BFI Analysis Document Final

See discussions, stats, and author profiles for this publication at: https://www.researchgate.
net/publication/353946805
Estimation Procedures of Indicators and Variables of the Bangladesh Forest

Inventory
Technical Report · June 2019
CITATIONS READS
0 57
11 authors, including:
Md. Akhter Hossain Nikhil Chakma

University of Chittagong Rangamati Science and Technology University
63 PUBLICATIONS 380 CITATIONS 12 PUBLICATIONS 31 CITATIONS
SEE PROFILE SEE PROFILE
Matieu Henry Rashed Jalal

Food and Agriculture Organization of the United Nations Food and Agriculture Organization of the United Nations
95 PUBLICATIONS 4,732 CITATIONS 23 PUBLICATIONS 151 CITATIONS
SEE PROFILE SEE PROFILE
Some of the authors of this publication are also working on these related projects:
Applications of UAV in Forest Resources Management: Case Study from AIT, Thailand View project
Vulnerability and Adaptive Capacity of Urban Communities in a Multi-hazard Coastal Area of Bangladesh View project
All content following this page was uploaded by Md. Akhter Hossain on 17 August 2021.
The user has requested enhancement of the downloaded file.

Estimation Procedures of
Indicators and Variables of the
Bangladesh Forest Inventory
June 2019
The Forest Department of Bangladesh leads actions to improve forest management and conservation,
adopting forward thinking, innovative approaches in its management of approximately 1.55 million
hectares of land across the country.
In 2015, the Forest Department began a process to establish a National Forest Inventory and Satellite Land
Monitoring System for improved forest and natural resource management. The process supports national
objectives related to climate change mitigation and provides information in support of the UN REDD
programme aimed at Reducing Emissions from Deforestation and Forest Degradation (REDD+). The
process also addresses domestic information needs and supports national policy processes related to forests
and the multitude of interconnected human and environmental systems that forests support.
The activities implemented under the Bangladesh Forest Inventory process are collaboration between
several national and international institutions and stakeholders. National partners from multiple government
departments and agencies assist in providing a nationally coordinated approach to land management.
International partners, including the United Stated Agency for International Development (USAID), the
Food and Agriculture Organization of the United Nations (FAO) and SilvaCarbon are supporting the
development of technical and financial resources that will assist in institutionalizing the process.
The results will allow the Forest Department to provide regular, updated information about the status of
trees and forests for a multitude of purposes including for assessment of role of trees for firewood,
medicines, timber, climate change mitigation.
CONTACTS:
Zaheer Iqbal Matieu Henry
National Project Coordinator Chief Technical Advisor
Forest Department Food & Agriculture Organization of the United
Email: z.iqbal60@gmail.com Nations
Email: matieu.henry@fao.org
Suggested Citation: Hossain, M.A., Aziz, A., Chakma, N., Johnson, K., Henry, M., Jalal, R., Carrillo, O.,
Scott, C., Birigazzi, L., Akhter, M., Iqbal, Z. 2019. Estimation Procedures of Indicators and Variables of
the Bangladesh Forest Inventory. Forest Department and Food and Agricultural Organization of the United
Nations, Dhaka, Bangladesh.
DISCLAIMER
This document is designed to reflect the activities related to the Bangladesh Forest Inventory process. It
does not reflect the official position of the supporting international agencies including USAID, FAO or
SilvaCarbon and should not be used for official purposes. Should readers find any errors in the document
or would like to provide comments for improving its quality they are encouraged to contact one of above
contacts.
2
Contents
1 Introduction ......................................................................................................................................... 6
2 An overview of the BFI ....................................................................................................................... 7
2.1 Biophysical inventory ................................................................................................................. 8
2.1.1 Overview of the collected data ........................................................................................... 9
2.2 Socio-economic survey .............................................................................................................. 15
2.2.1 Overview of the survey design ......................................................................................... 16
2.2.2 Data collection, management, and cleaning .................................................................... 17
2.3 Integration of biophysical, socio-economic and remote sensing data................................... 19
3 Statistical methodologies for biophysical data analysis ................................................................. 19
3.1 Estimations of plot level forest attributes ............................................................................... 19
3.2 Area estimations ........................................................................................................................ 19
3.2.1 Extent of trees and forests ................................................................................................... 19
3.2.2 Surveyed area estimations ................................................................................................... 21
3.3 Growing stock estimation ......................................................................................................... 21
3.3.1 Tree volume ........................................................................................................................ 22
3.3.2 Dead wood volume ............................................................................................................. 24
3.4 Biomass estimation.................................................................................................................... 25
3.4.1 Above ground biomass........................................................................................................ 25
3.4.2 Below ground biomass ........................................................................................................ 26
3.4.3 Dead wood biomass ............................................................................................................ 27
3.5 Carbon estimation ..................................................................................................................... 29
3.5.1 Carbon in above ground biomass ........................................................................................ 30
3.5.2 Carbon in below ground biomass ........................................................................................ 31
3.5.3 Carbon in dead wood .......................................................................................................... 31
3.5.4 Carbon in litter .................................................................................................................... 32
3.5.5 Soil carbon .......................................................................................................................... 32
3.6 Estimation of additional parameters ....................................................................................... 33
3.6.1 Stem density ........................................................................................................................ 33
3.6.2 Basal area and stem density ................................................................................................ 33
3.6.3 Diversity indices ................................................................................................................. 33
3.7 Population level estimations of means, total, variance, and confidence interval ................. 33
3.7.1 Ratio-to-Size Estimation ..................................................................................................... 34
3
3.7.2 Attributes from Different Plot Sizes ................................................................................... 34
3.7.3 Estimation of means and totals ........................................................................................... 34
4 Estimation of socio-economic indicators ......................................................................................... 37
4.1 Indicator 1: Quantity of each of the primary tree and forest products collected ............... 38
4.2 Indicator 2: Economic value of primary tree and forest products collected ....................... 39
4.3 Indicator 3: Quantity of each of the collected primary tree and forest products sold........ 40
4.4 Indicator 4: Income from selling primary tree and forest products .................................... 41
4.5 Indicator 5: Quantity of the processed tree and forest products .......................................... 42
4.6 Indicator 6: Quantity of each of the processed tree and forest products sold ..................... 43
4.7 Indicator 7: Income from selling processed tree and forest products .................................. 43
4.8 Indicator 8: Involvement with tree and forest related activities ........................................... 44
4.9 Indicator 9: Proportion of households receiving different tree and forest related services45
4.10 Indicator 10: Value of collected tree and forest products used for cooking and heating ... 45
4.11 Indicator 11: Cost of buying tree and forest products used for cooking and heating ........ 46
4.12 Indicator 12: Presumed amount of purchased tree and forest products used for energy .. 47
4.13 Indicator 13: Total annual income from tree and forest ....................................................... 47
5 References ........................................................................................................................................... 49
4
List of Acronyms
AE Allometric Equation
AGB Above Ground Biomass
BD Bulk Density
FD Forest Department
BFI Bangladesh Forest Inventory
BFIS Bangladesh Forest Information System
BGB Below Ground Biomass
CAGB Carbon in AGB
CBGB Carbon in BGB
CDWM Carbon Down Woody Matter
CWD Coarse Woody Debris
DBH Diameter at Breast Height
DGPS Differential Global Positioning System
DWM Down Woody Matter
FAO Food and Agriculture Organization of the United Nations
FGD Focus Group Discussion
FRA 2020 Global Forestry Resource Assessment 2020
FWD Fine Woody Debris
GS Growing stock
HH Household
NLRS National land class representation system
QAQC Quality Assurance and Quality Checking
RFID Radio-frequency Identification
SFM Sustainable Forest Management
SOC Soil Organic Carbon
SP Subplot
TC Tree Cover
WD Wood Density
5
Estimation Procedures of Indicators and Variables of the
Bangladesh Forest Inventory
1 Introduction
Historically, forest inventories have been commodity oriented with more emphasis on estimating the
commercial aspects of forests such as area, timber volume etc. Over time the conceptual basis of the national
forest inventories has been changed. Now-a-days forest inventory data focuses on broader ecological, social
and environmental issues such as biodiversity, forest restoration, recreation and aesthetics, ecosystem
services, livelihoods, disturbances and vulnerabilities.
The Bangladesh Forest Inventory (BFI) was implemented from 2016 to 2019 by the Forest Department
(FD) with technical support from Food and Agriculture Organization of the United Nations with cross-
disciplinary collaboration from affiliated government agencies, academic institutions, non-government
organizations, private industry and development partners. It’s a multi-purpose process and the centerpiece
of forest monitoring system with approaches to socio-economic, forest resources and tree cover monitoring.
The ultimate goal is to support the Government’s activities towards sustainable forest management (SFM)
by producing reliable datasets on forests, trees and livelihoods which inform planning, interventions, and
national and international frameworks such as Sustainable Development Goals (SDG’s).
In order to better manage the interactions between people and natural resources for SFM, the BFI is
integrated with field inventory, remote sensing and socio-economic survey information. Through field
inventory biophysical data has been collected from pre-designed sample plots whereas socio-economic data
has been collected from households across the country. Remote sensing technology was applied for land
cover classification and mapping through collaborative efforts of different expert agencies. The wide ranges
of data produced through these surveys are the basis for estimation of the criteria, indicators and variables
outlined in GoB (2017a).
Criteria and indicators are the key for monitoring the achievements or process towards sustainable forest
management whereas variables are established by data that enhances the specificity of an indicator. C&I
framework delineated in GoB (2017a) was developed under BFI for sustainable management of trees and
forest resources, fulfilling the national and international reporting requirements as well as policy makers’
data needs.
This document describes the methods for estimating the BFI indicators. It also outlines the integration
techniques of selected socio-economic information to the biophysical ones. Ratio to size estimators were
used for uncertainty estimation across the strata and subpopulations. The document references a number of
supplementary documents which provide details about all aspects of the BFI design, enumeration methods,
data management, relevant trainings and sample-based estimation systems. The primary intended audience
for this document is the BFI program itself with purposes to ensure sustainability, adoptability, adaptability
and feasibility of the estimation process considering the manpower and organizational setup of FD.
Nevertheless, other user and stakeholder will find it useful for understanding technical details and statistical
techniques of the BFI indicator and variables estimation as well as the integration process.
6
2 An overview of the BFI
The BFI is a constant and comprehensive process that assesses, evaluates, interprets and reports on the
status of trees and forest resources nationally. Field implementation of BFI is completed in three sessions
(2016-17, 2017-18 and 2018-19) starting with biophysical survey from November 2016 and completed in
April 2019. On the other hand, households’ interviews of socio-economic survey were conducted during
2017-18. The characteristic features of BFI are-
- Integration of remote sensing, biophysical and socio-economic survey for relating trees and forest
to society and environment,
- Dissemination of knowledge on forestry technologies by using latest equipment and open sourced
tools for data management,
- Quality assurance and quality control through hot and cold check of 10% sample plots of
biophysical survey and hot checking 5% household interviews in socio-economic survey,
- Establishment of a Bangladesh Forest Information System (BFIS) for providing user access to the
information.
In both surveys, field data collection, data cleaning, quality control, and data archiving were part of a
simultaneous process performed both in the field and in the central office (BFD, 2016b, 2016e, 2017b; M.
F. Kumar, Costello, Mahamud, Henry, & Johnson, 2017). Biophysical estimation were done in R statistical
software version 3.5.0, while socio-economic estimations were done using Stata statistical software. The
R-scipts are be found in the BFIS e-library (Section 2.4) (M. A. Hossain, 2017; M.A. Hossain, Laurent, &
Birigazzi, 2017; M.A. Hossain, Laurent, Sola, Birigazzi, & Aziz, 2017; Laurent & Hossain, 2017). Figure
1 shows an overview of the process and more details that are specific to the surveys are presented below.
7
Figure 1. Flow diagram of the data management, data cleaning, QA/QC and estimation processes used in
the biophysical and socio-economic surveys.
2.1 Biophysical inventory
The biophysical inventory follows a stratified systematic random sample design (Iqbal et al., 2016)
consisting of five zones based on the climatic and geophysical properties of Bangladesh (BFD, 2016f). The
design of BFI is a pre-stratified systematic sampling with specified sampling intensities for each zone. In
order to capture the national level data within an acceptable range of error with relatively low sampling
intensity, a total of 1858 permanent sample plots were fixed for field survey. Neyman allocation was
followed for allocation of plots at different ecozones (Cochran, 1977). The principle was to measure each
plot in one day and which were consisted of 3 and 5 subplots of 19m radius in Sundarbans and the other
zones respectively. The measurement of the field plots was performed by 13 field inventory teams and 4
quality assurance and quality checking (QAQC) teams were responsible for verification of data by ensuring
the implementation of the 7% hot and 3% cold checks provided that the budget is available (Iqbal et al.,
8
2016). Moreover, to record the precise location of the samples plots another manual on Differential Global
Positioning System (DGPS) use and Radio-frequency Identification (RFID) chip installation was prepared
for the respective team. The inventory teams including the field and quality checking teams were trained
on the species identification and recording, calibration and use of the instrument, using field forms for data
input, land feature and issues identification.
2.1.1 Overview of the collected data
Two versions of Open Foris were used in the BFI program whereas Open Foris Mobile was installed
handheld electronic tablets for data collection and Open Foris Collect was installed in centralized desktop
computer for data receiving and processing. The data collected through biophysical survey relate to plot
and subplot location, land feature, trees and sapling details, seedlings, down woody debris, soil and litter
measurement, bamboo etc. Trees with DBH ≥ 30 cm were measured in 19m plot whereas trees with DBH
≥ 10 and DBH ≥ 2 cm were measured in 8 m and 2.5 m radius plots respectively. The trees were categorized
as live, dead standing, live stump and dead stump status. Soil samples were collected at 8m at 270° bearing
from the subplot center. Moreover, from each plot detail data were collected on the plot location, land
feature, land ownership, down woody debris, bamboo and seedlings. Description and technical details of
data collection are presented in BFD (2016) and Iqbal et al. (2016). Table 1 presents more detailed
documentation of survey design, field instructions, soil sampling, data management and trainings.
Table 1: List of manuals, trainings, and R-scripts provided to the field teams for smooth and consistent data
collection, recording and management.
Manuals
BFD (2016). Field Instructions for the Bangladesh Forest Inventory (Version 1.2). Forest
Department and Food and Agricultural Organization of the United Nations, Dhaka, Bangladesh.
BFD. (2016). Manual for Soil Measurements for The Bangladesh Forest Inventory. Bangladesh
Forest
Department and Food and Agricultural Organization of the United Nations. Dhaka, Bangladesh.
BFD (2016). Quality assurance and quality control for the Bangladesh Forest Inventory. Forest
Kumar, M. F., Costello, L., Mahamud, R., Henry, M., Johnson, K. (2017) Bangladesh Forest
Inventory Data Management Protocol. Forest Department and Food and Agricultural Organization
of the United Nations. ISBN: 978-984-34-4275-8
Kumar, M. F., Mahamud, R., Costello, L., Sarkar, N., Jhonson, K., Hossain, A.M., Henry, M.
(2017). Field Manual on DGPS and RFID chip for Bangladesh Forest Inventory. Forest Department
and Food and Agricultural Organization of the United Nations. Dhaka, Bangladesh.
BFD. 2016. Protocol for Describing Land Features in Bangladesh, Data Collection Field Manual.
Forest Department and Food and Agriculture Organization of the United Nations, Dhaka,
Bangladesh.
Trainings
Akhter, M. and L. Costello (2016). Proceedings of the Equipment training for the implementation
of BFI, Forest Department and Food and Agriculture Organization of the United Nations.
9
Akhter, M. & Costello, L. 2016. Proceedings of the Training on Bangladesh Forest Inventory. 6-12
October 2016, BRAC CDM, Gazipur, Forest Department and Food and Agriculture Organization
of the United Nations.
Chakma, N. (2016). Proceedings of the Information sharing meeting on Bangladesh Forest

