringal Valparaiso (Google Earth, 2013). TerraCarbon, Version 1.1 / February 19,

5 Figure 2.1b. Map of the Russas and Valparaiso Project Areas. 3.0 Inventory Design 3.1 Objectives and approach Precision target The forest biomass inventory was designed to produce a stock estimate with a precision level not exceeding +/-15% of the mean with 95% confidence to meet the requirements of both the VCS and VM Carbon pools and forest components The inventory targeted live aboveground biomass, belowground biomass, standing dead wood, and lying dead wood. Bamboo and lianas were not measured and conservatively excluded from estimation of biomass stocks. Further, trees in the Cecropia genus have been conservatively excluded TerraCarbon, Version 1.1 / February 19,

6 from the forest inventory because of lack of verifiable allometric equations. The minimum diameter at breast height (dbh) for all live trees and minimum diameter for dead trees included in the inventory was 10 cm. In addition to collecting diameter data for live trees, the stem heights (i.e. height to the base of the crown) of three dominant trees in each plot were measured Definition and delineation of strata Stratification of the project area reduces overall variability and improves sampling efficiency. The project area was stratified using a vegetation map from the Acre State 1 publication Ecological and Economical Zoning where land cover is classified using the Brazilian Forest Classification System 2. Stratification resulted in five strata as listed in Table 3.1 Table 3.1. Five strata present in the Russas/Valparaiso Forest Inventory. Stratum Stratum Description Area (hectares) FAP Open palm forest 16,211 FAB + FAP Open forest with bamboo and palm 11,779 FAP - Alluvial Open alluvial forest with palm 6,741 FAP + FAB + FD and FAP + FD + FAB FAP + FD and FD + FAP Open forest with bamboo and palm and 18,130 dense forest Dense forest and open palm forest 19,171 1 Acre State, Zoneamento Ecológico-Econômico do Estado do Acre Fase II Documentos Síntese. Rio Branco, Acre. 2 Veloso, H.P., Rangel FO, A.L.R., Lima, J.C.A., Classificação da vegetação brasileira, adaptada a um Sistema Universal. IBGE, Rio de Janeiro. TerraCarbon, Version 1.1 / February 19,

7 Figure 3.1. Stratification of the inventoried area Sampling design The project employs stratified random sampling using clusters of five 23-meter radius circular (fixed area) plots for live aboveground and belowground biomass and standing dead wood. Lying dead wood was sampled using line intersect sampling (with two transects 92m +92m per cluster). The inventory employs cluster sampling to reduce variability and achieve greater precision. Each cluster of five 23-meter radius plots is arranged in a cross configuration oriented in the cardinal directions (Figure 3.2). The center plot of each cluster is permanently marked, while the four surrounding satellite plots centers are un-marked. TerraCarbon, Version 1.1 / February 19,

8 Figure 3.2. Diagram of cluster monitoring plot design with the permanent sample plot in the middle. North 92m 92m Center point of cluster Calculations of the required sample size referenced estimates of biomass carbon stocks and coefficients of variation (CV s) from a 2011 forest inventory carried out nearby in Acre, Brazil 3. Based on the analysis of the reference data available, a sample of 25 clusters should be sampled to target a precision of +/-15% of the mean with 95% confidence. 3 J. Eaton and I. Agapejev de Andrade Forest biomass carbon inventory for the Purus REDD Project, Acre State, Brazil. TerraCarbon, Version 1.1 / February 19,

9 Table 3.2. Estimates and parameters for estimating sample size (sample unit = a cluster of five 23 m radius plots). Parameters for estimating sampling size FAP FAB + FAP FAP - Alluvial FAP + FAB + FD and FAP + FD + FAB FAP + FD and FD + FAP Area (hectares) 16,211 11,779 6,741 18,130 19,171 Mean live tree biomass (t/ha) Weighted mean Estimated variance Estimated standard deviation Estimated coefficient of variation 30% 30% 28% 30% 30% Calculated Sample size FINAL SAMPLE SIZE Sample units were allocated at random within each stratum using ArcGIS. GPS coordinates of each monitoring plot correspond to center point of the cluster. Details of each individual cluster sampled are located in Section Field measurement protocols Direct field measurements of live aboveground and belowground biomass, standing dead wood, and lying dead wood followed field protocols detailed in the Standard Operating Procedures (See Appendix A). 3.3 Quality control and data archiving QA/QC procedures outlined here served to minimize errors in measurement and data recording Field measurements Field crews were fully trained in all aspects of field data collection and adhered to field measurement protocols. Field crew leaders were responsible for ensuring that field protocols were followed to ensure accurate and consistent measurement. Pilot sample plots were measured before the initiation of formal measurements to appraise field crews and identify and correct any errors in field measurements. To ensure accurate measurements, the height of diameter at breast height (1.3 m) was periodically re-assessed by personnel during the course of the inventory Data entry TerraCarbon, Version 1.1 / February 19,

