# Generating A Database Of Regional Biomass Models Biology Essay

**Published:** **Last Edited:**

This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.

Carbon is the major component as building organic material for living things. Carbon is retained in earth and ocean as well in the atmosphere. Higher plants gained carbon during the photosynthesis process as a result of biological carbon sink. The establishment of biological carbon sink is required by the forest ecosystem as well by other vegetation (Grace, 2004). Forest retained the biomass carbon for 50% from other vegetation (Brown, 1997, Gibbs, Brown, Niles, & Foley, 2007) therefore, it is important to understand the rate of carbon accumulation that is reserve in this case by the forest ecosystem that has a connection with the climate change (Makila, Saarnisto, & Kankainen, 2001). One of the functions of forest is as storage and sequestration of carbon as well as maintaining the green house effect (Yuniawati, Budiaman, & Elias, 2011). Thus, role of tropical forest of a biogeochemical cycle that promotes the carbon cycle mostly is based on the estimation of individual standing trees. Tropical forest itself based on (Chave, et al., 2005) stores large amount of carbon. The outcome of deforestation and degraded forest are releasing their carbon sequester into carbon dioxide (Gibbs, Brown, Niles, & Foley, 2007).

During the 1990s, the tropical forest deforestation event that includes the land use transformation is estimated to have released of 15-35% annual fossil fuel or 1-2PgC/year; furthermore, the prediction tempt to increase 85 to 130 PgC over the next 100 years (Moutinho & Schawrtzman, 2005).

The total amount of carbon is called biomass that contain in an individual tree and organic matter found on the forest floor. The advantage of biomass observation is to identify the ecosystem structure with the source of biomass, and to identify the forest productivity (Zianis, 2008).

In the tropical forest carbon pools are living biomass of tress, include the under storey vegetation, dead mass of litter, woody debris, and soil organic matter. The largest pool is found in the aboveground where trees influence the amount of the carbon stored. Thus, estimating the aboveground biomass as well carbon stock is important in tropical forest.

To quantify the carbon stored in aboveground forest is by harvesting the trees, this destructive sampling, and measure the dry-weight. Later on, the dry weighted tree biomass is converted to carbon content, 50 % of biomass will show the amount of carbon. Such method is considered expensive and time consuming. However, there is no direct method for measuring the carbon stock, by developing a model from destructive data that can estimate an area (Gibbs, Brown, Niles, & Foley, 2007).

Estimating biomass above ground it is such a method to explore the C stock and C sequestration in deforestation and transformation area (Ketterings, Coe, van Noordwijk, Ambagau, & Palm, 2001). The gained amount of carbon that emitted is triggering the awareness globally and special attention of the UN framework Convention on Climate Change (UNFCCC) as well has mentioned in the Kyoto Protocol (Somogyi, Cienciala, Makipaa, Muukkonen, Lechtonen, & Weiss, 2007).

## Biomass estimation and Assessment

Biomass is the renewable energy resources. It is derived from animals and plants include forest product, agriculture crops, aquatic plants, and human. Brown (1997) defined that the total of the above-and belowground living organic matter in trees as oven-dry tons per unit are as biomass.

Tree biomass is the amount of carbon fixed from the process of photosynthesis and minus the loss during respiration (Johnsen, Samuelson, Teskey, McNulty, & Fox, 2001) and for the total for forest biomass itself is over the reduction of the harvesting process (Brown, 1997).

Estimation of a land cover area biomass it is necessarily important for information of carbon stock (sequestration). The highest carbon stock and content of biomass are highly found in the tropical area (Ponce-Hernandez, Koohafkan, & Antoine, 2004).

The assessment of biomass estimation is important in order to describe forest structure ecosystem, and informed of forest productivity (Zianis, 2008). Others application of biomass estimation is important for countries especially tropic countries for reporting status of forest resources and carbon stock to the United Nations Framework Convention on Climate Change (UNFCCC) (UNFCCC, 2008).