Inventory implementation in the Chittagong Hill Tracts (Rangamati, Khagrachari and Bandarban
hill district), Forest Department and Food and Agriculture Organization of the United Nations.
Falgoonee, K. M. & Henry, M. 2016. Proceedings of the Training of Trainers for the Bangladesh
forest inventory. 02-08 October 2016, Dhaka, Forest Department and Food and Agricultural
Organization of the United Nations.
Kumar, M. F., Iqbal, M. Z., Mahmood, H., Costello, L., Henry, M., Rahman, L. M., Jalal, R., Das,
S., Sidik, F., Hayden, H., Birigazzi, L., Uddin, M., Uddin, N., Akhter, M., Newaz, Y., Siddiqui, B,
N., Sola, G., Ahmed, I., Nishad, H, M., Salahuddin, M., Rahman, M., Chowdhury, R, M., Hossain,
B., Siddique, A, B., Rashed, A, Z, M, M., Misbahuzzaman, K., Siddique, M. R. H., Hoque, S.,
Hasan, M. N. (2017), Training materials for the biophysical component survey of the Bangladesh
Forest Inventory – concepts, planning and procedures, Forest Department, Ministry of Environment
and Forests, Government of the People’s Republic of Bangladesh.
Mahamud, R., et al. (2017). Proceedings of Second DGPS training for the Bangladesh Forest
Inventory, Forest Department and Food and Agriculture Organization of the United Nations.
Kumar, M. F., et al. (2017). Proceedings of The Bangladesh Forest Inventory field teams Refresher
training, Forest Department and Food and Agricultural Organization of the United Nations.
Uddin, N. (2016). Proceedings for the training on tree species identification. Forest Department and
Food and Agriculture Organization of the United Nations. Dhaka, Bangladesh.
R-scripts
Hossain, M. A. 2017. R-Script for Quality Assurance and Quality Checking of Bangladesh Forest
Inventory Soil and Litter Data. Food and Agriculture Organization of the United Nations, Dhaka,
Bangladesh.
Hossain, M. A., Laurent, S. & Birigazzi, L. 2017. R-Script for Bangladesh Forest Inventory Data
Analysis. Food and Agriculture Organization of the United Nations, Dhaka, Bangladesh.
Hossain, M. A., Laurent, S., Sola, G., Birigazzi, L. & Aziz, T. 2017. R-Script for Bangladesh Forest
Inventory Data Quality Assurance and Quality Checking. Food and Agriculture Organization of the
United Nations, Dhaka, Bangladesh.
Field data collection concluded by visiting 1781 out of 1858 plots (1480 accessible, 301 partially
accessible), or 96% of the total. Among those plots which were not visited were 42 inaccessible plots (2.3%)
and 35 nonsampled plots (1.9%) (Figure 2). Partially accessible plots were cases where at least one subplot
could be measured but other subplots were not possible to measure due to water (44%), hazardous condition
such as steep slopes (25%), wall or building (21%), or other reasons such as denied access, restricted areas
or border areas (10%). In the case of inaccessible plots, a visit was attempted but tree measurements were
not possible, most commonly due to the plot center falling inside hazardous conditions such as steep slopes
(52%), water (33%), or other reasons such as denied access, restricted areas, or wall or building (15%).
10
Still, some information could be collected from inaccessible plots such as parameters for assigning land
cover classes. Finally, a nonsampled plot status means that no parameter was measured on any subplot,
usually due to plot falling in an extremely remote location (69%), restricted area (20%), or other reason
such as water or border area (11%). The effect of inaccessible and nonsampled plots is not expected to
substantially bias national or zone level estimates because they were relatively few (12% of the Hill zone
plots; 4% of the total) and mostly randomly distributed.
Figure 2. The distribution of plots according to the accessibility status.
2.1.1.1 Data management and cleaning

The data collected though Open Foris Mobile are is exported to a Dropbox folder in ‘.collect-data’ files and
then imported to the Open Foris desktop version every day for conversion to other data formats (e.g. .csv,
.xml). At this point, the data are used for analysis. In cases of unavailability of communication networks or
11
other problems, field teams submit data to BFI unit through USB toolkits (i.e. pen drive etc.) or email. After
importing, the data is archived into the mass storage for future backup. MS Access and R code were used
for updating the database regularly by doing necessary queries. For data cleaning and cold checking
individual plot reports were prepared as .pdf from MS Access. Preliminarily, data cleaning was done in the
Open Foris platform manually based on the errors identified in the plot report. After cleaning data in Open
Foris platform they were then extracted in .csv form for identifying the inconsistencies or outliers through
customized R-scripts. Then, based on the QC reports from R-scripts, data cleaning is done for 2nd time in
Open Foris. After the cleaning is completed, cleaned data is then exported in the.csv format for analysis
using customized R-script.
2.1.1.2 Allometric models
Volume and biomass were the two most important variables as revealed in different stakeholder
consultations. Species or site specific Allometric Equations (AE) that represent as much as possible
ecological conditions are essential for the estimation of biomass. Under the BFI program, the existing 517
allometric equations were considered and reviewed based on statistical credibility, applicability, operational
and conceptual validity (Table 2). A total of 222 models belonging to 39 plant species were reported valid
among which were 189 species specific tree allometric equations (representing 35 tree species). These
equations were suitable to use in the calculation of parameter of interest (i.e. volume, biomass etc.) at
different levels but still the number of AE was insufficient considering the diversity of tree species and
ecosystems in Bangladesh (M. Hossain, 2016; M Hossain, Siddique, & Akhter, 2016). The details of model
validation process is described in M. Hossain (2016) along with the list of all 222 varied models.
Table 2: Number allometric equations according to tree forms and nature of equation (Mahmood et al. 2016) [here, T = total
number of equations, V= Number of verified equations]
Tree forms Number of equations based on their nature
Volume Green Oven-dried Air-dried Carbon Nutrients Length of

biomass biomass biomass split leaf
T V T V T V T V T V T V T V
Tree 360 138 78 44 11 10 0 0 25 1 3 3 0 0
Palm 0 0 2 0 0 0 0 0 0 0 0 0 1 0
Bamboo 0 0 3 0 0 0 3 0 0 0 0 0 0 0
Total 360 138 83 44 11 10 3 0 25 1 3 3 1 0
2.1.1.3 Development of new allometric models

Among the validated models, there were 20% green-biomass equations, 12% oven-dry biomass equations,
and 62% were volume equations which were not deemed suitable to apply to the more than 300 tree species
found in the inventory. Therefore, to meet this need, five initiatives were taken to develop four zone specific
allometric equations and six major tree species-specific equations (Akhter, Hossain, & Birigazzi, 2013; M
Hossain et al., 2016). Several training programs were organized for capacity development of the national
stakeholders for developing the equations. Then, field work was conducted to collect tree data using semi-
destructive methods including, wood density, biomass and carbon expansion factors (Table 3). The list of
newly developed allometric equation that are not in M. Hossain (2016) are provided in Table 7.
12
2.1.1.4 Decision tree for selection of allometric equation
A decision tree for selecting allometric equation is designed that is divided into two parts i) Route A and
ii) Route B (Figure 3). In route A, there are 7 levels based on the area of consideration i.e. species, genus,
diameter range and zone. Since there is no country specific equation, so Pan-tropical equations will be
considered if there is no equation, for any species, upto level VI. According to this route if, for a species,
equation is available at species level then following levels will be ignored. If, for any species, equation is
not available at species level then equations at immediately next, “genus”, level will be searched if no
equation is available at that level also then equation under the next level “species without diameter range”
will be looked for and this procedure will be followed upto "zone equation without diameter range” level.
If no equation is found at that level also then pantropical equation will be used for tree biomass calculation.
Route B is to be followed after the levels, at which equation is available, being selected. This route is same
for all levels.
In route B there are four ways for each level which are mainly consisted of forest zone, tree forms (palms,
bamboos and other trees) and diameter range. Ways situated hierarchically in upper positions and with more
options is given priority. However, after following any of the ways, in case of multiple equations for a
certain species, the criteria at the right corner of route B will be followed in order to select one final equation
for biomass calculation of that species. Among the criteria, extent of biomass calculation (total or partial),
greater diameter range, vastness of sample size, higher coefficient of determination will be given priority
13
based on the hierarchical positions as shown in the Route B (Figure 3). The decision tree is expected to fit
the Bangladesh perspective for accurate assessment of tree biomass.
Figure 3: Decision tree for selecting allometric equation to estimate tree biomass
14
Table 3: Manuals, training proceedings along with other publications produced from the allometric equation
development activities of BFI
SN List of manuals, protocols and training proceedings on allometric equation development

1 Hossain, M. (2016). Improved National Tree Allometric Equation Database to Support Forest
monitoring and Assessment of Bangladesh. Forest Department, Food and Agricultural
Organization of the United Nations, Dhaka and Forestry and Wood Technology Discipline,
Khulna University, Khulna, Bangladesh.
2 Hossain, M. (2016). Proceedings of the training on tree allometric equation development and use,
Dhaka, Forest Department and Food and Agriculture Organization of the United Nations.
3 Hossain, M. (2016). Proceedings of the National Consultation Workshop on Tree Allometric
Equations in Bangladesh, Dhaka, Forest Department and Food and Agriculture Organization of
the United Nations and Forestry and Wood Technology Discipline, Khulna University, Khulna,
Bangladesh.
4 Hossain, M. (2016). Proceedings of the Training on Sample Processing and Laboratory Analysis
for the Development of Allometric Equation, Forest Department, Food and Agriculture
Organization of the United Nations, and Forestry and Wood Technology Discipline, Khulna
University, Bangladesh.
5 Mahmood, H., Siddique, M.R.H., Abdullah, S.M.R., Akhter, M. and Islam, S.M.Z. (2016).
Manual for Building Tree Volume and Biomass Allometric Equation for Bangladesh. Forest
6 Mahmood, H., Siddique, M.R.H., Abdullah, S.M.R., Matieu, H (2016). Training Manual: Sample
Processing and Laboratory Analysis for the Development of Allometric Equation. Forest
7 Mahmood, H., Siddique, M.R.H., Abdullah, S.M.R., Akhter, M., Islam, S.M.Z. 2016. Field
measurement protocol on tree allometric equations for estimating above-ground biomass and
volume in Bangladesh. Food & Agriculture Organization of the United Nations, Rome, Italy and
Forest Department, Government of the People's Republic of Bangladesh, 95 pp.
8 Mahmood, H. (2018). Common tree allometric equations for the Sundarbans, Coastal and Village
zone of Bangladesh, 52 pp.
9 Mahmood, H. (2018). Proceedings of the training on- Data and sample analysis for the hill zone
allometric equation development. 17 pp.
2.2 Socio-economic survey

The socio-economic survey is designed to provide information about the interactions between people and
tree and forest resources as well as the valuation of tree and forest ecosystem services. Specifically, the
design aimed to achieve both national estimates of and also the spatial comparisons between areas of greater
versus lower impact or dependence on tree and forests. The survey was designed by multiple partners and
approved by Bangladesh Bureau of Statistics (GoB 2017b).
15
2.2.1 Overview of the survey design
A multi-stage random sample was used for the stratification of the survey (GoB 2017b). The assumption
was made that tree and forest ecosystem services and their relationship with local households depends on
the quantity of tree and forest resources and tree cover per household. Tree cover was obtained from
remotely sensed Landsat data from 2014 (Potapov et al., 2017), and household data from the ‘Population
and Household Census - 2011’ dataset (BBS, 2016). The variable Household Tree Availability or percent
tree cover per household (%TC HH-1) was calculated for each union or ward:
Household Tree Availability = ((TC area)/(Union area))/(Number HH's)=(%TC)/HH.
This metric was then defined into four Household Tree Availability Classes (HTAC) within each zone by
quartiles. The unions were classified using those four HTAC classes. Thus, the total sub-strata were 4
classes * 5 zones = 20 strata. The first quartile represents lowest %TC but highest number of HH’s per
union. In other words, strata 1 represents the lowest availability of tree and forest resources, or the highest
impact of the people on their resources, and strata 4 may represent the highest availability due to lowest
impact. Once the strata were defined, and to give the equal allocation of union to each stratum, 16 unions
(or wards) were randomly sampled from each strata (GoB 2017b).
To select which households would be surveyed within the pre-selected unions, 10 GPS points (five first
option and five second option) were placed randomly within the Rural Settlement land cover class (Figure
4). Interviewers then navigated to five GPS points and chose the nearest four households to interview so
that 20 households were visited in each union. A total of 6400 household from 320 unions were surveyed
(20 strata * 16 unions * 20 households = 6400 total households). The 2015 Land Cover Map was used by
enumerators to ask respondents from which land cover certain primary products were collected (BFD,
2017). The full questionnaire can be found in the Appendix 2.62.
16
Figure 4. Left: Map of 320 unions selected for socio-economic field data collection stratified by 20 HTAC’s.
Right: Map of a single union indicating 10 GPS points where field teams chose to visit either even or odd
numbers, but not both, for a total
In addition, 100 a qualitative survey was conducted through conducting 100 FGDs (Focus Group
Discussion) across the zones (20 in each zone) where 8-10 participants including community leader and
special forest user groups participated. In FGDs the location specific forest and socio-economic attributes
were focused in the discussion (GOB, 2017b).
2.2.2 Data collection, management, and cleaning
A questionnaire including all the variables meeting the objective of the socioeconomic survey was designed,
pretested in the field and then used for interviewing the selected households (Rahman & Jashimuddin,
2017). Enumerator manual and Open Foris form were developed for data collection and 5 days intensive
training were provided to the enumerators and QAQC experts for accurate and quality data collection from
the field. After being trained, the 5 enumerator teams (5 enumerators in each team) collected data across
the country. The household survey data were collected by household interview using Samsung Tablet while
the FGD community surveys used printed forms for recording the responses to specific queries. The
procedure of data gathering and organization for developing a database was done following M. F. Kumar,
Costello, L., Mahamud, R., Henry, M., Johnson, K (2017) which is also similar to that of biophysical survey.
17
For verification of household interviews and ensuring consistent field data collection, QAQC was
conducted in the 5% interviews (BFD, 2017). The process involved 7 experts from relevant universities,
government and non-government organizations. Quality controlling activities was also conducted indoor
after receiving the data through cleaning in Open Foris platform and checking inconsistencies or errors
through customized R-script. The whole process is well documented all the design materials, manuals, field
instructions, training proceedings as listed in Table 4.
Table 4: List of manuals, trainings, R-scripts and maps related to socio-economic survey design, training
on household interview and field instructions
Manuals
BFD (2017). Socioeconomic Field Instructions for the Enumerators of Bangladesh Forest
Inventory. Forest Department and Food and Agriculture Organization of the United Nations.
Dhaka, Bangladesh.
BFD (2017). Quality Assurance and Quality Control for the Socio-economic component of
Bangladesh Forest Inventory. Forest Department and Food and Agricultural Organization of the
United Nations, Dhaka, Bangladesh.
Chakma, N. (2017). Proceeding of National Consultation on Socioeconomic Survey Design of

the Bangladesh Forest Inventory, Dhaka, Forest Department and Food and Agriculture
GOB (2017). The socio-economic survey of the Bangladesh Forest Inventory, Forest
Department, Ministry of Environment and Forests, Government of the People's Republic of
Bangladesh.
Rahman, M. and M. Jashimuddin (2017). Field Test Report on Socioeconomic Survey, Forest
Department.
Trainings
Chakma, N., et al. (2017). Proceedings of the training of survey enumerators, and quality
assurance and quality control team members for the socio-economic component of the
Bangladesh Forest Inventory, Forest Department and Food and Agriculture Organization of the
United Nations.
R-scripts
Hossain, M.A. 2019. Socio-economic data quality assurance and quality check script. Dhaka,
Bangladesh, Food and Agriculture Organization of the United Nations.
Maps
Hira, S. and R. Jalal (2017). Land cover maps of 320 selected unions for the socioeconomic survey
of Bangladesh Forest Inventory. Food and Agriculture Organization of the United Nations, Dhaka,
Bangladesh: These maps are based on the land cover 2015 data.
Hira, S. and R. Jalal (2017). Basemaps of 320 selected unions for the socioeconomic survey of
Bangladesh Forest Inventory. Food and Agriculture Organization of the United Nations, Dhaka,
Bangladesh: These maps are based on the administrative boundary, road, river, etc.
18
Jalal, R. (2017). Selected unions for the socioeconomic survey of the BFI. Food and Agriculture
Organization of the United Nations, Dhaka, Bangladesh: Unions were selected based on stratified
random sampling based on tree cover, household density and union area data.
2.3 Integration of biophysical, socio-economic and remote sensing data