10 Data was recorded on field datasheets and then transcribed to electronic media. To minimize errors in data entry, personnel involved in data entry were consulted by personnel involved in measurement to clarify anomalous values and ambiguities. A subset of the field sheets were checked to ensure that data transcribed to electronic media were consistent with data on the field datasheets. Database searches were made following data entry to identify and correct anomalous values. 4.0 Field Implementation Field measurements for the forest inventory were carried out in February and March All fieldwork was completed by TECMAN employees, including Francisco Bezerra Fernandes, Manoel Valmir Pereira da Silva, Rizomar Lopes de Araújo, and Valdimar de Souza Azevedo, under the leadership of MSc forester Igor Agapejev de Andrade and Fabio Thaines with ongoing guidance from James Eaton. 5.0 List of forest inventory plots A total of 25 clusters 4, all listed in Table 5.1, were sampled as part of the forest biomass inventory. One of the original sampling plots 101-C could not be reached due to logistical challenges in the field and was not completed, however the remaining sample points are well-distributed (Figure 5.1). Coordinates represent the center point of the center plot. Table 5.1. GPS coordinates and strata of cluster centers (i.e. permanent sample plot centers). Cluster Plot Strata Easting (UTM18S;SAD69) Northing (UTM18S;SAD69) 102 C FAP C FAP C FAP C FAP C FAB + FAP C FAB + FAP C FAB + FAP C FAB + FAP C FAB + FAP C FAP + FD and FD + FAP C FAP + FD and FD + FAP C FAP + FD and FD + FAP C FAP + FD and FD + FAP C FAP + FD and FD + FAP C FAP + FD and FD + FAP C FAP + FD and FD + FAP C FAP - Alluvial One of the original sampling plots 101-C could not be reached as the area was to wet to safely sample. TerraCarbon, Version 1.1 / February 19,

11 402 C FAP - Alluvial C FAP - Alluvial C FAP + FAB + FD and FAP + FD + FAB C FAP + FAB + FD and FAP + FD + FAB C FAP + FAB + FD and FAP + FD + FAB C FAP + FAB + FD and FAP + FD + FAB C FAP + FAB + FD and FAP + FD + FAB C FAP + FAB + FD and FAP + FD + FAB TerraCarbon, Version 1.1 / February 19,

12 Figure 5.1. Location of field monitoring plots within Seringal Russas and Seringal Valparaiso. TerraCarbon, Version 1.1 / February 19,

13 6.0 Data analysis For live trees, biomass was calculated as a function of diameter at breast height (DBH; in cm) using the predictive model developed by Brown 5 for tropical moist forest stands, below (n=170 trees, R- squared=0.84). Application of the moist equation reflects the annual precipitation for the inventoried area, 2230 mm. aboveground biomass (kg) = (( *(DBH)+1.242*(DBH)^2)) For palms, we used height and dbh (a conservative estimate of basal diameter) measurements to estimate the aboveground volume of a paraboloid and then applied mean (species level) Amazonian palm specific gravity of 0.31 g/cm 3 estimated by Baker et al 6. The estimate of biomass was therefore limited to the main trunk (bole) of the palm. Thus, for palms aboveground biomass (Mg) = 0.5*Π*(basal diameter(cm)/200) 2 *height(m)*0.31 Root biomass density was estimated at the cluster level applying the equation developed by Cairns et al. 7 (1997; R-squared=0.83), where Root Biomass Density (t/ha) = EXP ( LN(aboveground biomass density)) The volume of lying dead wood per unit area was estimated using the equation (Warren and Olsen ) as modified by Van Wagner 8 separately for each dead wood density class: V LDW where: N 2 * D n 1 8* L 2 n V LDW Volume of lying dead wood per unit area; m 3 ha -1 D n Diameter of piece n of dead wood along the transect; cm 5 Brown, S., Estimating biomass and biomass change of tropical forests: A primer. FAO Forestry Paper: vii, 55 p. 6 Baker TR., Phillips OL., Malhi Y., Almeida S., Arroyo L., Di Fiore A., Killeen TJ., Laurance SG., Laurance, WF., Lewis SL., Lloyd J., Monteagudo A., Neill DA., Patiño S., Pitman NC., Silva JN., Vásquez Martínez R Variation in wood density determines spatial patterns in Amazonian forest biomass. Global Change Biology 10: Cairns, M. A., S. Brown, E. H. Helmer, and G. A. Baumgardner Root biomass allocation in the world s upland forests. Oecologia 111, Van Wagner, C.E. (1968). The line intersect method in forest fuel sampling. Forest Science 14: TerraCarbon, Version 1.1 / February 19,