Based on Brown (1997) the first approach is by the existing data of volume measured in the fields and the second by using the mathematical equation by combining the dry weight per tree as a single function with the variable of the tree also named as allometric equation. Allometric equation is a method to estimate biomass in a forest stand. According to (Ketterings, Coe, van Noordwijk, Ambagau, & Palm, 2001)allometricg the biomass with allometric equation is the correlation of tree diameter at breast height or any variable that is used in forest inventory (Zianis, 2008). However, allometric equation is restricted to species specific or site specific; therefore, there are difficulties to integrate site specific heterogeneously collected data to draw regional-scale conclusions about biomass stocks (Chave, et al., 2005). Therefore, in order constructing the biomass equation based on catering, et al (2001) some qualification should be considered such as the form of the model equation, determining the parameter in equation, input of the variables, and using the specific allometric equation to later on extrapolate into such area observation.

Technical approaches to estimate biomass in tropical forest. (1) Destructive sampling in situ; (2) Non-destructive sampling based forest inventory; (3) remote sensing assessment; (4) model construction.

Destructive sampling

The practical measuring biomass is by a destructive method and measuring the dry weight of each part of a tree. The accuracy of the biomass estimation itself has a big role in the context of the establishing growth and carbon sequestration of the forest in a period of time (Basuki, van Laake, Skidmore, & Hussin, 2009).

Non-destructive sampling based on forest inventory

The aboveground biomass estimation and carbon stock change for a large forest area is based on forest inventory, which require variables such as diameter at breast high, tree-height, and volume of three per total area. Parameter of volume of three per area is used to estimate aboveground of stand biomass using the Biomass expansion factor (BEF) (Somogyi Z. , et al , 2007).

However, most of the national forest inventory is based on the commercial tree species.

The United Nation Food and Agriculture Organization (FAO) have compiled some nation's forest inventory especially for commercial timber resources. The inventory present data such stand a table that is used for harvesting the timber. Thus, for biomass estimation such data compilation of commercial timber is not required to estimate the aboveground biomass or carbon sequestrate. The compilation of stand table of commercial timber is based on tree with diameter > 35 centimeter, while, for aboveground biomass estimation it requires certain range diameter from the minimum until certain value of the diameter to be measured (Brown, Gillespie, & Lugo, 1989; 1992)

Remote sensing approach

Estimation of biomass using the remote sensing approach is one of the important sources for analyzing. The approaching using the remote sensing is by providing such method to analyze spatial distributional local forest biomass into regional information (Zhang & Kondragunta, 2006). Forest biomass data are analyzed with two kinds of approaches, direct approach, and indirect relationship approach. The direct relation approaches are using the multiple regression analysis, k-nearest neighbor (Fehrmann, Lehtonen, Kleinn, & Tomppo, 2008), and neural networks. In other hand, indirect approaches are leaf area index (LAI), structure of the crown and height, and shadow fraction (Wulder, White, Fournier, Luther, & Magnussen, 2008).

Model construction

Model construction here means allometric equation that usually specific to certain site observation and species specific. However, the use of allometric equation will be needed in different type of approaches. Since, the allometric equation is referring to a diameter, and dry weight of every part of a single tree. It establishes a precise estimation of biomass estimation. The equation itself constructs with variable of tree diameter, tree-height, dry weight, and wood density.

In the tropical forest with the highest number of species is adjusted by using the allometric equation and mixed species tree regression to estimate the aboveground biomass (Chave, et al., 2005).

## Biomass Equation (regression model)

Many studied in developing biomass equations are related with the variable of diameter at breast height, tree-height, and dry biomass from destructively sample (Basuki, et al., 2009; Chave, et al., 2005; Ketterings, et al, 2001; and Zianis, 2008). The use of allometric equation is an accurate approach to estimate aboveground biomass. In tropical forest yet is challenging, estimating of such area in tropical forest with high variety species. The use of allometric equation for mixed species is suitable to use as estimation (Chave, et al., 2005).

The most common equation to construct the allometric equation is

(1)

Where Y= tree biomass (kilogram)

X= diameter of 1.30 meters

a, b= coefficient parameter

e= error term

This equation is known a power function (Brown, 1997).