Application of both natural and social sciences is required for sustainable natural resource management
through development of common indicators as they are interrelated and interdependent (De Lange, Wise,
Forsyth, & Nahman, 2010). An integrated approach is required for making decisions concerning the
management of the natural resources, development of socio-economic condition of forest dependent
communities, strengthening forest co-management, reducing conflicts with stakeholder, and ensuring
local knowledge-based planning (Conley & Moote, 2003; Curtis, Byron, & MacKay, 2005; Lockwood,
2000).
Particular attention was put in the sampling design of socio-economic survey to ensure that the results are
compatible and consistent with the data collected through the surveys in order to facilitate the socio-
economic and biophysical integration. A total of 79 indicators were identified under eight criteria for the
BFI where a significant number of these indicators require information coming from both socio-economic
and biophysical components. The purpose is to know the interactions between socio-economic status and
the dynamics of trees and forest status. The integration will eventually support the development of national
policies, strategy processes and forestry planning. The integration process is described with examples in
GoB (2017a). However, some of the indicators that integrate both the biophysical and socio-economic
information are-
- Fuel wood demand and supply,
- Average income from primary tree and forests products collected by households from different
forest cover classes,
- Average value of tree and forest products collected by HHs
- Total value of the primary tree and forest products collected by households
3 Statistical methodologies for biophysical data analysis

3.1 Estimations of plot level forest attributes
This section describes the estimation procedures of the important forest attributes such as biomass, carbon,
volume, basal area and stem density at plot level. The tree distribution by diameter and height classes were
estimated at zone level. Forest area is estimated from both map and field survey at different levels including
plot, zone, forest type etc.
3.2 Area estimations

3.2.1 Extent of trees and forests
The BFI uses two compatible sources of information to understand changes in land area, volume, biomass,
and carbon stocks. First, the extent, spatial distribution and changes in land cover were assessed through
development of land cover map for the whole country used multi-spectral ortho (level 3) SPOT6/7 images
of 6-meter resolution with maximum 10% cloud coverage. The legend classes were derived from the
National Land Representation System (NLRS) of Bangladesh based on the distinction of classes from
satellite image interpretation, availability of ancillary data, and expert knowledge. Multi resolution
segmentation algorithm was used to develop image objects using the bands green, red and near-infrared
(NIR) with equal weights as input layers. The image objects developed were then used as the basic unit of
19
classification and land cover code was assigned to each segment by trained photo interpreters. Based on the
NLRS class hierarchy, the legend classes of the land cover map 2015 are aggregated for assessing extent
of forest, other wooded land, etc.
Land cover classes can be further categorized into Forest and Other Land following the FRA definitions
(FAO 2018). The determination of which classes to include in either class was the decision of the FD and
are outlined in Table 5. A separate column is included for BFI Land Cover Classes because Single Crop
and Multiple Crop could not be distinguished from the field data, so they were combined into one Permanent
Crop class. Also included are the land cover classes used in land cover changes estimates for the years
2000, 2005, and 2010.
Table 5: Land cover assignments to FRA categories and
BFI Land Cover FRA

Land Cover 2015 Land Cover (2000, 2005 and 2010)
Classes class
Forest Tree Dominated Area (Terrestrial) and
Bamboo Forest Bamboo Forest Forest Cultivated Trees (Hill Forest, Bamboo Forest, Forest
Plantation)
Forest Plantation Forest Plantation Forest Cultivated Trees (Hill Forest, Bamboo Forest, Forest
Plantation)
Hill Forest Hill Forest Forest Cultivated Trees (Hill Forest, Bamboo Forest, Forest
Plantation)
Mangrove Forest Mangrove Forest Forest Mangrove Forest
Mangrove Mangrove
Forest Mangrove Plantation
Plantation Plantation
Plain Land Forest Plain Land Forest
Forest Plain Land Forest (Sal Forest)
(Sal Forest) (Sal Forest)
Rubber Plantation Rubber Plantation Forest Rubber Plantation
Shrubs with Shrubs with Shrub Dominated Area (Terrestrial) (Orchards and
Forest
scattered trees scattered trees Other Plantations (Shrub), Shifting Cultivation)
Forest Tree Dominated Area (Aquatic/ Regularly
Swamp Forest Swamp Forest Forest
Flooded) (Swamp Forest, Swamp Plantation)
Other Artificial Surfaces (Air Port, Built-Up Non-Linear and
Air Port Air Port
Land Dump Sites / Extraction Sites)
Other
Baor Baor Baor
Land
Brackish Water Brackish Water Other
Aquaculture
Aquaculture Aquaculture Land
Other
Brickfield Brickfield Brickfield
Land
Built-Up Non- Built-Up Non- Other Artificial Surfaces (Air Port, Built-Up Non-Linear and
Linear Linear Land Dump Sites / Extraction Sites)
Dump Sites/ Dump Sites/ Other Artificial Surfaces (Air Port, Built-Up Non-Linear and
Extraction Sites Extraction Sites Land Dump Sites / Extraction Sites)
Fresh Water Fresh Water Other
Aquaculture
Aquaculture Aquaculture Land
Herb Dominated Herb Dominated Other
Herb Dominated Area
Area Area Land
Other
Lake Lake Lake
Land
20
Mud Flats or Mud Flats or Other
Mud Flats or Intertidal Area
Intertidal Area Intertidal Area Land
Other
Multiple Crop Permanent crop Permanent crop
Land
Orchards and Other Orchards and Other Other Shrub Dominated Area (Terrestrial) (Orchards and
Plantations (Shrub) Plantations (Shrub) Land Other Plantations (Shrub), Shifting Cultivation)
Orchards and Other Orchards and Other Other
Orchards and Other Plantations (Trees)
Plantations (Trees) Plantations (Trees) Land
Perennial Perennial Other
Perennial Beels/Haors
Beels/Haors Beels/Haors Land
Other
Ponds Ponds Ponds
Land
Other
River Banks River Banks Sand
Land
Other
Rivers and Khals Rivers and Khals Rivers and Khals
Land
Other
Rural Settlement Rural Settlement Rural Settlement
Land
Other
Salt Pans Salt Pans Salt Pans
Land
Other
Sand Sand Sand
Land
Other Shrub Dominated Area (Terrestrial) (Orchards and
Shifting Cultivation Shifting Cultivation
Land Other Plantations (Shrub), Shifting Cultivation)
Other
Single Crop Permanent crop Permanent crop
Land
Other
Swamp Reed Land Swamp Reed Land Swamp Reed Land
Land
Other
Forest Tree Dominated Area (Aquatic/ Regularly
Swamp Plantation Swamp Plantation Wooded
Flooded) (Swamp Forest, Swamp Plantation)
Land
3.2.2 Surveyed area estimations

Area estimates used in the denominator of ratios for volume, biomass, and carbon estimates are based on
the sampled areas, not the Land Cover Map. Nonetheless, the land cover classes of both the sample and the
map use the same system (Land Representation System – Bangladesh) and the resulting areas from both
sources were comparable. The plot wise land feature proportions recorded for the completely surveyed
subplots during the field measurement were used for calculations of actually surveyed area by land cover
class and zones. Subplots other than completely sampled ones were not considered in the sampled area
calculation as well as data analysis process.
3.3 Growing stock estimation

The growing stock (GS) is defined as “Volume over bark of all living trees with a minimum diameter of 10
cm at breast height (or above buttress if these are higher). Includes the stem from ground level up to a top
diameter of 0 cm, excluding branches” following the definitions of FAO (2018b) for global Forest Resource
Assessment (FRA) 2020. The GS is estimated differently for trees, bamboos and stumps (live) as described
in the following paragraphs.
21
3.3.1 Tree volume
3.3.1.1 Individual tree volume
Similar individual tree volume computation method was used for both live and dead standing trees
(DBH ≥ 10 cm) whereas live tree volume was considered as growing stock and dead standing tree
volume was dead wood volume. Validated models as mentioned in paragraphs 2.1.2 were used for
total volume estimation of individual trees. The associated features were considered to select
equation for a certain species following the decision tree described in Figure 3. For a certain
species the model to model features were compared to select the appropriate equation since there
was no automatic method or software program based on the decision tree. In total 31 species-
specific volume models for 30 species were selected and applied (Table 6). These species are most
important in Bangladesh as they consisted most of the tree population. However, for the other
species Huber’s volume equation associated with an average form factor of 0.5 for gross volume
was used (Equation 1). Though commercial volume of trees has significant values but total gross
volume would also serve many purposes since the demand for the natural resources in Bangladesh
is very high. From that consideration we focused only on the gross volume. The computation
procedure can also be followed in Section 2.2 of analysis R-script.
2
𝐷𝑏ℎ
𝑉𝑔 = × 𝜋 × 𝐻𝑡𝑜𝑡 × 𝑓𝑔𝑟𝑜𝑠𝑠 ------------------------Equation 1
4
Here,
𝑉𝑔 = Gross volume (m3) of individual tree,
𝐷𝑏ℎ = 𝑇𝑟𝑒𝑒 𝑑𝑖𝑎𝑚𝑒𝑡𝑒𝑟 𝑎𝑡 𝑏𝑟𝑒𝑎𝑠𝑡 ℎ𝑒𝑖𝑔ℎ𝑡 (𝑚),
𝜋 = 3.1416,
𝐻𝑡𝑜𝑡 = Tree total height (m),
𝑓𝑔𝑟𝑜𝑠𝑠 = Average form factor of trees here 0.50.
Table 2: List of models used for estimation of tree gross volume over bark [here, D = Diameter at Breast
Height (cm), C = Girth at Breast Height (cm), H = Height (m), D in= Diameter at Breast Height (inch), Hft
= Total height (ft), V = Volume (m3), Vcft = Volume (cft)]
SN Species Volume models R2 n DBH range

1 Acacia auriculiformis Log (V) = -11.506528 + 1.973377 × Log (C) + 0.623823 × Log (H) 0.979 124 25-125
2 Acacia mangium Log (V) = -10.7488 + 2.2178 × Log (C) 0.98 132 16-79
3 Acacia nilotica Log (V) = -11.875835 + 1.8823999 × Log (C) + 1.0819988 × Log (H) 0.91 128 25-115
4 Albizia saman Log (V) = -11.37623 + 2.26924 × Log (C) 0.981 153 <15 to >180
5 Albizia richardiana Log (V) = -10.996396 + 2.247808 × Log (C) 511 0.98 20->240
6 Albizia procera Log (V) = -11.6632 + 1.941989 × Log (C) + 0.754839 × Log (H) 221 0.991 20-> 130
7 Albizia spp. Log (V) = -11.19651 + 1.85690 × Log (C) + 0.67878 × Log (H) 140 0.979 <15-165
8 Aphanamixis Log (V) = -11.25528 + 1.98544 × Log (C) + 0.47163 × Log (H) 105 0.987 <15-180
polystachya
9 Artocarpus chaplasha Log (V) = -8.9449526 + 1.82851 × Log (D) + 0.735381 × Log (H) 427 0.984 5-66
10 Artocarpus Log (V) = -10.99533 + 1.80823 × Log (C) + 0.68951 × Log (H) 119 0.983 <15 to >180
heterophyllus
11 Avicennia officinalis V = 0.0089 + 0.0000264 × D2 × H 308 0.859 10-28
12 Azadirachta indica Log (V) = -11.42823 + 1.89235 × Log (C) + 0.71493 × Log (H) 36 0.985 15-105
13 Breonia chinensis Log (V) = -10.4647 + 2.3911 × Log (D) + 0.6373 × Log (H) 51 0.991
Neolamarckia cadamba
22
SN Species Volume models R2 n DBH range
14 Dalbergia sissoo Log (V) = -12.5189939 + 1.9800535 × Log (C) + 1.0775148 × Log 202 0.934 25-145
(H)
15 Dipterocarpus Log (V) = -8.5116354 + 2.35556 × Log (D) 436 0.979 8-58
turbinatus
16 Eucalyptus Log (V) = -9.3520 + 1.8055 × Log (D) + 0.8590 × Log (H) 0.986 511 3 - 18
camaldulensis
17 Eucalyptus V = 0.076339 - 0.00058066 × H + 0.000016216 × C2 + 0.0000032565 0.978 94 <60, >120
camaldulensis × C2 × H
18 Gmelina arborea Log (V) = -8.4687076 + 1.63502 × Log (D) + 0.784847 × Log (H) 0.966 486 8-76
19 Hevea brasiliensis Log (V) = -11.2768 + 1.8795 × Log (C) + 0.6928 × Log (H) 0.97 583 40-240
20 Lagerstroemia speciosa Log (V) = -9.6744 + 2.1065 × Log (D) + 0.6675 × Log (H) 0.986 74
21 Lannea coromandelica Log (V) = -11.519102 +2.01724 × Log (C) + 0.56356 × Log (H) 0.971 87 15-105
22 Mangifera indica Log (V) = -11.25377 + 1.96697 × Log (C) + 0.52237 × Log (H) 0.981 343 <15 to >180
23 Pinus caribaea Log (V) = -9.39412 + 1.867386 × Log (D) + 0.839034 × Log (H) 0.995 122 5 - >25
24 Senna siamea Log (V) = -11.6557 + 1.871 × Log (C) + 0.897 × Log (H) 0.99 120 16-79
25 Shorea robusta Log (V) = -12.0554 + 2.5178944 × Log (C) 0.967 499 31-189
26 Sonneratia apetala Log (V) = -9.29715 + 1.70514 × Log (D) + 0.95088 × Log (H) 0.98 236
27 Swietenia macrophylla Log (V) = -11.27102 + 1.88064 × Log (C) + 0.64629 × Log (H) 0.99 105 <15-165
28 Syzygium cumini Log (V) = -11.24854 + 2.24804 × Log (C) 0.966 99 <15-150
39 Tectona grandis Vcft = 0.000465 × Din1.58 × Hft1.603 0.92 645 7-62
30 Terminalia arjuna Log (V) = -11.3794 + 1.896423 × Log (C) + 0.653558 × Log (H) 0.997 177 20 to 120
31 Xylia xylocarpa Log (V) = -9.4303 + 2.0988 × Log (D) + 0.6042 × Log (H) 0.987 94
3.3.1.2 Stump volume

For stumps (live or dead), total length, diameter and diameter measurement height point was collected. But,
BFI data showed that, in some cases of stump total length (tree_total_lgt) was not filled but stump height
measurement point (tree_htdmp) is collected and entered in the form. There may be some technical flaws
in the form due to which the field teams couldn’t entered the total length of stumps. In those cases, the
stump height measurement point was assumed as total stump length. Stump volume is calculated assuming
it as cylindrical shaped (Equation 2). The form factor for stump is ignored because of it’s small size. The
computation procedure can also be followed in Section 2.3.2 of analysis R-script.
𝑆𝑣 = 𝑆𝑏𝑎 × 𝐻 -------------------------- Equation 2
Here, 𝑆𝑣 = Stump volume (m3), 𝑆𝑏𝑎 = Stump basal area (m2), 𝐻 = Stump height (m).
3.3.1.3 Bamboo volume