14 N L Total number of wood pieces intersecting the transect; dimensionless Length of the transect; m Length of transect was corrected for slope. The volumes per unit area of each dead wood density class were then summed by class within each plot and multiplied by their respective densities (see below) to convert to a mass per unit area for each plot. Biomass of standing dead wood in the decomposition class 1 is estimated using the allometric equation for live trees. In decomposition class 2, the estimate of biomass was limited to the main trunk (bole) of the tree, converting volume of the bole to biomass using dead wood density classes. Volume of bole was estimated as the volume of a cone, as specified in the VM0007 module, Estimation of carbon stocks in the dead wood pool. Density of dead wood was determined through sampling and laboratory analysis. Discs were collected in the field and volume determined as per standard protocols (see Appendix A for more details). Discs were then oven-dried to constant weight, and the resulting dry weight was recorded and used to calculate dead wood density as oven-dry weight (g) / green volume (cm 3 ) for each sample. Laboratory work was performed by the Centro de Tecnologia da Madeira e Mobiliário 9 in Rio Branco, Acre. Dry mass was converted to carbon using the default carbon fraction of 0.47 t C/t d.m. (as recommended by IPCC 10 Guidelines for National Greenhouse Gas Inventories). 9 This translates as the Center of technology of Wood and Furniture". 10 IPCC 2006 Guidelines for National Greenhouse Gas Inventories. Chapter 4 AFOLU (Agriculture, Forestry and Other Land-use). TerraCarbon, Version 1.1 / February 19,

15 7.0 Results and Discussion 7.1 Dead wood density results The dead wood density estimates (in g/cm3) are detailed in Table 7.1. Table 7.1. Results of dead wood density measurements, expressed in g/cm3. Dead wood Sample Number density (g/cm3) Rotten (P) Intermediary (I) Solid (S) Mean Variance Standard Deviation Coefficient of Variation (%) Standard Error % Confidence Interval 95% Confidence Interval Number of Samples 10% of the mean Values used in the calculation of dead wood stocks are as follows: The value for the rotten wood density class is g/cm3. This value represents the mean for this class, as the 90% confidence interval is equal to or less than 10% of the mean TerraCarbon, Version 1.1 / February 19,

16 The value for the intermediate wood density class is g/cm3. This value represents the mean for this class, as the 90% confidence interval is equal to or less than 10% of the mean The value for solid wood density class is g/cm3. This value represents the mean for this class, as the 90% confidence interval is equal to or less than 10% of the mean. 7.2 Descriptive statistics and precision Detailed biomass stocks by pool for each cluster are provided in Table 7.2. Descriptive statistics are then presented on the carbon stocks (t C/ha) of live tree (aboveground and belowground) biomass, standing dead wood, and lying deadwood (Table ) for each forest strata. On average, live aboveground and belowground tree biomass, standing dead wood, and lying deadwood make up 96%, 1% and 3% of total forest biomass, respectively. Table 7.2 Cluster level estimates of biomass stocks. Cluster ID Live aboveground tree biomass (t/ha) Belowground biomass (t/ha)) Standing dead wood (t/ha) Lying dead wood (t/ha) TerraCarbon, Version 1.1 / February 19,