The variances of tree-diameter and tree-height are based on the site condition, tree species, and technical measurement. In order to reduce heteroscedasticity of data while constructing the equation, logarithmic transformation is the potential approach (Brown, et al, 1989 & LaBarbera, 1989). The construction of biomass based on Brown, et al (1989) equation is using the nonlinear regressions, with common form

(2)

In addition, both sides are using the natural logarithm (Brown, et al, 1989).

The use of the equation (2) is by linearizing the equation with the function of natural logarithm for both variables. Establishing the linear equation is by performing de-transformed. However, based on Miller (1984) after the transformation the linear model takes place on variables will give a result of biased in the model. Once the linear equation is constructed, it is important to validate to a set of data and statistically tested the error.

Some published allometric equations are available. Although the allometric equation is an accurate approach for estimating biomass or carbon, only few available equations are published especially in tropical forest. Nevertheless, Brown (1997) established some allometric equation for tropical area based on the Brazilian Amazonian forest. Kettering, et al (2001) established allometric equation based on research in the mixed Sumatra forest

## Errors in Biomass Equations

Errors in tree biomass regression occurred in the field and during data processing. Based on Cunia T (year) there are four major sources of error in tree biomass regression functions, which are:

Selection of sample trees.

For estimating the same set of tree population, it is required selecting the same procedure to apply.

Measurement of the sample trees.

An error that occurs in the data field collection such as measuring of DBH, and height. The errors could be triggered by the error device and human error during measurement.

Statistical model

The model that is applied to estimate the biomass should be synchronized both sample and population due to the validity of the use of the model.

Application of the biomass regression

The error of the component is important when applying to forest population that is differ from the estimated. This argument is mainly because biomass regression never applied to the true population of the forest because the dynamic in the forest itself is always changing.

## Problem Statement

The biomass estimation assessment in a forest and transformed vegetation requires accuracy. A method such as biomass equation and remote sensing-based estimation are requiring extensive process. Developing a biomass equation with such method of destructive sampling is time consuming and expensive. However, once the equation is established it able to estimate forest biomass. The biomass equations are restricted to species specific and site specific (Chave, et al., 2005). There are published biomass equations for forest tropical region (Brown, 1997; Ketterings, Coe, van Noordwijk, Ambagau, & Palm, 2001). However, to apply one of the published equations to estimate biomass it is extremely requires validation.

Land use type and geographical location will affect on biomass estimation of a certain area. Moreover, species specific, a sample size of tree harvested, and diameter range are variables for constructing the equation is limited. Hence, the validation before applying biomass equation is needed.

## Research Hypotheses

A data set of 144 trees of destructive sample is compiled from scattered sources in Sumatra forest will generate a certain allometric equation that fit with the area of observation to estimate biomass.

We use this database to test the generality of simple models, and ask where common allometric patterns can be found for trees grown in different environments.

We test the assumption that a single pan-tropical allometry can be used in AGB estimation procedures.

Specifically, we ask to what extent the observed differences between site-derived allometries are due to the limited sample size used to construct the allometry.

## Objective

The main objective of the research is to compile a relevant base of allometric biomass models in a comprehensive database that is the basis for further research on carbon density for different forest-or land use types in the designated study areas.

To compile a set of relevant biomass models in a comprehensive database and to develop a general model suitable with study site in Jambi.

The database contains the model formulation, model coefficients, and all available information on the model quality (standard errors).

## Methodology

## Site Information

## Forest in Sumatra

Sumatra is dominated by the mountainous area called the Bukit Barisan, spread along for 1.700 km. The characteristic itself makes Sumatera is the second largest archipelago after Borneo (Laumonier, Uryu, Stuwe, Budiman, Setiabudi, & Hadian, 2010). The large area of Sumatera provides the opportunity for the migration program by government of Indonesia also derived by the spontaneous reaction from the public. The migration areas are along Jambi and Lampung. The large number of migrants is increasing the recruitment of the area for living, most of the migrants move into forested areas. To support their living, most of the people do the agriculture systems, which are oil palm and rubber-agro forestry. The system of slash and burn is the particular method for clearing the land and it has contributes to clearing forest areas (Partohardjono, Pasaribu, & Fagi, 2005).