The Bamboo Culm Woody Volume has been calculated using the following Equation 3. The
detail analysis procedure can also be explored from Section 2.2.2 of analysis R-script.
2 (𝐷𝑏ℎ ×7)2
𝑉𝑏 = 𝐷𝑏ℎ × × 𝜋 × 𝐻𝑙𝑒𝑛𝑔𝑡ℎ × 𝑓𝑏𝑎𝑚𝑏𝑜𝑜 ----------------------------------Equation 3
4
Here,
𝑉𝑏 = Volume of bamboo culm (m3)
23
𝐷𝑏ℎ = Diameter of bamboo at breast height (m)
𝐻𝑙𝑒𝑛𝑔𝑡ℎ = Total height (m)
𝑓𝑏𝑎𝑚𝑏𝑜𝑜 = Form factor of bamboo (0.80)
3.3.2 Dead wood volume

Dead wood volume includes the volume of dead standing tree, down woody debris and dead stump
(Equation 4). The method varied with the type of dead wood volume. The detail analysis procedure of all
the dead wood volume can be explored from section of 2.3 of data analysis r-scrip.
𝑌𝑑𝑣 = 𝑌𝑠𝑑𝑡 + 𝑌𝑐𝑤𝑑 + 𝑌𝑓𝑤𝑑 + 𝑌𝑑𝑠 ------------------------------- Equation 4
Here, 𝑌𝑑𝑣 = Dead wood volume, 𝑌𝑠𝑑𝑡 = Standing dead tree volume, 𝑌𝑐𝑤𝑑 = Volume of Coarse Woody
Debris, 𝑌𝑓𝑤𝑑 = Volume Fine Woody Debris and 𝑌𝑑𝑠 = Volume of dead stump
3.3.2.1 Standing dead tree

In case of standing dead tree, the species names were recorded from the field. So, volume of standing dead
tree volume is computed using the same procedure (paragraph 3.3.1.1) followed for trees. See section 2.2.1
of analysis R-script for r-codes of standing dead tree volume.
3.3.2.2 Down Woody Material volume
Down woody debris (DWD) are of two types, i) Coase Woody Debris (CWD) and ii) Fine Woody Debris.
Volume computation process for each of the types at plot, subplot, land feature and transact level is same.
The following equation 5 proposed by Marshall, Davis, and LeMay (2000) was used for computing the
volume of both Coarse and Fine Woody Debris. See section 2.3.3.1 of analysis R-script for details of DWM
volume estimation.
𝜋2
𝑌𝐷𝑊𝑀𝑣𝑖 = × ∑𝑛𝑗=1 𝑑𝑖𝑗
2
------------------------------- Equation 5
8×𝐿
Here, 𝑌𝐷𝑊𝑀𝑣𝑖 = Volume of Down Woody Material (t/ha) for ith transact
𝐿 = Length (m) of transact where the CWD or FWD is surveyed. In BFI, there are 4 transacts in
each plot and length of each transact is 8m for CWD. On the other and, for FWD the transact
number is same but the length of transact is 8m for large FWD and 3m for small and medium FWD.
𝑛 = Number DWD in ith transact.
𝑑𝑖𝑗 = Diameter (cm) of the jth DWD of ith transact. In case of CWD the recorded diameter is used
but for FWD the average diameter was used for each of the classes following the FWD class
definition.
The transact level volume was then averaged at subplot and land feature level for each plot using equation
6. The volume density was then multiplied with the sampled area (𝐴𝑖𝑗𝑙 ) for getting total DWD volume at
subplot and land feature level (Equation 7).
1 𝑛
̅̅̅̅̅̅̅̅̅
𝑌𝐷𝑊𝑀𝑣 = ∑ 𝑌 ------------------------------- Equation 6
4 𝑖=1 𝐷𝑊𝑀𝑣𝑖
Here, ̅̅̅̅̅̅̅̅̅
𝑌𝐷𝑊𝑀𝑣 = DWD volume (t/ha) at subplot level
𝑌𝑡𝐷𝑊𝑀𝑣𝑖𝑗𝑙 = ̅̅̅̅̅̅̅̅̅
𝑌𝐷𝑊𝑀𝑣 × 𝐴𝑖𝑗𝑘 ------------------------------- Equation 7
24
Here, 𝑌𝑡𝐷𝑊𝐷𝑣𝑖𝑗𝑙 = total DWD volume (t) of lth land feature of jth subplot under ith plot
𝐴𝑖𝑗𝑙 = Sampled area of lth land feature of jth subplot under ith plot
3.3.2.3 Stump (dead) volume

Stump (dead) volume was estimated following the volume Equation 2 mentioned under stump biomass
estimation section (paragraph 3.3.1.2). See section 2.3.2 of analysis R-script for more details.
3.4 Biomass estimation

Biomass is estimated following the definition of Above Ground Biomass (AGB) and Below Ground
Biomass (BGB) proposed by FAO (2018b) for global forest resource assessment 2020 (FAO, 2018a). All
live tree (DBH ≥2 cm), bamboo and live stump data collected in the BFI was used for estimating above
ground and below ground biomass. See section 2.4 of analysis R-script for more details.
3.4.1 Above ground biomass

The above ground biomass is the function of tree diameter, height and wood density. Above ground biomass
contains the biomass of trees, saplings, bamboos and live stumps (Equation 8). Estimation procedure of
above ground biomass for trees and saplings, bamboos and stumps are different. See section 2.4.1 to 2.4.3
of analysis R-script for more details.
𝑌𝑎𝑔𝑏 = 𝑌𝑇𝑆𝑎𝑔𝑏 + 𝑌𝐵𝑎𝑔𝑏 + 𝑌𝑆𝑎𝑔𝑏 ------------------------------- Equation 8
Here, 𝑌𝑎𝑔𝑏 = Above ground biomass,
𝑌𝑇𝑆𝑎𝑔𝑏 = Tree and Sapling AGB,
𝑌𝐵𝑎𝑔𝑏 = Bamboo AGB, and
𝑌𝑆𝑎𝑔𝑏 = Stump (live) AGB.
3.4.1.1 Tree and sapling above ground biomass

A number of allometric equations were developed under the BFI for the five zones and 6 common species
of Bangladesh. In addition to these there were existing allometric equations verified under the BFI process
which were also considered for use to compute tree biomass. The decision tree described under paragraph
2.1.2.3 was used to select allometric equations. Table 7 showed the list of allometric equations selected
finally for biomass computation. See section 2.4.1 of analysis R-script for more details.
25
Table 7: Allometric equations used for estimation of above ground biomass [here, Yagb = above
ground biomass (kg), D = DBH (cm), H = total tree length (m), 𝜌 = wood density (kg/m3)]
SN Species AE equations selected for biomass estimation R2 Sample DBH range
size
1 Sonneratia apetala 𝑙𝑛 𝑌𝑎𝑔𝑏 = 1.7608 + 2.0077 𝑙𝑛(𝐷) + 0.2981 𝑙𝑛(𝐻) 0.9744 61 4-45
2 Excoecaria 𝑙𝑜𝑔 𝑌𝑎𝑔𝑏 = −0.8572 + 1.0996 𝑙𝑜𝑔 𝐷 2
agallocha
3 Acacia 𝑙𝑜𝑔(√𝑌) = −0.475 + 0.614 𝑙𝑜𝑔 𝐷2
auriculiformis
4 Acacia mangium 𝑙𝑜𝑔(√𝑌𝑎𝑔𝑏 ) = −0.497 + 0.606 𝑙𝑜𝑔 𝐷2
5 Heritiera fomes 𝑙𝑛 𝑌𝑎𝑔𝑏 = −2.1324 + 2.3895 𝑙𝑛(𝐷) + 0.1367 𝑙𝑛(𝐻) 97 2-40
6 Shorea robusta 𝑙𝑛 𝑌𝑎𝑔𝑏 = −3.3592 + 2.1830 𝑙𝑛(𝐷) + 0.6787 𝑙𝑛(𝐻) 42 8-38
7 Gmelina arborea ln Yagb = −3.028 + 0.925 × ln(D2 H) 0.991 16 5-15
8 Sal zone ln 𝑌𝑎𝑔𝑏 = −2.46 + 2.17 ln 𝐷 + 0.367 ln 𝐻 + 0.161 ln 𝜌 59 6-31
9 Sundarbans zone ln 𝑌𝑎𝑔𝑏 = −1.956299 + 2.163361 ln 𝐷 0.9852 82 2-78

+ 0.375219 ln 𝐻 + 0.689466 ln 𝜌
10 Village zone ln 𝑌𝑎𝑔𝑏 = −6.0325 + 1.9715 ln 𝐷 + 0.8193 ln 𝜌 0.9455 817
11 Hill zone ln 𝑌𝑎𝑔𝑏 = −6.9531 + 0.8250 ln(𝐷2 𝐻𝜌) 0.9334 175 15-58
3.4.1.2 Bamboo above ground biomass

There is no country specific allometric equation for biomass estimation of bamboo but two equations from
a literature search were found suitable to be used. Equation 9 proposed by Sohel, Alamgirb, Akhter, and
Rahman (2015) is suitable for Baijja bansh (Bambusa vulgaris) whereas the equation 10 proposed by de
Melo, Sanquetta, Dalla Corte, and Mognon (2015) is used for the other bamboos. See section 2.4.2 of
analysis R-script for more details.
𝑌𝐵𝑎𝑔𝑏 = 11.403 + 0.0006(𝐷 2 𝐻)-----------------------------Equation 9
𝑌𝐵𝑎𝑔𝑏 = 0.67545 + 0.02813(𝐷2 𝐻)-----------------------------Equation 10
3.4.1.3 Stump above ground biomass

Only live stump is considered for biomass estimation. Stump biomass is calculated following Equation 3
and Equation 11. At first, stump volume is calculated using Equation 2 then it is converted into biomass
using Equation 11. Stumps were assumed as cylindrical shaped. The form factor for stump is ignored
because of it’s small size. See section 2.4.3 of analysis R-script for more details.
𝑌𝑆𝑎𝑔𝑏 = 𝑆𝑣 × 𝜌 ------------------------- Equation 11
Here, 𝑆𝑣 = Stump volume (m3), 𝑌𝑆𝑎𝑔𝑏 = Above ground biomass of stump, (kg), 𝜌 = Wood density of
respective tree species which stump is left.
3.4.2 Below ground biomass

Similar to that of above ground biomass, the below ground biomass is consisted of tree, sapling, bamboo
and live stump. As described below the estimation procedure for all these tree and sapling, bamboo and live
stump is different (Equation 12). See section 2.6 of analysis R-script for more details.
26
𝑌𝑏𝑔𝑏 = 𝑌𝑇𝑆𝑏𝑔𝑏 + 𝑌𝐵𝑏𝑔𝑏 + 𝑌𝑆𝑏𝑔𝑏 ------------------------------- Equation 12
Here, 𝑌𝑏𝑔𝑏 = Above ground biomass, 𝑌𝑇𝑆𝑏𝑔𝑏 = Tree and Sapling AGB, 𝑌𝐵𝑏𝑔𝑏 = Bamboo AGB, and 𝑌𝑆𝑏𝑔𝑏
= Stump (live) AGB.
3.4.2.1 Tree and sapling below ground biomass

Below ground biomass of trees in Sundarbans and other 4 forest zones were estimated following two
different equation. The Pearson, Brown, and Birdsey (2007) formula (Equation 13) was used for estimating
below ground carbon for the Hill, Sal, and Village zones. See section 2.6.1 of analysis R-script for more
details.
𝑌𝑇𝑆𝑏𝑔𝑏 = 𝑒𝑥𝑝[−1.0587 + 0.8836 𝑙𝑛(𝑌𝑇𝑆𝑎𝑔𝑏 )] ---------------------------------- Equation 13
Here, 𝑌𝑇𝑆𝑏𝑔𝑏 = Below ground biomass (t/ha) in Hill, Sal, Coastal and Sundarbans zone
Besides, below ground biomass of Sundarbans was estimated using the equation proposed by Komiyama,
Ong, and Poungparn (2008), as given below (Equation 14).
𝑌𝑇𝑆𝑏𝑔𝑏 = 0.199 × 𝜌0.899 × (𝑑𝑏ℎ)2.22 ---------------------- Equation 14
Here, 𝑌𝑇𝑆𝑏𝑔𝑏 = below ground biomass in Sundarbans zone (kg); ρ = wood density; DBH = tree diameter at
breast height (cm), and 0.199, 0.899 and 2.22 are constants.
3.4.2.2 Bamboo below ground biomass

Below ground biomass of Bambusa vulgaris is calculated based on the conversion factor proposed by
Stokes, Lucas, and Jouneau (2007) and Bijaya and Bhandari (2008) who showed that the rhizome of
bamboo is 5% of the culms. For the other bamboo species, the ratio is 26% as reported by Lobovikov, Ball,
Guardia, and Russo (2007). Therefore, to get the bamboo BGB following formula (Equation 15 and
Equation 16) were used. See section 2.6.2 of analysis R-script for more details.
𝑌𝐵𝑏𝑔𝑏 = 𝑌𝐵𝑎𝑔𝑏 × 0.05------------------------Equation 15
𝑌𝐵𝑏𝑔𝑏 = 𝑌𝐵𝑎𝑔𝑏 × 0.26------------------------Equation 16
Here, 𝑌𝐵𝑎𝑔𝑏 = Above ground biomass (kg), 𝑌𝐵𝑏𝑔𝑏 = Bamboo below ground biomass (kg).
3.4.2.3 Stump (live) below ground biomass

Roots of live stumps possess substantial amount of biomass which doesn’t depend on the above ground
biomass of the stump. But, the diameter of the stump has a positive relationship with the extent of root
biomass. That’s why below ground biomass of stump is estimated following the equation prescribed by
Hjelm (2015). We used Equation 17 where DBH is the independent variable that indicates the relations
with the below ground wood (root) biomass. See section 2.6.3 of analysis R-script for more details.
0.00001×𝐷2.529
𝑌𝑆𝑏𝑔𝑏 = 1000
------------------------- Equation 17
3.4.3 Dead wood biomass

The dead biomass, also known as dead matter, is composed of standing dead tree, down woody debris, dead
stump biomass and litter dry weight (Equation 18). Estimation procedure of the dead woods is described in
the following sections. See section 2.5 of analysis R-script for more details.
27
𝑌𝑙𝑖𝑡
𝑌𝑑𝑏 = 𝑌𝑆𝐷𝑇𝑏 + 𝑌𝐷𝑊𝐷𝑏 + 𝑌𝐷𝑆𝑏 + ------------------------------- Equation 18
100
Here, 𝑌𝐷𝑏 = Dead wood biomass, 𝑌𝑆𝐷𝑇𝑏 = Standing dead tree biomass, 𝑌𝐷𝑊𝐷𝑏 = Biomass of Down Woody
Debris, 𝑌𝐷𝑆𝑏 = Dead stump biomass, 𝑌𝑙𝑖𝑡 = litter oven dry matter (g/m2) and 100 is used to divide litter oven
dry weight in order to convert g/m2 unit to t/ha.
3.4.3.1 Standing dead tree biomass

Standing dead tree biomass (𝑌𝑆𝐷𝑇𝑏 ) is calculated following the same procedure that followed for live trees.
But, a reduction factor (calculated temporarily from the biomass data of Sal and Sundarbans forests of
Bangladesh) for wood density (𝜌) was used based on the wood decay classes prescribed in BFI manual.
Wood density of the recorded species was multiplied by the reduction factor following Harmon, Woodall,
Fasth, Sexton, and Yatkov (2011) to get reduced wood density under three defined decay classes (Table 8
and Equation 19). The reduced wood density was then used in the selected allometric equations (in
necessary cases) for biomass calculation of standing dead trees. See section 2.5.1 of analysis R-script for
more details.
Table 8: Wood density reduction factors for standing dead trees
Decay class WD reduction factor (𝜌𝑟𝑓 ) Composition

Decay class 1 0.99 Bole + all branches (no leaves)
Decay class 2 0.8 Bole + big branches (no leaves and small branches)
Decay class 3 0.54 Bole without bark (No leaves, branches and bark)
𝜌𝑟 = 𝜌𝑟𝑓 × 𝜌 -------------------------- Equation 19
Here, 𝜌 = Wood density, 𝜌𝑟𝑓 = Wood density reduction factor. 𝜌𝑟 = Reduced wood density (kg/m3)
3.4.3.2 Biomass of Down Woody Material