17 Table 7.3. Estimates of live tree biomass carbon stock by strata. Descriptive Statistic FAB + FAP (tons C/ha) FAP (tons C/ha) FAP - Alluvial (tons C/ha) FAP + FAB + FD or FAP + FD + FAB (tons C/ha) FAP + FD or FD + FAP (tons C/ha) Mean (t C/ha) Standard Error (t C/ha) Variance Standard Deviation Coefficient of Variation (%) Number of Samples Table 7.4. Estimates of standing deadwood biomass carbon stock by strata. Descriptive Statistic FAB + FAP (tons C/ha) FAP (tons C/ha) FAP - Alluvial (tons C/ha) FAP + FAB + FD or FAP + FD + FAB (tons C/ha) FAP + FD or FD + FAP (tons C/ha) Mean (t C/ha) Standard Error (t C/ha) Variance Standard Deviation Coefficient of Variation (%) Number of Samples Table 7.5. Estimates of lying deadwood biomass carbon stock by strata. Descriptive Statistic FAB + FAP (tons C/ha) FAP (tons C/ha) FAP - Alluvial (tons C/ha) FAP + FAB + FD or FAP + FD + FAB (tons C/ha) FAP + FD or FD + FAP (tons C/ha) Mean (t C/ha) Standard Error (t C/ha) Variance Standard Deviation TerraCarbon, Version 1.1 / February 19,

18 Coefficient of Variation (%) Number of Samples Area (ha) 11,779 16,211 6,741 18,130 19,171 Area weighted averages and uncertainty estimates for each strata are listed in Table 7.6, below. Table 7.6. Mean CO2 stock estimates and uncertainty for individual forest strata. Strata FAP FAB + FAP FAP + FD or FD + FAP FAP - Alluvial FAP + FAB + FD or FAP + FD + FAB Mean (t CO2/ha) EBSL,SS,t,I 7,426,595 5,204,885 7,467,390 2,497,727 8,741,282 (tco2e) UncertaintyBSL,SS,I (%) 15.3% 34.1% 11.7% 63.9% 14.4% Uncertainty of the forest biomass estimates are calculated using equation 4 and equation 5 of the X- UNC module of the VM0007 methodology. Overall, the inventory produced an estimate of CO2 stocks at the project level of t CO2/ha with a precision level of +/- 9.7% of the mean at the 95% confidence level (Table 7.7). The forest inventory thus meets the precision requirements of the methodology (+/- 15% of the mean at a 95% confidence level). Table 7.7. Project level statistics for total CO2 stocks in the 2013 forest inventory employing stratified random sampling. Descriptive Statistic Total CO2 stock Mean (t CO2/ha) UncertaintyBSL,SS (%) 9.7% 7.3 Validation of allometric equations Short of cost-intensive destructive sampling, accuracy of allometric equation results was assessed through a quasi-validation. Estimates of live aboveground biomass calculated using the allometric equation (with dbh as the sole independent variable) were compared with estimates of aboveground biomass (tons) calculated as follows, which served as the observed value: Live aboveground biomass = Stem Volume * Wood Density * BEF Where, TerraCarbon, Version 1.1 / February 19,

19 Stem volume (m3) = EXP(( )+( *(LN(DBH(cm))))+( *(LN(Commercial height(m))))) 11 DBH measured in the field Commercial height measured in the field Wood Density (species-specific) = in g/cm 3 ; Chave et al database of wood density for species naturally occurring in Central and South America. Trees were identified and recorded in the field by local foresters familiar with the composition of the project area forest. Where species-specific wood density was not available, average wood density for the genus was applied; Chave et al Where the tree was unidentified, a wood density of g/cm 3 was applied, corresponding to the species level mean for the east Amazon region; Chave et al 2006 Biomass Expansion Factor (BEF) = 1.38 for trees 20-40cm, 1.33 for trees 40-80cm, and 1.25 for trees 80cm, as stipulated in REDD Modules Methodology, VM0007. The Brown (1997) 13 allometric equation meets the requirements of methodology VM0007, whereby trees are distributed both above and below the curve and less than 75% of the predicted biomass estimates are above the observed values. 11 Commercial volume equation used in Acre, Brazil (pers comm.. Igor Agapejev, TECMAN, Rio Branco, Acre, Brazil). Thaines, F., E. Muñoz Braz, P. Povoa de Mattos, A.A. Ribeiro Thaines Equações para estimativa de volume de madeira para a região da bacia do Rio Ituxi, Lábrea, AM (Equations for estimating timber volume in the region of the River Basin of Ituxi, Lábrea, Amazon, Brazil). Pesquisa Florestal Brasileira, Volume 30; Issue J. Chave, H. Muller-Landau, T. Baker, T. Easdale, H. ter Steege, CO Webb Regional and phylogenetic variation of wood density across 2,456 neotropical tree species. Ecological Applications 16, Brown, S., Estimating biomass and biomass change of tropical forests: A primer. FAO Forestry Paper: vii, 55 p. TerraCarbon, Version 1.1 / February 19,