Jambi is a province that is located in the eastern part of Sumatra, with an area of 49.578kmÂ² and is sparsely populated relative to the rest of Sumatra. Based on data of 1990, total area of Jambi Province is 53.436 square kilometers with the population of 2.018.463 inhabitants.

Total district in Jambi is Kerinci, Bungo-Tebo, Sarolangun-Bangko,Batang Hari, Tanjung Jabung and Kotamadya Jambi. Overview the national park in Jambi, which is mostly in this area, is still forested. Jambi has 4 areas which are designated as a national park; these are Taman Nasional Berbak (TNB), Taman Nasional Kerinci Sebelat (TNKS), Taman Nasional Bukit Dua Belas (TNBD) and Taman Nasional Bukit Tigapuluh (TNBT). Nevertheless, although it has the four largest national's parks in Sumatra, deforestation in Jambi has reached a total area of 459,856.67 ha (Kementerian Kehutanan Republik Indonesia, 2010). The industrial sectors responsible for large area land use changes (1) are palm oil plantations, (2) rubber plantation and (3) Acacia plantations.

For an overall evaluation of the effects of this massive land use change the impact on biodiversity and functional biodiversity on landscape scale but also the effects on other ecosystem services, and carbon sequestration is of interest.

The forest of Sumatra caught on fire during period of the year 1992-1993, the cause of the fire mainly set by the climatic condition and the human interference especially for the infrastructure both for large and small holders in forestry companies (Stolle, Chomitz, Lambin, & Tomich, 2003).

The focus area of the research is a province of Jambi. Database of destructive sample is collected from enclose to Jambi province.

## Data collection

Data collection is gathered from compiling published journal and unpublished report, which research is conducted in Sumatra. In order to gain the database of biomass estimation in Sumatra, data of destructive sampling is gained from the scattered region of Sumatra. Each of the regions has different type of forest, in table 2.1 describing the sites of destructive data scattered in Sumatra.

Table 2. Description of source database

## Suite no.

## Province

## Region

## Site name

## Lat (E)

## Long (N)

## Precip.

## MAT

## No. of trees

## Soil type

1

## Â

South Sumatra

Marian Peat Dome Forest

Â Ni

Â Ni

2454 mm

26.4-27.50C

20

Histosol

2

## Â

South Sumatra

Marian

Â n.i

n.iÂ

2304 mm

Â n.i

30

Histosol

3

## Â

South Sumatra

IUPHHK-HT PT. SBA WI

n.iÂ

Â n.i

Â n.i

n.iÂ

36

n.iÂ

4

## Â

North Sumatra

Sektor Habinsaran PT.Toba Pulp Lestari Tbk

9905'-18'

207'-21'

128 mm

Â n.i

30

Â n.i

5

Jambi

Middle Sumatra

Sepunggur area

102o14'E

1o29'S

3000mm

22.1-32.3oC

29

Â n.i

Soil characteristic is one of source of research to estimating carbon stock. The soil type determines the amount of carbon stock retained. The parameters of calculating carbon stock are compound of organic carbon and bulk density. Database of soil in Indonesia (Shofiyati, Las, & Agus, 2010), stated that bulk density in peat forest on range 0.4-0.6 g cm-3. Peat is also finding in Sumatra forest especially in Province Jambi forest. Hence, in the compilation one of the site soil characteristic is Histosol, based on Van Noordwijk (YEAR) has low bulk density as well in range 0.1-0.8 Mg m-3.

## Destructive data set

For this study, no destructive sample is done. A data set of 143 trees destructive sampling of existing research in Sumatra is compiled. The database of destructive sample is based upon published and unpublished research conducted in Sumatra, which is compiled from the library of the Institut Pertanian Bogor (IPB). The information about each site or each author is single tree is dry weight of stem, branch, twigs, stumps, leaves, flower, and wood density. As it is seen on Table 2.2