Total biomass of down woody debris is the sum of fine woody debris and coarse woody debris. It is
calculated using the following formula (Equation 20). See section 2.5.3 -2.5.4 of analysis R-script for more
details.
𝑌𝐷𝑊𝑀𝑏 = 𝑌𝐶𝑊𝐷𝑏 + 𝑌𝐹𝑊𝐷𝑏 ---------------------------- Equation 20
Here, 𝑌𝐷𝑊𝑀𝑏 = Biomass of Down Woody Material, 𝑌𝐶𝑊𝐷𝑏 = Biomass of Coarse Woody Debris, 𝑌𝐹𝑊𝐷𝑏 =
Biomass of Coarse Woody Debris.
Biomass of Down Woody Matter was computed at plot, subplot, land feature, transact and decay class
level using the Equation 21 following Marshall et al. (2000).
𝜋2
𝑌𝐷𝑊𝑀𝑏𝑖 = × ∑𝑛𝑗=1 𝑑𝑖𝑗𝑘
2
𝜌𝑟𝑘 ------------------------------- Equation 21
8×𝐿
Here,
𝑌D𝑊M𝑏𝑖 = Biomass of Down Woody Matter (CWD or FWD) (t/ha) for ith transact,
𝐿 = Length (m) of transact where the DWM (CWD or FWD) is sampled. In there are 4 transacts in each
plot and length of each transact is 8m for CWD. On the other and, for FWD the transact number is same
but the length of transact is 8m for large FWD and 3m for small and medium FWD.
28
𝑑𝑖𝑗𝑘 = Diameter (cm) of the jth DWD of kth decay class under ith transact. In case of CWD the recorded
diameter is used but for FWD the average diameter was used for each of the classes following the FWD
class definition.
𝜌𝑟𝑘 = Reduced wood density of the kth decay class. It was calculated using Equation 19 where the 𝜌𝑟𝑓
(wood density reduction factor) for CWD is taken from the following Table 9 which was reported by
Harmon et al. (2011). In is relevant to mention here that the wood density used here is 0.6133782 g/cm2.
No decay class is recorded for the FWD so the wood density (0.6133782 g/cm3) was multiplied with an
average reduction factor (0.51) was to get an approximate reduced wood density (𝜌𝑟 ).
Table 9: Average density reduction factor (decayed density/undecayed density) for Downed Woody
Material (DWM) trees by decay class for hardwood species (Adjusted from Harmon et al. 2011).
Type Decay class Downed dead wood N

Hardwood 1 0.95 (0.01) 51
2 0.74 (0.02) 58
3 0.51 (0.03) 60
4 0.29 (0.02) 55
5 0.22 (0.02) 31
Note: Values are means with standard errors in parentheses. “n” is the number of DDM studied.
The transact level biomass was then averaged at subplot and land feature level for each plot using equation
22. The biomass density was then multiplied with the sampled area for getting total biomass of DWM at
subplot level and land feature level (Equation 23).
1 𝑛
̅̅̅̅̅̅̅̅̅
𝑌DWMb = ∑ 𝑌 ------------------------------- Equation 22
4 𝑖=1 DWMbi
Here, ̅̅̅̅̅̅̅̅
𝑌𝐷𝑊𝐷𝑏 = Mean DWD biomass (t/ha) at subplot level
𝑌𝑡𝐷𝑊𝑀𝑏𝑖𝑗𝑙 = ̅̅̅̅̅̅̅̅̅
𝑌𝐷𝑊𝑀𝑏 × 𝐴𝑖𝑗𝑙 ------------------------------- Equation 23
Here, 𝑌𝑡𝐷𝑊𝑀𝑣𝑖𝑗𝑙 = total DWM biomass (in ton) in lth land feature of jth subplot under ith plot
𝐴𝑖𝑗𝑙 = Sampled area of lth land feature of jth subplot under ith plot
3.4.3.3 Stump (dead) biomass

Stump (dead) biomass estimation procedure is similar to that described for live stump (paragraph 3.4.1.3).
See section 2.5.2 of analysis R-script for more details.
3.4.3.4 Litter oven dry weight
Litter samples were collected from the sample plots at plot, subplot and land feature level. The collected
samples were then analyzed in the laboratory and received data as oven dry weight directly was in g/m2
unit. See section 2.7.4 of analysis R-script for more details.
3.5 Carbon estimation

Carbon stock was estimated for 5 different pools - carbon in above ground biomass (CAGB), carbon in
below ground biomass (CBGB), carbon in down woody matter (CDWM), carbon in litter and soil organic
carbon (SOC). The methods of estimating carbon density and stocks in different pools is described in the
following paragraphs. See section 2.7 of analysis R-script for more details.
29
3.5.1 Carbon in above ground biomass
Carbon in AGB is the sum of carbon from AGB of trees and saplings, bamboos and live stumps (Equation
24). The proportion of carbon in AGB of trees and saplings, bamboo and live stump are different. Hence
the computation method is also different as described in the following paragraphs 3.5.1.1 to 3.5.1.3. See
section 2.7.1 of analysis R-script for more details.
𝐶𝑎𝑔𝑏 = 𝐶𝑇𝑆𝑎𝑔𝑏 + 𝐶𝐵𝑎𝑔𝑏 + 𝐶𝑆𝑎𝑔𝑏 ------------------------------- Equation 24
Here, 𝐶𝑎𝑔𝑏 = Carbon in above ground biomass,
𝐶𝑇𝑆𝑎𝑔𝑏 = Carbon in tree and sapling AGB,
𝐶𝐵𝑎𝑔𝑏 = Carbon in bamboo AGB, and
𝐶𝑆𝑎𝑔𝑏 = Carbon in stump (live) AGB. Live stumps occurred when branching and leaves from a cut
stump was observed.
3.5.1.1 Tree and sapling carbon
Carbon in above ground biomass of trees and saplings is estimated using the allometric equations mentioned
in Table 10. As mentioned earlier a number of allometric equations were developed under the BFI program
and selected based on the decision tree described in paragraph 2.1.2.3. The common allometric equation
for computing carbon in the trees and saplings of village zone was used to compute carbon for all the species
of coastal zone except the species having specific carbon allometric equation (i.e. Sonneratia apetala, E.
agallocha etc.). See section 2.7.1.1 of analysis R-script for more details.
Table 10: Allometric equations used for estimating the carbon estimation (Here, AGB= Above-ground
Biomass, CAGB = Carbon in AGB, D = Diameter at Breast Height, H = Total Height, WD = Wood Density)
SN zone Equation R2 Sample DBH

size range
1 Acacia log10 (√CAGB) = -0.630 + 0.614×log10(D2)
auriculiformis
2 Acacia mangium log10 (√CAGB) = -0.652 + 0.607×log10
(D^2)
3 Heritiera fomes Ln (CAGB) = -2.7488 + 2.4723×Ln (D)
4 Shorea robusta Ln (CAGB) = -3.9802 + 2.1660×Ln (D) + 42 8-38
0.6984×Ln (H),
5 Sonneratia Ln (CAGB) = -2.5035 + 2.0042×Ln (D) + 0.9754 61 4-45
apetala 0.3188×Ln (H)
6 Hill zone Ln (CAGB) = -7.7129 + 0.8268×Ln 0.9342 175 15-58

(D2×H×WD)
7 Sundarbans zone 0.9853 82 2-78
Ln (CAGB) = -7.5236 + 2.1628 × Ln (D) +
0.3834 × Ln (H) + 0.7004 × Ln (WD).
8 Sal zone 59 6-31
ln (CAGB) = -3.014 + 2.206×ln (D) +
0.302×ln (H) + 0.262×ln (WD)
30
3.5.1.2 Carbon in bamboo AGB
Bamboo carbon is estimated using the conversion factor of Allen (1986) that says carbon content is 54%
of bamboo oven dry biomass (Equation 25). See section 2.7.1.2 of analysis R-script for more details.
𝐶𝐵𝑎𝑔𝑏 = 𝑌𝐵𝑎𝑔𝑏 × 0.54 ------------------------- Equation 25
Here, 𝐶𝐵𝑎𝑔𝑏 = Carbon in bamboo AGB (t/ha), 𝑌𝑎𝑔𝑏 = bamboo AGB (t/ha) and 0.54 is the conversion factor
to convert biomass into carbon.
3.5.1.3 Carbon in stump (live) AGB

Carbon in stump (live) AGB was estimated as 50% of AGB (Equation 26). See section 2.7.1.3 of analysis
R-script for more details.
𝐶𝑆𝑎𝑔𝑏 = 𝑌𝑆𝑎𝑔𝑏 × 0.5 ------------------------- Equation 26
Here, 𝐶𝑆𝑎𝑔𝑏 = Carbon in stump (live) AGB (t/ha), 𝑌𝑎𝑔𝑏 = Stump (live) AGB and 0.5 is the conversion factor
to convert biomass into carbon.
3.5.2 Carbon in below ground biomass

Carbon in BGB is the sum of carbon in BGB of trees and saplings, bamboos and live stumps (Equation 27).
Carbon in BGB of trees and saplings, bamboos and live stumps was estimated as the 50% of the BGB
(Hairiah, Sitompul, van Noordwijk, & Palm, 2001; Matthews, 1997). See section 2.7.2 of analysis R-script
for more details.
𝐶𝑏𝑔𝑏 = 𝐶𝑇𝑆𝑏𝑔𝑏 + 𝐶𝐵𝑏𝑔𝑏 + 𝐶𝑆𝑏𝑔𝑏 ------------------------------- Equation 27
Here, 𝐶𝑏𝑔𝑏 = Carbon in BGB,
𝐶𝑇𝑆𝑏𝑔𝑏 = Carbon in tree and sapling BGB,
𝐶𝐵𝑏𝑔𝑏 = Carbon in bamboo BGB, and
𝐶𝑆𝑏𝑔𝑏 = Carbon in stump (live) BGB.
3.5.3 Carbon in dead wood

Dead wood carbon was computed from the dead wood biomass stock. The carbon concentration
in dead wood biomass is approximately 50% (Kauffman, Cummings, Ward, & Babbitt, 1995).
Applying this conversion factor, the carbon content of dead wood is computed from the plot,
subplot and land feature level dead biomass (Equation 14). See section 2.7.3 of analysis R-script
for more details.
𝐶𝑑𝑏 = 𝑌𝑑𝑏 × 0.50--------------------------Equation 28
Here, 𝐶𝑑𝑏 = Carbon stock in dead wood (t), 𝑌𝑑𝑏 = Biomass of dead wood (t)
31
3.5.4 Carbon in litter
Litter sample was collected and analyzed in the laboratory to get litter dry weight at plot, subplot and land
feature level in g/m2. In addition to the litter dry weight, the same litter sample was used to estimate carbon
percentages at plot and land feature levels. Then the litter carbon percentage is used to compute litter carbon
(Equation 29). Taking into account the sampled area per plot, subplot and land feature the litter carbon data
(t/m2) was then extrapolated for estimations of mean and variance. In cases the litter was not found the
value for litter of the respective subplots was let 0 and in cases the litter data was absent for any reasons it
was treated as missing value during estimations. See section 2.7.4 of analysis R-script for more details.
𝑌𝑙𝑖𝑡
𝐶𝑙𝑖𝑡 = × 𝐶𝑙𝑖𝑡𝑃 ----------------------------- Equation 29
106
Here, 𝐶𝑙𝑖𝑡 = Litter carbon (t/m2), 𝑌𝑙𝑖𝑡 = litter dry biomass (g/m2), 𝐶𝑙𝑖𝑡𝑃 = Litter carbon (%)
3.5.5 Soil carbon

Soil samples were analyzed in the laboratory for bulk density (g/cm3) and soil organic carbon (SOC)
percentage (%). Soil organic carbon was determined by multiplying %LOI by 0.5. Note that SOC analysis
was done by the Loss on Ignition (LOI) method without removing carbonates, which yields a higher
estimate of SOC than other methods such as Walkley-Black. This is especially true in highly calcareous
soils of the Sundarban and Coastal zone. If more comparable estimates to Walkley-Black are desired, one
should multiply the %SOC by 0.53 in the Sundarban zone and 0.34 in the Coastal zone.
Soil samples for bulk density and SOC were collected from the first three subplots (SP 1 to SP 3) of each
plot. For bulk density, individual samples were collected for analysis whereas the SOC sample was a mixed
for each land feature within the plot. In case of inaccessibility of any of these plots field were allowed to
collect soil sample from the 4th or 5th subplots. In Sundarbans and Coastal zones soil samples were collected
from three depths (i.e. 5-10 cm, 20-25 cm and 65-70 cm for bulk density and 0-15cm, 15-30 cm and 30-
100 cm for SOC) whereas in the remaining three zones (Hill, Sal and Village zones) soil samples were
collected from the first two depths (i.e. 5-10 cm and 20-25 cm for bulk density and 0-15cm and 15-30 cm
for SOC). For each layer, soil carbon was calculated from soil organic carbon and bulk density data
following method (Equation 30) proposed by Donato, Kauffman, and Stidham (2009). SOC in t/ha was
derived for each plot, subplot and land feature which was the interpolated in ton unit taking into account
the sampled area at plot, subplot and land feature level. See section 2.7.5 of analysis R-script for more
details.
𝑆𝑂𝐶𝑖𝑗𝑘𝑝 = 𝐵𝐷𝑖𝑗𝑘𝑝 × 𝐶𝑗𝑘𝑝 × 𝑆𝐷𝑖 × 100------------------------ Equation 30
Here,
𝑆𝑂𝐶𝑖𝑗𝑘𝑝 = Soil organic carbon (t/ha) in ith soil layer under jth land feature of kth subplot and pth plot,
𝐵𝐷𝑖𝑗𝑘𝑝 = Bulk density (g/cm3) of ith soil layer under jth land feature of kth subplot and pth plot,
𝑆𝐷𝑖 = Soil depth interval of ith layer (cm). Soil depth interval is 15 cm for 0-15 cm soil layer, 15 cm for
15-30 cm soil layer, and 70 cm for 30-100 cm soil layer,
𝐶𝑖 = Organic carbon (%) of jth land feature under kth subplot and pth plot, and
100 is a conversion factor to convert the units to t/ha.
32
At plot, subplot and land feature level the total soil carbon is the sum of carbon in all the subsequent layers
of respective zone. In case of Sundarbans and Coastal zone the total depth is 100 cm (Equation 31) whereas
in Hill, Sal and Villages zones the total depth is 30 cm (Equation 32) (BFD, 2016; Iqbal et al., 2016).
𝐶𝑠𝑖 = 𝐶𝑙1𝑖 + 𝐶𝑙2𝑖 ------------------------------------ Equation 31
𝐶𝑠𝑖𝑗 = 𝐶𝑙1𝑖 + 𝐶𝑙2𝑖 + 𝐶𝑙3𝑖 ------------------------------------ Equation 32
Here, 𝐶𝑠𝑖 = Total soil carbon in ith subplot, 𝐶𝑙1𝑖 = Carbon stock (t/ha) in 0-15 cm layer, 𝐶𝑙2𝑖 = Carbon
stock (t/ha) in 15-30 cm layer, 𝐶𝑙3𝑖 = Carbon stock (t/ha) in 30-100 cm layer
3.6 Estimation of additional parameters

Growing stock, biomass and carbon in 5 different polls were the key estimations of BFI. In addition to these
basal area, stem density, diversity index, seedlings density and recruitment percentage of seedlings were
estimated at different levels.
3.6.1 Stem density

Stem density is the simple count of tree or sampling individuals per unit area. Number of tree and sapling
individuals sampled at small (2.5 m radius) and medium (8 m radius) plots were extrapolated to large plot
(19 m radius) or subplot level. Stem density was then estimated at different levels i.e. species, land cover
classes, zone etc. following methods of Scott (2018) as described in the paragraph 3.7.
3.6.2 Basal area and stem density