20 Live aboveground biomass (tons)- Observed Russas and Valparaiso Forest Inventory Figure 7.1. Estimates of live aboveground biomass: observed versus predicted (Brown 1997). Solid line = 1:1 line Live aboveground biomass (tons)- Predicted TerraCarbon, Version 1.1 / February 19,

21 Appendices Appendix A. Standard Operating Procedures Monitoring Forest Carbon in Acre, Brazil Delivered to CarbonCo Authors / James Eaton and David Shoch Date / November 2012 Version / 1.0 TerraCarbon, Version 1.1 / February 19,

22 1.0 Objective and monitoring approach 1.1 Objective The inventory objective is to produce an estimate of forest biomass carbon stocks per unit area with precision of +/-15% of the mean with 95% confidence. 1.2 Carbon pools The inventory will sample and/or estimate forest carbon stocks in the following pools: Aboveground live tree biomass (including palms) Belowground live tree biomass Standing dead wood Lying dead wood 1.3 Sampling design The inventory employs stratified random sampling with clusters of five 23m radius circular plots 0.83 ha (configuration of the cluster is detailed below). TerraCarbon, Version 1.1 / February 19,

23 1.4 Cluster configuration The project employs cluster sampling, using the configuration below: North 92m 92m Center point of cluster Map coordinates of sample points correspond with center point of the cluster above. TerraCarbon, Version 1.1 / February 19,

24 2.0 Standard operating procedures 2.1 Marking cluster center Once a cluster center location is reached, the center point will be marked with a stake securely planted in the ground, to which an aluminum tag is attached. The tag is labeled with the cluster number. UTM coordinates of the cluster center will be recorded, if altered from the prescribed location. 2.2 Measuring and recording slope The slope (in %) of each plot will be measured with a clinometer. The slope will be recorded so the plot dimensions can later be adjusted to calculate the equivalent horizontal area. 2.3 Measurement of live trees Within each plot all stems > 10 cm dbh will be measured and species recorded. Diameter of all trees will be measured at breast height (1.3 m above ground level, see Figure 2.1). Diameter of trees with buttresses will be measured directly above the point of termination of the buttress (Figure 2.2). Species (or genera or common name) will also be recorded. In each plot, height to the base of the crown of the three tallest trees will be measured with a clinometer. Where palms are encountered that meet the minimum dbh threshold, two measurements will be taken: basal diameter and height to the top of the stem. TerraCarbon, Version 1.1 / February 19,

25 Figure 2.1. Point of measurement of diameter at breast height (from Pancel 14, 1993). 14 Pancel, L., ed Tropical forestry handbook. Berlin, Germany, Springer-Verlag. Volume 1, 738 pp. TerraCarbon, Version 1.1 / February 19,

26 Figure 2.2. Point of measurement for diameter at breast height of a buttress tree. Point for measurement for DBH TerraCarbon, Version 1.1 / February 19,

27 2.4 Boundary Issues In the event that a plot overlaps the project property boundary or strata boundary, the plot will be corrected using the mirage method 15 (Figure 2.3). The solid-lined circle is the actual plot border. The portion of the circle above the horizontal line is outside of the forest strata being sampled. After sampling all the trees within the plot within the forest strata (e.g. below the line), the trees within the grey shaded area will then be registered twice on the data sheet to account for the same area which is above the horizontal line and outside the plot. Figure 2.3. Diagram of mirage method (Avery and Burkhart, 1994). outside inside Where the 92 meter lines of transit from the cluster center cross the project or strata boundary prior to terminating, lines will be deflected from the boundary back into the project area using a ricochet method to complete the 92 m, where the line of transit will ricochet back into the project area to the right of the original bearing at a 45 degree angle. 15 Avery, T.E. and H.E. Burkhart Forest Measurements. Fourth Edition. McGraw Hill, Boston, Massachusetts, USA. 408 pp. TerraCarbon, Version 1.1 / February 19,