Table 2. Description of destructive sample

## Author

## Location

## Forest Type

## No of Species

## N

## Max DBH [cm]

## Max Height [m]

Eka Widyasari H., 2010

Sumatera Selatan

Secondary Forest

11

20

30.2

19.1

Novita, 2010

Sumatera Selatan

Secondary Forest

18

30

64

31.2

Limbong, H. D., 2009

Sumatera Selatan

Plantation

1

36

25.7

26

Siahaan, A. F., 2009

Sumatera Utara

Plantation

1

30

21.02

22.23

Ketterings,et.al., 2001

Sumatra

Secondary Forest

15

29

48.1

32.4

## Allometric model in Sumatra

The analysis relies upon a compilation of tree harvest studies carried out since 1999. It is compiled from 5 published and unpublished research. The compilation of destructive sample is data from different regions in Sumatra. North Sumatra, Southern Sumatra, and middle Sumatra. The compilation is 144 numbers of trees, which categorized into two major groups. Nature forest, which is abbreviated with N in the Table 2.4 and Plantation forest, is P. The grouping is due to analyzing and generating an allometric model for a different type of forest.

The biomass analysis in this research is mainly focused on total aboveground dry weight. Therefore, in Table 2.4 is showing the allometric model for total aboveground from such authors.

Table 2. Total aboveground allometric equation 5 different authors

## Author

## P/N

## Equation

## a

## B

## c

## d

## Î²3

## R2

Eka Widyasari H., 2010

N

w=aDb

0.153108

2.4

## Â

## Â

## Â

97.8

N

w=exp{a+b[ln(D)]+c[ln(D)]2+d[ln(D)]3}

-1.51

2.08

-0.002

0.023

## Â

97.9

N

w=a(D2H)b

0.095499

0.897

## Â

## Â

## Â

97.7

N

w=exp{a+b[ln(D2H)]+c[ln(D2H)]2}

-2.5

0.858

0.23

## Â

## Â

97.7

Novita, 2010

N

w=aDb

0.020628

2.4511

## Â

## Â

## Â

96.1

N

w=exp{a+b[ln(D)]+c[ln(D)]2+d[ln(D)]3}

3.465

-2.948

1.861

-0.207

## Â

96.2

N

w=a(D2H)b

0.0746

0.949

## Â

## Â

## Â

96.1

N

w=exp{a+b[ln(D2H)]+c[ln(D2H)]2}

-3.06

1.06

-0.0064

## Â

## Â

96.1

N

w=aDbÏc

0.39772

2.34637

0.6302

## Â

## Â

97.5

N

w=exp{a+b[ln(D)]+c[ln(D)]2+d[ln(D)]3+Î²3[ln(Ï)]}

1.351

-0.618

1.238

-0.162

0.7224

97.8

Limbong, H. D., 2009

P

w=aDb

1.464

1.549

## Â

## Â

## Â

87

P

w=a+bD+cD2

179

-24.89

1.278

## Â

## Â

96.7

P

w=a(D2H)b

190.4

0.593

## Â

## Â

## Â

80.7

P

w=a+b(D2H)+c(D2H)2

44.46

89.16

202.9

## Â

## Â

88.9

P

w=aDb

0.003

3.519

## Â

## Â

## Â

81.8

P

w=a+bD+cD2

-1374

108.3

-1.735

## Â

## Â

88.4

P

w=a(D2H)b

146.2

1.298

## Â

## Â

## Â

86.6

P

w=a+b(D2H)+c(D2H)2

-133.4

340.7

-58.47

## Â

## Â

87.2

P

w=aDb

0.724

1.856

## Â

## Â

## Â

85.9

P

w=a+bD+cD2

1920

-170.9

4.226

## Â

## Â

89.7

P

w=a(D2H)b

210

0.713

## Â

## Â

## Â

78.9

P

w=a+b(D2H)+c(D2H)2

337.2

-341.6

203.1

## Â

## Â

83.5

Siahaan, A. F., 2009

P

w=aDb

288.4032

1.94

## Â

## Â

## Â

93.9

Ketterings,et.al., 2001

N

w=aDb+Îµ

0.0661

2.591

## Â

## Â

## Â

## Â

## Regression models

Analysis of this research study is by using the regression analysis to determinate the best-fit model that is suitable. First, is by determining the coefficient to generate an allometric equation. There are methods to establish coefficient, by scatter plot, linear regression analysis, and the least square method. It is important to observe the relation linear regression of diameter and dry weight biomass by applying scatter plot analysis. In addition, regression line on the scatter plot is giving result of a mathematical equation-containing the coefficient of x-independent variable and y-independent variable. The linear regression analysis is able to determine the coefficient from more than one predictor variable, in this term variable such as diameter, height, and wood dry density. The application of predictive depends on the type of regression model that is used. The outcomes of linear regression analysis are coefficient, R square, and standard error of each coefficient. The least square method to identify coefficient is using the form of linear regression model, in other word is by transforming the actual data to logarithmic.