Basal area of individual tree and sapling is calculated using the following Equation 33.
𝐷2
𝐵𝐴 = 𝜋 4
-------------------------------- Equation 33
3.6.3 Diversity indices

Four common diversity indices were computed at zone and Land Cover Class (LCC) level following the
ecosystem level equations as described below. The diversity indices named Shannon-Wiener Diversity
Index, Margalef’s Richness Index, Simpson’s Dominance Index and Pilou’s Evenness Index were
computed following Shannon (1963), Margalef (1958), Simpson (1949) and Pielou (1966). For reference
the equations (Equation 34 - 37) are mentioned below.
𝐻 = − ∑𝑆𝑖=1 𝑃𝑖 (ln 𝑃𝑖 ) -------------------------------- Equation 34

(𝑆−1)
𝑅= -------------------------------- Equation 35
ln(𝑁)
∑𝑆𝑖=1 𝑛𝑖 (𝑛𝑖 −1)

𝑆𝐷 = (1 −
𝑁(𝑁−1)
) -------------------------------- Equation 36
𝐻
𝐸= ln(𝑆)
-------------------------------- Equation 37
Here, H = Shannon-Wiener Diversity Index, R = Margalef’s Richness Index, SD = Simpson’s Dominance

Index, E = Pilou’s Evenness Index, N = total number of individuals of all the species; n = number of
individuals of ith species; Pi = proportional occurrence of ith species (ni/N); S = total number of species.
3.7 Population level estimations of means, total, variance, and confidence interval
After having the estimates at plot, subplot and land feature level the means, totals, variance and sampling
error were done following the ratio-to-size estimators applicable for Stratified Random Sampling. The
33
estimators were developed by Scott (2018). It is also relevant to mention that these estimators are based on
the assumptions and formula of Scott (2018) and Korhonen & Salmensuu (2014).
3.7.1 Ratio-to-Size Estimation

The Ratio-to-Size estimator is a ratio estimator that is used when the cluster size of a two-stage sample
varies, thus taking advantage of the relationship between the total of the attribute observed and the size of
the sampling unit (cluster). Korhonen and Salmensuu (2014) used the estimator to take advantage of the
relationship between any two attributes measured on the cluster. Specifically, this estimator is efficient
when the attribute of interest is correlated with the size of the attribute in the denominator. A common
example in forest inventory is the tree volume or biomass on a plot is related to the portion of the plot that
is forested. This situation occurs when plots are randomly placed on the landscape without regard to whether
the location is forested in order to determine the total forest area and volume within the population. A given
plot may straddle the boundary between forest and non-forest or between two different domains of interest,
such as forest types. This may also happen when a portion of the plot cannot be measured for safety or other
reasons (source: adapted from Scott 2018).
3.7.2 Attributes from Different Plot Sizes

Often in forest inventory, nested or concentric plots are used to sample trees of different sizes as used in
BFI. Seedlings and sapling were sampled on small plots (2.5 m radius) while medium trees (30 cm > DBH
≥ 10 cm) were sampled from medium sized plots (8 m radius) and large trees (DBH ≥ 30 cm) were sampled
on large plots (19 m radius). In order to estimate an attribute of interest across plot sizes, such as total
number of trees. Conceptually the total number of trees can be estimated using Equation 45 for each size
class, then summing across the size classes. The variance of this would be very difficult to determine due
to the covariances between the size class estimates. Instead, a simplification is to rescale each size class, j,
to the largest size class, J. The plot attribute becomes (Equation 38):
∑𝑛 𝑎𝑖𝐽
𝑦𝑖𝑑 = ∑𝐽𝑗 𝑦𝑖𝑗𝑑 ∑𝑖𝑛 -------------------------------- Equation 38
𝑖 𝑎𝑖𝑗
where:
yijd = sum of the attribute of interest in domain of interest d on plot i on plot size j
aij = plot size j area measured on plot i.
3.7.3 Estimation of means and totals

3.7.3.1 Estimation of attributes of interest for simple random sampling
In BFI, under each strata (zone) the survey method is simple random sampling. Hence, the mean, total,
ratio and variance estimations of the attributes of interest at stratum (zone) level were done following the
Equations 39 to 42. See section 3.1 to 3.14 of analysis R-script for more details.
The mean of the attribute of interest, such as biomass, in domain d across all land is computed with a
numerator as the total biomass measured that is in the domain of interest and the denominator as the total
area measured:
∑𝑛
𝑖 𝑦𝑖𝑑 ∑𝑛
𝑖 ∑𝑘 𝑦𝑖𝑘𝑑
𝑦̅𝑑 = ∑𝑛
= ∑𝑛
-------------------------------- Equation 39
𝑖 𝑎𝑖 𝑖 𝑎𝑖
where:
34
yid = sum of the attribute of interest in domain of interest d on plot i
yikd = the attribute of interest in domain d in plot i in condition k. The domain can refer to area
characteristics or can refer to the observational unit, such as tree species. Note that
observation unit values are simply summed and are not expressed on a per unit area basis,
such as the sum of sampled tree volumes.
When estimating values for trees sampled using different plot sizes, the largest plot size is used in the
denominator. For trees observed with smaller plot sizes, then the attribute is expanded to the larger plot
size by multiplying it by the larger plot area divided by the smaller plot area.
The attribute total (or the value for any category of interest) is estimated as:
𝑌̂𝑑 = 𝐴 ∙ 𝑦̅𝑑 -------------------------------- Equation 40

Where:
A = total area in the population
The variance of the estimated mean, (Equation 39), is obtained using Equation 41 as:
𝑛 ∑𝑛 ̅ 𝑑 )2
𝑖 (𝑦𝑖𝑑 −𝑎𝑖 𝑦 𝑛 ∑𝑛 2
̅ 𝑑 ∑𝑛
𝑖 𝑦𝑖𝑑 −2𝑦 ̅𝑑2 ∑𝑛
𝑖 𝑦𝑖𝑑 𝑎𝑖 +𝑦
2
𝑖 𝑎𝑖
𝑣(𝑦̅𝑑 ) = 𝑛−1 2 = 𝑛−1 2 --------------------------- Equation 41
(∑𝑛
𝑖 𝑎𝑖 ) (∑𝑛
𝑖 𝑎𝑖 )
and for the total is (Equation 42):
𝑣(𝑌̂𝑑 ) = 𝐴2 ∙ 𝑣(𝑦̅𝑑 ) -------------------------------- Equation 42
Ratio Estimates: Often estimates of means are of more interest for a particular subcategory of land, rather
than the whole land area, such as biomass per hectare of forest land. It can be estimated as the ratio of the
mean of the attribute across all land (Equation 39) divided by the mean area proportion across all land:
∑𝑛
𝑖 𝑦𝑖𝑑⁄
𝑦̅𝑑 ∑𝑛
𝑖 𝑎𝑖 ∑𝑛 𝑦𝑖𝑑
𝑅̂𝑑𝑑′ = = 𝑛
∑𝑖 𝑥𝑖𝑑′
= ∑𝑛𝑖 -------------------------------- Equation 43
𝑥̅𝑑′ 𝑖 𝑥𝑖𝑑′
⁄ 𝑛
∑𝑖 𝑎𝑖
where:
yid = the attribute of interest in domain of interest d on plot i where d is a subdomain of d’, for
example d is the teak volume on forest land.
xid’ = the attribute of interest in domain of interest d’ on plot i, where d’ is the primary domain
such as forest land.
The simplification is performed by noting that the denominators for both x and y are the same. Often the
denominator is a specific domain of an area attribute, such as forest area or forest type.
The variance of the ratio estimate can be computed using the same approach as in Equation 41, but
replacing ai with xid’. That is, instead of summing across all measured areas, only those areas in the
domain of interest are summed.
𝑛 ∑𝑛 ̂
𝑖 (𝑦𝑖𝑑 −𝑅𝑑𝑑′ 𝑥𝑖𝑑′ )
2 𝑛 ∑𝑛 2 ̂ 𝑛 ̂2 𝑛 2
𝑖 𝑦𝑖𝑑 −2𝑅𝑑𝑑′ ∑𝑖 𝑦𝑖𝑑 𝑥𝑖𝑑′ +𝑅𝑑𝑑′ ∑𝑖 𝑥𝑖𝑑′
𝑣(𝑅̂𝑑𝑑′ ) = 𝑛−1 2 = 𝑛−1 2 ------------------- Equation 44
(∑𝑛
𝑖 𝑥𝑖𝑑′ ) (∑𝑛
𝑖 𝑥𝑖𝑑′ )
35
3.7.3.2 Estimation of attributes of interest for stratified random sampling
The biophysical survey of BFI, as a whole, is a stratified random sampling. Hence, the mean, total, ratio
and variance estimations of the attributes of interest at stratum (zone) level were done following the
Equation 45 to Equation 53. See section 3.1 to 3.14 of analysis R-script for more details.
The mean across the entire population area is estimated as (Equation 45):
𝑛
∑𝑖 ℎ 𝑦ℎ𝑖𝑑
𝑌̅𝑑 = ∑𝐻 ̅ 𝐻
ℎ 𝑊ℎ 𝑌ℎ𝑑 = ∑ℎ 𝑊ℎ 𝑛 ------------------------------- Equation 45
∑𝑖 ℎ 𝑎ℎ𝑖
where:
Wh = weight for stratum h = Nh/N
Nh = area or number of first-phase sample points in stratum h
N= total area or total number of first-phase sample points across all strata
nh = total number of sample plots in stratum h
n= total number of sample plots across all strata
yhid = the attribute of interest in domain of interest d on plot i in stratum h
The total is estimated by multiplying by the total area (Equation 46):
𝑌̂𝑑 = 𝐴 𝑌̅𝑑 ------------------------------- Equation 46

Often estimates of means are of more interest for a particular subcategory of land, rather than the whole
land area, such as biomass per hectare of forest land. The ratio is computed by dividing the total of attribute,
Y, in domain d by the total of attribute, X, in domain of interest, d′ where d is a subset of d′ (Equation 47):
𝑌̂𝑑 𝐴 ∑𝐻 ̅
ℎ 𝑊ℎ 𝑌ℎ𝑑 ∑𝐻 ̅
ℎ 𝑊ℎ 𝑌ℎ𝑑
𝑅̂𝑑𝑑′ = 𝑋̂𝑑′
= 𝐴 ∑𝐻 𝑊 ̅
𝑋
= ∑𝐻 𝑊 ̅
𝑋
------------------------------- Equation 47
ℎ ℎ ℎ𝑑′ ℎ ℎ ℎ𝑑′
Note: It is important to first estimate the totals of Y and X across strata then create the ratio estimate,
rather than doing the ratio estimate by strata then averaging. The first method is unbiased, the second is
not.
The variances of these depend on the sampling design used. Stratified Random Sampling uses a
predetermined sample size in each stratum. Samples are drawn randomly within each stratum (Cochran
1977). The sizes of each stratum are known. The variance estimator of the mean for Stratified Random
Sampling (when N is large relative to n) is (Equation 48):
2 2
𝑊ℎ 𝑠ℎ
𝑣(𝑌̅𝑑 ) = ∑𝐻
ℎ ------------------------------- Equation 48
𝑛ℎ
where the stratum variance is (Equation 49):

𝑛 2 𝑛 2 ∑ ℎ 2 𝑛
𝑛ℎ2 ∑𝑖 ℎ 𝑦ℎ𝑖𝑑 −2𝑦̅ℎ𝑑 ∑𝑖 ℎ 𝑦ℎ𝑖𝑑 𝑎ℎ𝑖 +𝑦̅ℎ𝑑 𝑎ℎ𝑖
𝑠ℎ2 = 𝑛ℎ −1 𝑛 2
𝑖
------------------------------- Equation 49
ℎ
(∑ 𝑎 )
𝑖 ℎ𝑖
36
The variance of the total is:
2 2
𝑊ℎ 𝑠ℎ
𝑣(𝑌̂𝑑 ) = 𝐴2 𝑣(𝑌̅𝑑 ) = 𝐴2 ∑𝐻
ℎ 𝑛 ------------------------------- Equation 50
ℎ
The variance of the ratio estimate is approximated as:

1 2
𝑣(𝑅̂𝑑𝑑′ ) = 𝑋̂ 2 [𝑣(𝑌̂𝑑 ) + 𝑅̂𝑑𝑑 ̂𝑑′ ) − 2 𝑅̂𝑑𝑑′ 𝑐𝑜𝑣(𝑌̂𝑑 , 𝑋̂𝑑′ )] ------------------------- Equation 51
′ 𝑣(𝑋
𝑑′
where the covariance is:

2
𝑊ℎ 𝑐𝑜𝑣(𝑌ℎ𝑑 , 𝑋ℎ𝑑′ )
𝑐𝑜𝑣(𝑌̂𝑑 , 𝑋̂𝑑′ ) = 𝐴2 ∑𝐻
ℎ 𝑛
---------------------------- Equation 52
ℎ
In the simple means case, the stratum covariance is:

∑𝑛 ̅ ℎ𝑑 𝑥̅ℎ𝑑′
𝑖 𝑦ℎ𝑖𝑑 𝑥ℎ𝑖𝑑′ −𝑛ℎ 𝑦
𝑐𝑜𝑣(𝑌ℎ𝑑 , 𝑋ℎ𝑑′ ) = (𝑛ℎ −1)
---------------------------- Equation 53a
and in the Ratio-to-Size case, the stratum covariance is:

𝑛
𝑛ℎ2 ∑𝑖 ℎ (𝑦ℎ𝑖𝑑 −𝑎ℎ𝑖 𝑦̅ℎ𝑑 )(𝑥ℎ𝑖𝑑′ −𝑎ℎ𝑖 𝑥̅ℎ𝑑′ )
𝑐𝑜𝑣(𝑌ℎ𝑑 , 𝑋ℎ𝑑′ ) = (𝑛 𝑛 2
ℎ −1) (∑ ℎ 𝑎 )
𝑖 ℎ𝑖
𝑛 𝑛 𝑛 𝑛2
𝑛ℎ2 ∑𝑖 ℎ 𝑦ℎ𝑖𝑑 𝑥ℎ𝑖𝑑′ − 𝑦̅ℎ𝑑 ∑𝑖 ℎ 𝑎ℎ𝑖 𝑥ℎ𝑖𝑑′ − 𝑥̅ℎ𝑑 ∑𝑖 ℎ 𝑎ℎ𝑖 𝑦ℎ𝑖𝑑′ + (∑𝑖 ℎ 𝑎ℎ𝑖 )𝑦̅ℎ𝑑 𝑥̅ℎ𝑑′
= (𝑛 𝑛 2 ------- Equation 54
ℎ −1) (∑ ℎ 𝑎 )𝑖 ℎ𝑖
4 Estimation of socio-economic indicators
The motivation for the socio-economic survey was to understand the interaction between human and tree
and forests. The survey and the data collection instruments were designed in light of the set criterion
which were developed in a participatory process where FD, forest experts, academicians and civil society
organizations actively participated. This section describes the estimation procedure and techniques used
in the analysis. The results are presented at zone and national level. The indicators are estimated through
using Bootstrap estimation using replicate weight variables in STATA 14 with 320 replications (i.e. total
number of unions surveyed) and setting number of PSUs to be 16 (i.e. number of unions selected from
each stratum). But, as limited number of households were involved in production and selling of processed
products, in indicators related to these products, the number of replicates was greater than the number of
available units for some strata, and ultimately Bootstrap estimation was not possible. For processed
products the estimates were derived using the Stata command svyset, declaring the appropriate weight
and setting zones as strata. During the survey the households were asked about quantity, price, days
employed, income earned, different services received from tree and forests and other livelihood related
questions. However, for estimating value and income from tree and forest products it was necessary to
37
know product price. But product price was not available for the households who did not sell the product.
Moreover some households did not report price, particularly when the product was not a regular selling
item. In such cases the zone level average product specific price was used. In the cases that no household
in the zone reported price, the average of other zones is used. Future estimations of socio-economic
indicators will be done using R-scripts.
4.1 Indicator 1: Quantity of each of the primary tree and forest products collected
Core Indicator: Average quantity of each of the primary tree and forest products collected by a
HH
Unit: Quantity/HH/year
∑𝑛𝑖(𝑞𝑖𝑘𝑗 𝑤𝑗 )
𝑞𝑘𝑗 = (54𝑎)
∑𝑛𝑖 𝑤𝑗
where,
𝑞𝑘𝑗 = average quantity of the 𝑘th primary tree and forest product annually collected from
zone 𝑗 (Quantity/HH/year);
𝑞𝑖𝑘𝑗 = annual quantity of the 𝑘th product collected by the 𝑖th HH in zone 𝑗 (Quantity/year);
𝑛 = number of HHs surveyed in zone 𝑗;