28 Figure 2.4. Diagram of boundary reflection method. Strata boundary Primary point of cluster X m 92 m X m 45º 2.5 Dead wood Dead wood measurements will be restricted to pieces of dead wood with a diameter > 10 cm Measurement of standing dead wood Standing dead trees will be measured using the same plots used for live trees. The decomposition class (not to be confused with dead wood density class) of the dead tree shall be recorded and the standing dead wood is categorized under two decomposition classes: 1) Tree with branches and twigs that resembles a live tree (except for leaves); 2) Tree with signs of decomposition (other than loss of leaves) including loss of twigs, branches, or crown. For decomposition class 1, diameter at breast height is measured and recorded as per protocols for live trees. For decomposition class 2, the following measurements/assignations are taken: dead wood density class (sound, intermediate or rotten) basal diameter TerraCarbon, Version 1.1 / February 19,

29 height to the base of the crown Measurement of lying dead wood Lying dead wood will be sampled using the line intersect method using the two 92-meter lines forming two axes of the cluster. Where exceeding 15%, the slope (in %) of each line will be recorded with a clinometer. Along the lines, the diameters of all lying dead wood 10 cm diameter intersecting the lines are measured at the point of intersection. A piece of lying dead wood should only be measured if (a) more than 50% of the log is aboveground and (b) the sampling line crosses through at least 50% of the diameter of the piece (where it intersects the end of a piece). Each piece of dead wood measured is also assigned to one of three dead wood density classes (sound, intermediate or rotten) using the machete test. 2.6 Determining the density of dead wood During the field inventory, a representative sample of dead wood should be collected to determine the average density for each density class. Thirty samples of dead wood should be collected for each density class, giving you a total of 90 samples. Cut a full disc of the selected piece of dead wood using a chain saw or a hand saw. Green volume (cm 3 ) is determined in the laboratory using a water displacement method standardized by ABNT 16 or the Brazilian Association of technical rules. Volume is determined by first saturating the sample in water for up to three days until a constant weight is reached. Next a beaker is placed on a balance and partially filled with distilled water. The sample is submerged in the beaker by pressing down with a needle/wire, the level of water therefore rises, and the reading on the balance is as if one has added the amount of water equivalent to the volume of the sample. Therefore the reading on the balance is equal to the volume of the sample (with the equivalence 1 g= 1 cm 3 ). This technique thus fills all the voids in the sample and gives the true volume of the sample. The disc is then dried in an oven ( C) in the laboratory to constant weight (g). Density is calculated as dry mass (g) divided by volume (cm 3 ). 16 The norm number is ABNT NBR ISO TerraCarbon, Version 1.1 / February 19,

30 3.0 Quality control Implementation of the monitoring plan will apply QA/QC procedures as outlined here to minimize errors in measurement and data recording. This section covers procedures for: (1) collecting reliable field measurements and (2) documenting data entry. 3.1 Field measurements Field crews will be fully trained in all aspects of the field data collection and adhere to field measurement protocols. Field crew leaders will be responsible for ensuring that field protocols are followed to ensure accurate and consistent measurement. Pilot sample plots shall be measured before the initiation of formal measurements to appraise field crews and identify and correct any errors in field measurements. To ensure accurate measurements, the height of diameter at breast height (1.3 m) will be periodically re-assessed by personnel during the course of the inventory. Field crews will have maps for use in the field to precisely interpret property and strata boundaries. 3.2 Data entry Data will be recorded on field sheets and then transcribed to electronic media. To minimize errors in data entry, where they are not the same, personnel involved in data entry and analysis will consult with personnel involved in measurement to clarify any anomalous values or ambiguities in transcription. A subset of the field sheets will be checked to ensure that data transcribed to electronic media is consistent with data on the field sheets. Database searches will be made following data entry to identify any anomalous values that require clarification or correction. TerraCarbon, Version 1.1 / February 19,