Normally, the regression model for biomass allometric equation is available in a published research journal. To generate the best-fit model is by applying coefficients into regression models, by combining of the predictor to obtain general allometric equation based on data.

In this research analysis, there are several regression models to obtain the best-fit model. The first model is using diameter as the predictor variable, the second model is by adding the parameter height as the predictor, the third model predictors are tree diameter, tree-height, and wood density, and the fourth model predictors are tree diameter and wood density.

## Biomass-diameter regression (model I)

The first model is by using the most common allometric equation, non-linear equation

Eq 2.

Where a and b are the coefficient in the equation. The estimation of the statistical coefficient of nonlinear equation (a and b) is analyzed by using the scatter plot and multiple regression with statistical software STATISTICA 10.

The natural logarithm is applied to the data. The exponents are the result from least square regression of non-linear parameter (least square estimation.) of natural logarithmic of tree diameter and dry weight biomass (Navar, 2010 & Zianis, 2008). The equation 2 is log linear equation

Eq 2.

The improvement for constructing an allometric equation is by transforming natural logarithmic variables in the equation. The logarithmic transformation is to reduce heterogeneous variance in the data. In order to construct allometric equation, the biomass unit is transformed back to its value, thus, requires a correction factor to fit in the model (Navar, 2010); biomass units might resulting bias in the model (Miller, 1984). Many regression models have been published; therefore, only several models are used for this research.

## Biomass-diameter-height regression (model II)

The regression analysis with additional parameter height is due to the heterogeneous height for each species. In this analysis, testing the model with tree-height parameter is necessary since the presence of tree-height information in both groups. Applying tree-height is to observe if the tree-height will give improvement in biomass estimation.

The regression model with tree-height parameter used in this analysis is

Eq 2.

The linear regression is the natural logarithmic of all biomass units

Eq 2.

## Biomass-diameter-height and wood density

The regression analysis using all predictors in the equation is intended due to the difference height in-group of nature forest and in other hand rather equal in group of plantation. The variety of species in nature forest differs in value of wood density. Therefore, to test best-fit model which including predictor's diameter, height, and wood density are expected to give a fine outcome. The regression model is given by the equation 2.5

Eq 2.

The logarithmic transformation of the equation

Eq 2.

## Biomass-diameter-wood density

The regression analysis wood density as the addition factor is important in calculating biomass (Chave, et al., 2005). In the existing data of harvested tree not all sites are representing the wood density. The information on the wood density database is taken through (Rahayu, S; Ketterings et al. 2001) the World Agroforestry web sourcece.

The regression model with predicted variable wood density is shown in the equation 2.7

Eq 2.

In order to generate coefficients in the equation by using least squares it is require transforming the data and equation to linear regression.

Eq 2.

## Model selection

In order to get the best-fit model concerning data collection model selection is applied by observing the value of the coefficient of determination r square (r2) (Basuki, van Laake, Skidmore, & Hussin, 2009), and the result of Akaike Information Criterion (AIC) (Chave, et al., 2005). The result of most coefficient determination for Nature and Plantation forest is above 90%. Chave, et al., (2005) is referring to (Burnham & Anderson, 2004) used AIC as the model selection from the best model. The AIC analysis for model selection is based on the minimum value of AIC is the "best" statistical model and parameter balance, the AIC formula as model selection which used in (Basuki, van Laake, Skidmore, & Hussin, 2009 and Chave, et al., 2005 is shown in equation 2.9

Eq 2.