𝑤𝑗 = 𝑤𝑢 + 𝑤ℎℎ weight assigned for the zone 𝑗. The Bootstrap weights for complex surveys is
calculated using the computer programme STATA 14 which is combination of the following
two components:
1) 𝑤𝑢 = the sample weight of the union selection within the zone = number unions in a zone
/ total number of unions surveyed in that zone
2) 𝑤ℎℎ = the sample weight of the HH selection = number of HH in the selected union / total
number of HHs surveyed in the union.
The national average was estimated as weighted sum of all the zones through the following
equation:
5
𝑁𝑗 𝑞𝑘𝑗
𝑞𝑘 = ∑ (54𝑏)
𝑁
𝑗
where,
𝑞𝑘 = quantity of the 𝑘th primary tree and forest product collected from all the zones
(Quantity/HH/year); and
𝑁 = total number of HHs in Bangladesh.
38
Derived Indicator: Total quantity of each of the primary tree and forest products collected by all
the HHs in a zone or nationally
Unit: Quantity/year
𝑡𝑞𝑘𝑗 = 𝑞𝑘𝑗 ∗ 𝑁𝑗 (54𝑐)
where,
𝑡𝑞𝑘𝑗 = total quantity of the 𝑘th primary tree and forest product annually collected from zone 𝑗
(Quantity/ year); and
𝑁𝑗 = total number of HHs in the zone 𝑗.
4.2 Indicator 2: Economic value of primary tree and forest products collected
Core indicator: Economic value of primary tree and forest products collected by a HH
Unit: BDT/HH/year
∑𝑛𝑖(𝑞𝑖𝑘𝑗 𝑝𝑖𝑘𝑗 𝑤𝑗 )
𝑎𝑣𝑗 = (55𝑎)
where,
𝑎𝑣𝑗 = average value of primary tree and forest products annually collected from zone 𝑗
(BDT/HH/year);
𝑞𝑖𝑘𝑗 = quantity of the 𝑘th product collected by the 𝑖th HH in zone 𝑗 per year (quantity/year);
𝑝𝑖𝑘𝑗 = price of the 𝑘th product for the 𝑖th HH in zone 𝑗 (BDT/unit);1
𝑛 = number of HHs surveyed in zone 𝑗; and

𝑤𝑗 = weight assigned for the zone 𝑗, which is same as described in Eq. 54a.
The national average was generated as the weighted sum of all the zones, similar to the Eq. 54b.
In similar way, average (BDT/HH/year) and total (million BDT/year) values of different primary
tree & forest products collected by households across zones were estimated.
Derived indicator: Total economic value of primary tree and forest products collected in zone 𝑗
Unit: BDT/year
𝑡𝑣𝑗 = 𝑎𝑣𝑗 ∗ 𝑁𝑗 (55𝑏)
1If the households did not sell the product or did not report price, the zone level average product specific price was used. In
extreme cases, where no household in the zone reported price, the average of other zones is used.
39
where,
𝑡𝑣𝑗 = total economic value of primary tree and forest products from zone 𝑗 (BDT/year); and
4.3 Indicator 3: Quantity of each of the collected primary tree and forest products sold
Core Indicator: Average quantity of each of the collected primary tree and forest products sold
by a HH
∑𝑛𝑖 ∑𝑚
𝑘 (𝑠𝑜𝑙𝑑𝑖𝑘𝑗 )
𝑠𝑜𝑙𝑑𝑘𝑗 = 𝑤𝑗 (55𝑎)
𝑛
where,
𝑠𝑜𝑙𝑑𝑘𝑗 = average quantity of the 𝑘th primary product annually sold in zone 𝑗
(Quantity/HH/year);
𝑠𝑜𝑙𝑑𝑖𝑘𝑗 = quantity of the 𝑘th primary product annually sold by the 𝑖th HH in zone 𝑗
(Quantity/year);
Derived Indicator: Total quantity of each of the collected primary tree and forest products sold
by the HHs
Unit: Quantity/year
𝑡𝑠𝑜𝑙𝑑𝑘𝑗 = 𝑠𝑜𝑙𝑑𝑘𝑗 ∗ 𝑁𝑗 (55𝑐)
where,
𝑡𝑠𝑜𝑙𝑑𝑘𝑗 = total quantity of the 𝑘th primary product sold in zone 𝑗 (Quantity/year); and
b) Share of each of the collected primary tree and forest products sold
Core Indicator: Share of each of the collected primary tree and forest products sold by a HH
Unit: %
𝑠𝑜𝑙𝑑𝑖𝑘𝑗
∑𝑛𝑖 ∑𝑚
𝑘 ( ⁄𝑞 )
𝑖𝑘𝑗
𝑠𝑘𝑗 = 𝑤𝑗 (55𝑑)
𝑛
40
where,
𝑠𝑘𝑗 = share of the 𝑘th tree and forest product sold in zone 𝑗 (%);
𝑠𝑜𝑙𝑑𝑖𝑘𝑗 = annual quantity of the 𝑘th product sold by the 𝑖th HH in zone 𝑗 (Quantity/year);2
𝑞𝑖𝑘𝑗 = annual quantity of the 𝑘th product collected by the 𝑖th HH in zone 𝑗 (Quantity/year);3

𝑤𝑗 = weight assigned for the zone 𝑗 which is same as described in Eq. 54a.
4.4 Indicator 4: Income from selling primary tree and forest products
Core indicator: Average annual income earned by a HH from selling primary tree and forest
products
Unit: BDT/HH/year
∑𝑛𝑖 ∑𝑚
𝑘 (𝑔_𝑖𝑛𝑐𝑜𝑚𝑒𝑖𝑘𝑗 − 𝑐𝑜𝑠𝑡𝑖𝑘𝑗 )𝑤𝑗
𝑎_𝑖𝑛𝑐𝑜𝑚𝑒𝑗 = 𝑤𝑗 (56𝑎)
where,
𝑎_𝑖𝑛𝑐𝑜𝑚𝑒𝑗 = average annual income earned by a HH from selling primary products in zone 𝑗
(BDT/HH/year);
𝑔_𝑖𝑛𝑐𝑜𝑚𝑒𝑖𝑘𝑗 = gross income earned by the 𝑖th HH from selling the 𝑘th primary product in
zone 𝑗 (BDT/year);
𝑐𝑜𝑠𝑡𝑖𝑘𝑗 = cost associated with the selling of 𝑘th primary tree and forest product for the 𝑖th HH
in zone 𝑗 (BDT/year);
In similar way, average (BDT/HH/year) and total (million BDT/year) income from different
primary tree and forest products collected by households across zones were estimated.
Derived indicator: Total annual income earned by all the HHs from selling primary tree and
forest products
2
Same as in Eq. 9.3.1
3
Same as in Eq. 54a
41
Unit: BDT/year
𝑡_𝑖𝑛𝑐𝑜𝑚𝑒𝑗 = 𝑎_𝑖𝑛𝑐𝑜𝑚𝑒𝑗 ∗ 𝑁𝑗 (56𝑏)
where,
𝑡_𝑖𝑛𝑐𝑜𝑚𝑒𝑗 = total annual income earned by all the HHs in zone 𝑗 from selling primary tree
and forest products (BDT/year); and
𝑁 = total number of HHs in the zone 𝑗.
4.5 Indicator 5: Quantity of the processed tree and forest products
Core Indicator: Average quantity of each of the processed tree and forest products produced
∑𝑛𝑖(𝑞𝑖𝑘𝑗 𝑤𝑗 )
𝑞𝑘𝑗 = (57𝑎)
where,
𝑞𝑘𝑗 = average quantity of the 𝑘th processed tree and forest product produced in zone 𝑗
(Quantity/HH/year);
𝑞𝑖𝑘𝑗 = quantity of the 𝑘th processed product annually produced by the 𝑖th HH in zone 𝑗
(Quantity/year);
The national level average was generated as the weighted sum of all the zones, similar to the Eq.
54b.
Derived Indicator: Total quantity of each of the processed tree and forest products produced by
all the HHs in a zone or nationally
Unit: Quantity/year
𝑡𝑞𝑘𝑗 = 𝑞𝑘𝑗 ∗ 𝑁𝑗 (57𝑏)
where,
𝑡𝑞𝑘𝑗 = total quantity of the 𝑘th processed tree and forest product produced in zone 𝑗
(Quantity/ year); and
42
4.6 Indicator 6: Quantity of each of the processed tree and forest products sold
Core Indicator: Average quantity of each of the processed tree and forest products sold by a HH
∑𝑛𝑖(𝑠𝑖𝑘𝑗 𝑤𝑗 )
𝑠𝑘𝑗 = (58𝑎)
where,
𝑠𝑘𝑗 = average quantity of the 𝑘th processed tree and forest product sold in zone 𝑗
(Quantity/HH/year);
𝑠𝑖𝑘𝑗 = quantity of the 𝑘th processed product annually sold by the 𝑖th HH in zone 𝑗
(Quantity/year);
Derived Indicator: Total quantity of each of the processed tree and forest products sold
Unit: Quantity/year
𝑡𝑠𝑘𝑗 = 𝑠𝑘𝑗 ∗ 𝑁𝑗 (58𝑏)
where,
𝑡𝑠𝑘𝑗 = total quantity of the 𝑘th processed tree and forest product sold in zone 𝑗
(Quantity/year); and
4.7 Indicator 7: Income from selling processed tree and forest products
Core indicator: Average income from selling processed tree and forest products
Unit: BDT/HH/year
∑𝑛𝑖 ∑𝑚
𝑘 (𝑔_𝑖𝑛𝑐𝑜𝑚𝑒𝑖𝑘𝑗 − 𝑐𝑜𝑠𝑡𝑖𝑘𝑗 )𝑤𝑗
𝑎_𝑖𝑛𝑐𝑜𝑚𝑒𝑗 = (59𝑎)
where,
𝑎_𝑖𝑛𝑐𝑜𝑚𝑒𝑗 = average income from selling processed tree and forest products in zone 𝑗
(BDT/HH/year);
43
𝑔_𝑖𝑛𝑐𝑜𝑚𝑒𝑖𝑘𝑗 = gross income earned by the 𝑖th HH from selling the 𝑘th processed tree and
forest product in zone 𝑗 (BDT/year);
𝑐𝑜𝑠𝑡𝑖𝑘𝑗 = cost (e.g. labour, raw materials, transportation, selling, etc.) associated with the
selling of 𝑘th processed tree and forest product for the 𝑖th HH in zone 𝑗 (BDT/year);
Derived indicator: Total income from selling processed tree and forest products
Unit: BDT/year
where,
𝑡_𝑖𝑛𝑐𝑜𝑚𝑒𝑗 = total income earned in zone 𝑗 from selling processed tree and forest products
(BDT/year); and
4.8 Indicator 8: Involvement with tree and forest related activities
a) Number of family members involved

Core indicator: Average number of household members involved in different tree and forest
related activities
Unit: no/HH/year
∑𝑛𝑖 𝑛𝑜𝑖𝑘𝑗 𝑤𝑗
𝑛𝑜𝑗𝑘 = (60𝑎)
where,
𝑛𝑜𝑗𝑘 = average number of family members involved with 𝑘th activity (e.g. collection,
processing, selling) in zone 𝑗 (no/year);
𝑛𝑜𝑖𝑘𝑗 = number of family members involved with 𝑘th activity in the 𝑖th HH in zone 𝑗
(no/HH/year);
44
Derived indicator: Total number of household members involved in different tree and forest
related activities
Unit: no/year
𝑡𝑛𝑜𝑘 = 𝑛𝑜𝑗𝑘 ∗ 𝑁𝑗 (60𝑏)
where,
𝑡𝑛𝑜𝑘 = total number of HH members involved with 𝑘th activity in zone 𝑗 (no/year); and
4.9 Indicator 9: Proportion of households receiving different tree and forest related
services
Core Indicator: Proportion of households receiving services

Unit: %
𝑛 𝑚
ℎ
𝑠ℎ𝑘𝑗 = ∑ ∑ ( 𝑖𝑘𝑗⁄𝑛) ∗ 𝑤𝑗 ∗ 100 (61)
𝑖 𝑘
where,
𝑠ℎ𝑘𝑗 = proportion of the households in zone 𝑗 receiving 𝑘th service (%);
ℎ𝑖𝑘𝑗 = total number of HHs in zone 𝑗 receiving 𝑘th service (no);

4.10 Indicator 10: Value of collected tree and forest products used for cooking and heating
Core indicator: Average value of collected tree and forest products used for cooking and heating
Unit: BDT/HH/year
∑𝑛𝑖(𝑞𝑖𝑘𝑗 𝑝𝑖𝑘𝑗 )𝑤𝑗
𝑎𝑣𝑗 = (62𝑎)
where,
𝑎𝑣𝑗 = average value of collected cooking and heating products used for cooking by a HH in
zone 𝑗 (BDT/HH/year);
𝑞𝑖𝑘𝑗 = quantity of the 𝑘th product (i.e. firewood and leaves) collected by the 𝑖th HH in zone 𝑗
per year (kg/year);
45
𝑝𝑖𝑘𝑗 = price of the 𝑘th product for the 𝑖th HH in zone 𝑗 (BDT/kg);4

Derived indicator: Total value of collected tree and forest products used for cooking and heating
Unit: BDT/year
where,
𝑡𝑣𝑗 = total value of tree and forest products used for cooking and heating in zone 𝑗
(BDT/year); and
4.11 Indicator 11: Cost of buying tree and forest products used for cooking and heating
Core indicator: Average cost of purchasing tree and forest products used for cooking and heating
Unit: BDT/HH/year
∑𝑛𝑖(𝑐𝑖𝑘𝑗 ∗ 12)𝑤𝑗
𝑎𝑐𝑗 = (63𝑎)
where,
𝑎𝑐𝑗 = average annual cost of purchasing energy products used for cooking and heating in zone
𝑗 (BDT/HH/year);
𝑐𝑖𝑘𝑗 = monthly cost of purchasing the 𝑘th energy product by the 𝑖th HH in zone 𝑗
(BDT/month);
Derived indicator: Total cost of purchasing tree and forest products used for cooking and heating
Unit: BDT/year
𝑡𝑐𝑗 = 𝑎𝑐𝑗 ∗ 𝑁𝑗 (63𝑏)
4 If the households did not report price, the zone level average product specific price was used.
46
where,
𝑡𝑐𝑗 = total cost of purchasing energy products used for cooking and heating in zone 𝑗
(BDT/year); and
4.12 Indicator 12: Presumed amount of purchased tree and forest products used for
energy
Core indicator: Average presumed amount of purchased tree and forest products used for energy
Unit: kg/HH/year
∑𝑛𝑖(𝑞𝑖𝑘𝑗 −𝑝𝑖𝑘𝑗 )𝑤𝑗
𝑎𝑣𝑗 = (64𝑎)
where,
𝑎𝑣𝑗 = average presumed amount of purchased tree and forest products used for energy in zone
𝑗 (kg/HH/year);
𝑞𝑖𝑘𝑗 = annual amount of the 𝑘th energy product (i.e. firewood and leaves) collected by the 𝑖th
HH in zone 𝑗 (kg/year);
𝑝𝑖𝑘𝑗 = annual quantity of the 𝑘th product (i.e. fuelwood and leaves) purchased by the 𝑖th HH
in zone 𝑗 (BDT/kg);
𝑛 = number of the HHs surveyed in zone 𝑗; and
Derived indicator: Total presumed amount of purchased tree and forest products used for energy
Unit: t/year
where,
𝑡𝑣𝑗 = total presumed amount of purchased tree and forest products used for energy in zone 𝑗
(t/year); and
4.13 Indicator 13: Total annual income from tree and forest
Core indicator: Average annual income from tree and forest
47
Unit: BDT/HH/year
∑𝑛𝑖 𝑓_𝑖𝑛𝑐𝑜𝑚𝑒𝑖𝑘𝑗 𝑤𝑗
𝑎_𝑖𝑛𝑐𝑜𝑚𝑒𝑗 = (65𝑎)
where,
𝑎_𝑖𝑛𝑐𝑜𝑚𝑒𝑗 = average annual income from tree and forest in zone 𝑗 (BDT/HH/year);
𝑓_𝑖𝑛𝑐𝑜𝑚𝑒𝑖𝑘𝑗 = net income (i.e. cost adjusted) earned by the 𝑖th HH from 𝑘th forest related
activities (i.e. income from primary and processed products, salary) in zone 𝑗 (BDT/year);
Derived indicator: Total annual income from tree and forest
Unit: BDT/year
where,
𝑡_𝑖𝑛𝑐𝑜𝑚𝑒𝑗 = total annual income earned by all the HHs in zone 𝑗 from tree and forest
(BDT/year); and
b) Percentage of total income received by HHs from tree and forest
Core indicator: Percentage of total annual income received by HHs from tree and forest
Unit: %
𝑓_𝑖𝑛𝑐𝑜𝑚𝑒𝑖𝑗
∑𝑛𝑖 ∑𝑚
𝑘 ( ⁄𝑡_𝑖𝑛𝑐𝑜𝑚𝑒 )
𝑖𝑘𝑗
𝑠𝑘𝑗 = 𝑤𝑗 (65𝑐)
𝑛
where,
𝑠𝑘𝑗 = share of 𝑖th HH’s total annual income earned from tree and forest in zone 𝑗 (%);
𝑓_𝑖𝑛𝑐𝑜𝑚𝑒𝑖𝑘𝑗 = total annual income earned by the 𝑖th HH in zone 𝑗 from forest
(BDT/HH/year);5
𝑡_𝑖𝑛𝑐𝑜𝑚𝑒𝑖𝑘𝑗 = total annual income for the 𝑖th HH in zone 𝑗 (BDT/HH/year);
5
Same as estimated in Eq. 9.13.1.
48
5 References
Uncategorized References
Akhter, M., Hossain, M., & Birigazzi, L. (2013). Tree volume and biomass allometric equations of
Bangladesh. Retrieved from
Allen, S. (1986). Chemical analysis. Methods in plant ecology, 285-344.
BBS. (2016). Statistical Yearbook of Bangladesh 2015. Retrieved from
BFD. (2016). Field Instructions for the Bangladesh Forest Inventory (Version 1.2). : Bangladesh Forest
BFD. (2016b). Quality assurance and quality control for the Bangladesh Forest Inventory. Dhaka,
Bangladesh: Bangladesh Forest Department, Food and Agriculture Organization of the United
Nations.
BFD. (2016e). The Bangladesh forest inventory design: Methodological approach (pp. 74). Dhaka:
Bangladesh Forest Department, Food and Agriculture Organization of the United Nations.
BFD. (2016f). Zoning for tree and forest assessment in Bangladesh. Retrieved from Dhaka:
BFD. (2017). National Land Representation in Bangladesh. Dhaka: Bangladesh Forest Department, Food
and Agriculture Organization of the United Nations.
BFD. (2017b). Quality assurance and quality control for the socio-economic component of the
Bangladesh Forest Inventory (pp. 20). Dhaka: Bangladesh Fores Department, Food and
Agriculture Organization of the United Nations.
Bijaya, D. G. C., & Bhandari, J. (2008). Carbon Sequestration Potential and uses of Dendrocalamus
strictus. IUFRO World Series Volume 27, 62.
Cochran, W. G. (1977). Sampling Techniques (thrid edition ed.). New York: John Willey & Sons.
Conley, A., & Moote, M. A. (2003). Evaluating collaborative natural resourcemanagement. Society
&Natural Resources, 16(5), 371-386.
Curtis, A., Byron, I., & MacKay, J. (2005). Integrating Socio‐Economic and Biophysical Data to Underpin
Collaborative Watershed Management. JAWRA Journal of the American Water Resources
Association, 41(3), 549-563.
De Lange, W. J., Wise, R., Forsyth, G., & Nahman, A. (2010). Integrating socio-economic and biophysical
data to support water allocations within river basins: An example from the Inkomati Water
Management Area in South Africa. Environmental Modelling & Software, 25(1), 43-50.
de Melo, L. C., Sanquetta, C. R., Dalla Corte, A. P., & Mognon, F. (2015). Methodological alternatives in
the estimate of biomass for young individuals of bambusa genus. Bioscience Journal, 31(3).
Donato, D. C., Kauffman, J. B., & Stidham, M. (2009). Protocols for Measuring & Reporting Carbon Stocks
in Mangrove Forests, With Special Reference to Carbon Assessment for Sundarbans Reserve
Forest, Bangladesh. Retrieved from
FAO. (2018a). Guidelines and specifications FRA 2020 Version 1.0. Retrieved from Rome, Italy:
FAO. (2018b). Global Forest Resources Assessment 2020. Terms and definitions. Rome, Italy: FAO
forestry department.
GoB. (2017a). Criteria and indicators of Bangladesh Forest Inventory: integration of socio-economic and
biophysical information. Retrieved from
GOB. (2017b). The socio-economic survey of the Bangladesh Forest Inventory. Retrieved from
49
Hairiah, K., Sitompul, S., van Noordwijk, M., & Palm, C. (2001). Methods for sampling carbon stocks
above and below ground: ICRAF Bogoi.
Harmon, M. E., Woodall, C. W., Fasth, B., Sexton, J., & Yatkov, M. (2011). Differences between standing
and downed dead tree wood density reduction factors: a comparison across decay classes and
tree species. Res. Pap. NRS-15. Newtown Square, PA: US Department of Agriculture, Forest
Service, Northern Research Station. 40 p., 15, 1-40.
Hjelm, B. (2015). Empirical models for estimating volume and biomass of poplars on farmland in Sweden
(Vol. 2015): Department of Crop Production Ecology, Swedish University of Agricultural ….
Hossain, M. (2016). Improved National Tree Allometric Equation Database to Support Forest Monitoring
and Assessment of Bangladesh. Retrieved from Dhaka:
Hossain, M., Siddique, M., & Akhter, M. (2016, 22-25 February). A critical review and database of
biomass and volume allometric equation for trees and shrubs of Bangladesh. Paper presented at
the International Conference on Water Resources and Environment (WRE2016), Shanhgai.
Hossain, M. A. (2017). R-script for Quality Assurance and Quality Checking of Bangladesh Forest
Inventory Soil and Litter Data (R-script). Retrieved 15 June 2019, from Food and Agriculture
Organization of the United Nations http://bfis.bforest.gov.bd/library/r-script-for-bangladesh-
forest-inventory-data-analysis/
Hossain, M. A., Laurent, S., & Birigazzi, L. (2017). R-script for Bangladesh Forest Inventory Data Analysis
[Rscript]. Dhaka, Bangladesh: Food and Agriculture Organization of the United Nations.
Hossain, M. A., Laurent, S., Sola, G., Birigazzi, L., & Aziz, T. (2017). R-script for Bangladesh Forest
Inventory Data Quality Assurance and Quality Checking [R-script]. Dhaka, Bangladesh: Food and
Agriculture Organization of the United Nations.
Iqbal, Z., Kuegler, O., Saket, M., Rahman, L., Siddiqui, R., Scott, C., . . . Henry, M. (2016). The Bangladesh
Forest Inventory Design: Methodological Approach. Retrieved from Dhaka:
Kauffman, J. B., Cummings, D., Ward, D., & Babbitt, R. (1995). Fire in the Brazilian Amazon: 1. Biomass,
nutrient pools, and losses in slashed primary forests. Oecologia, 104(4), 397-408.
Komiyama, A., Ong, J. E., & Poungparn, S. (2008). Allometry, biomass, and productivity of mangrove
forests: A review. Aquatic Botany, 89(2), 128-137.
Korhonen, K. T., & Salmensuu, O. (2014). Formulas for estimators and their variances in NFI. [An internal
paper for the Forest and Agricultural Organization of the United States].
Kumar, M. F., Costello, L., Mahamud, R., Henry, M., & Johnson, K. (2017). Bangladesh Forest Inventory
Data Management Protocol. Dhaka: Bangladesh Forest Department, Food and Agriculture
Kumar, M. F., Costello, L., Mahamud, R., Henry, M., Johnson, K. (2017). Bangladesh Forest Inventory
Data Management Protocol.
Laurent, S., & Hossain, M. A. (2017). R-script for Data Corrections of Bangladesh Forest Inventory [R-
script]. Dhaka, Bangladesh: Food and Agriculture Organization of the United Nations.
Lobovikov, M., Ball, L., Guardia, M., & Russo, L. (2007). World bamboo resources: a thematic study
prepared in the framework of the global forest resources assessment 2005: Food & Agriculture
Org.
Lockwood, A. C., Michael. (2000). Landcare and catchment management in Australia: lessons for state-
sponsored community participation. Society & Natural Resources, 13(1), 61-73.
Margalef, R. (1958). Information theory in biology. General Systems Yearbook, 3, 36-71.
Marshall, P. L., Davis, G., & LeMay, V. M. (2000). Using line intersect sampling for coarse woody debris:
Vancouver Forest Region.
Matthews, E. (1997). Global litter production, pools, and turnover times: Estimates from measurement
data and regression models. Journal of Geophysical Research: Atmospheres, 102(D15), 18771-
18800.
50
Pearson, T. R., Brown, S. L., & Birdsey, R. A. (2007). Measurement guidelines for the sequestration of
forest carbon. Gen. Tech. Rep. NRS-18. Newtown Square, PA: US Department of Agriculture,
Forest Service, Northern Research Station. 42 p., 18.
Potapov, P., Siddiqui, B., Iqbal, Z., Aziz, T., Zzaman, B., Islam, A., . . . Turubanova, S. (2017).
Comprehensive monitoring of Bangladesh tree cover inside and outside of forests, 2000–2014.
Environmental Research Letters, 12(10), 104015.
Rahman, M., & Jashimuddin, M. (2017). Field Test Report on Socioeconomic Survey. Retrieved from
Scott, C. (2018). Estimation Using Ratio-to-Size Estimator Across Strata and Subpopulations.
Shannon, C. E. a. W., W. . (1963). The Mathematical Theory of Communities. University of Illinois Press,
Urbana, pp. 111-117).
Simpson, E. H. (1949). Measurement of diversity. Nature, 163-688).
Sohel, M. S. I., Alamgirb, M., Akhter, S., & Rahman, M. (2015). Carbon storage in a bamboo (Bambusa
vulgaris) plantation in thedegraded tropical forests: Implications for policy development. Land
Use Policy, 49, 142–151.
Stokes, A., Lucas, A., & Jouneau, L. (2007). Plant biomechanical strategies in response to frequent
disturbance: Uprooting of Phyllostachys nidularia (Poaceae) growing on landslide‐prone slopes
in Sichuan, China. American Journal of Botany, 94(7), 1129-1136.
51
View publication stats

BFI Analysis Document Final

Uploaded by

Copyright:

Available Formats

BFI Analysis Document Final

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

BFI Analysis Document Final

Uploaded by

Copyright:

Available Formats

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

Estimation Procedures of Indicators and Variables of the Bangladesh Forest

Technical Report · June 2019

Md. Akhter Hossain Nikhil Chakma

SEE PROFILE SEE PROFILE

Matieu Henry Rashed Jalal

SEE PROFILE SEE PROFILE

The user has requested enhancement of the downloaded file.

Chakma, N. (2016). Proceedings of the Information sharing meeting on Bangladesh Forest

Figure 2. The distribution of plots according to the accessibility status.

2.1.1.1 Data management and cleaning

Tree forms Number of equations based on their nature

Volume Green Oven-dried Air-dried Carbon Nutrients Length of

Tree 360 138 78 44 11 10 0 0 25 1 3 3 0 0

Total 360 138 83 44 11 10 3 0 25 1 3 3 1 0

2.1.1.3 Development of new allometric models

SN List of manuals, protocols and training proceedings on allometric equation development

2.2 Socio-economic survey

Chakma, N. (2017). Proceeding of National Consultation on Socioeconomic Survey Design of

2.3 Integration of biophysical, socio-economic and remote sensing data

3 Statistical methodologies for biophysical data analysis

3.2 Area estimations

BFI Land Cover FRA

3.2.2 Surveyed area estimations

3.3 Growing stock estimation

SN Species Volume models R2 n DBH range

3.3.1.2 Stump volume

3.3.1.3 Bamboo volume

3.3.2 Dead wood volume

3.3.2.1 Standing dead tree

3.3.2.3 Stump (dead) volume

3.4 Biomass estimation

3.4.1 Above ground biomass

Here, 𝑌𝑎𝑔𝑏 = Above ground biomass,

𝑌𝑇𝑆𝑎𝑔𝑏 = Tree and Sapling AGB,

𝑌𝐵𝑎𝑔𝑏 = Bamboo AGB, and

𝑌𝑆𝑎𝑔𝑏 = Stump (live) AGB.

3.4.1.1 Tree and sapling above ground biomass

8 Sal zone ln 𝑌𝑎𝑔𝑏 = −2.46 + 2.17 ln 𝐷 + 0.367 ln 𝐻 + 0.161 ln 𝜌 59 6-31

9 Sundarbans zone ln 𝑌𝑎𝑔𝑏 = −1.956299 + 2.163361 ln 𝐷 0.9852 82 2-78

10 Village zone ln 𝑌𝑎𝑔𝑏 = −6.0325 + 1.9715 ln 𝐷 + 0.8193 ln 𝜌 0.9455 817

3.4.1.2 Bamboo above ground biomass

𝑌𝐵𝑎𝑔𝑏 = 11.403 + 0.0006(𝐷 2 𝐻)-----------------------------Equation 9

𝑌𝐵𝑎𝑔𝑏 = 0.67545 + 0.02813(𝐷2 𝐻)-----------------------------Equation 10

3.4.1.3 Stump above ground biomass

3.4.2 Below ground biomass

3.4.2.1 Tree and sapling below ground biomass

𝑌𝑇𝑆𝑏𝑔𝑏 = 𝑒𝑥𝑝[−1.0587 + 0.8836 𝑙𝑛(𝑌𝑇𝑆𝑎𝑔𝑏 )] ---------------------------------- Equation 13

𝑌𝑇𝑆𝑏𝑔𝑏 = 0.199 × 𝜌0.899 × (𝑑𝑏ℎ)2.22 ---------------------- Equation 14

3.4.2.2 Bamboo below ground biomass

𝑌𝐵𝑏𝑔𝑏 = 𝑌𝐵𝑎𝑔𝑏 × 0.26------------------------Equation 16

3.4.2.3 Stump (live) below ground biomass

3.4.3 Dead wood biomass

3.4.3.1 Standing dead tree biomass

Decay class WD reduction factor (𝜌𝑟𝑓 ) Composition

𝜌𝑟 = 𝜌𝑟𝑓 × 𝜌 -------------------------- Equation 19

3.4.3.2 Biomass of Down Woody Material

𝑌𝐷𝑊𝑀𝑏 = 𝑌𝐶𝑊𝐷𝑏 + 𝑌𝐹𝑊𝐷𝑏 ---------------------------- Equation 20

Type Decay class Downed dead wood N

3.4.3.3 Stump (dead) biomass