Where,

L is the log likelihood of the fitted model, and p is the number parameter in the model.

The AIC analysis is conducted by function of Generalized Linear Models in Statistica program.

## Result and Discussion

## Total above-ground biomass-diameter regression

Comparing mixed and plantation for regression model 1

The allometric equation for the first model is by using the parameter diameter to total aboveground biomass (LN (AGB) = LN (a) + b LN (D)). The value of the coefficient allometric equation is described in the table 3.1. The coefficient determinate of adjusted r2 for mixed forest is 0.946 and 0.945 for plantation. Based on result standard error of estimation of model in Plantation forest is showing the best fit in comparison to mixed forest. In addition, there is a slight difference between coefficients in two models.

Table 3. Result of the regression analysis with model 1 (Biomass-Diameter) Nature and Plantation forest

## Site

## Model

## Coefficient

## Standard error of coefficient

## Adjusted R2

## Standard error of estimate

N

AGB=aDb

a

0.15

0.03

0.946

0.47

## Â

## Â

b

2.44

0.06

P

AGB=aDb

a

0.16

0.03

0.945

0.32

## Â

## Â

b

2.29

0.07

Based on ANOVA test, the result of regression model 1 both of result N and P from this are significant differences.

Table 3. ANOVAs result of regression model 1 between Nature and Plantation forest

The influence of regression model 1 to both of group resulting difference result for the estimated total aboveground. In the plantation group effects after application of regression model 1 is resulting of the smaller range of total above ground compare to effect on total aboveground on nature forest. It shows clearer on the figure 3.1 where the comparison between total aboveground after applying regression model 1. Although, the diameter ranges both groups of forest is not significantly different.

Figure 3. Graphic of model 1 between group Nature and Plantation forest

As it is seen in the figure 3.1 the result of first model where the predictor is only the diameter. There is a difference in the dispersion of the result from model 1 both nature and plantation. In the plantation group, showing the trend of total above ground of biomass is shorter, where the range of total above ground biomass is wider in nature forest. This possibility mainly causes by the trunk of the trees in the nature forest is bigger than in the plantation group.

## Total aboveground biomass-diameter-height regression

Table 3.

## Site

## Model

## Coefficient

## SE

## R2

## SEE

N

AGB=aDbHc

a

0.09

0.02

0.95

0.40

## Â

## Â

b

2.03

0.16

## Â

## Â

c

0.58

0.21

P

AGB=aDbHc

a

0.29

0.07

0.96

0.29

## Â

## Â

b

3.06

0.20

## Â

## Â

c

-0.95

0.23

## Total aboveground biomass-diameter-height-wood density

Table 3.

## Site

## Model

## Coefficient

## Standard error of coefficient

N

AGB=aÏD2H

a

0.07

0.005

P

AGB=aÏD2H

a

0.47

0.02

## Total aboveground biomass-diameter-wood density

Table 3.

## Site

## Model

## Coefficient

## Standard error of coefficient

## R2

## Standardr error of estimate

M

AGB=aDbÏc

a

0.182

0.231

0.958

0.459

## Â

## Â

b

2.405

0.069

## Â

## Â

c

0.235

0.174

P

AGB=aDbÏc

a

1.940

0.261

0.980

0.193

## Â

## Â

b

1.872

0.057

## Â

## Â

c

3.175

0.300

## Model selection using the Akaike Information Criterion

Nature forest

Table 3.

## Nature Forest

## Model

## Coefficient

## Standard error of coefficient

## Adjusted R2

## Standard error of estimate

## Loglikelihood

## AIC

AGB=aDb

a

0.145

0.03

0.946

0.47

-578

1162

## Â

b

2.435

0.06

AGB=aDbHc

a

0.09

0.02

0.95

0.40

-575

1155

## Â

b

2.03

0.16

## Â

c

0.58

0.21

AGB=aÏD2H

a

0.07

0.005

## -

## -

-574

1154

AGB=aDbÏc

a

0.182

0.231

0.958

0.459

-581

1167

## Â

b

2.405

0.069

## Â

c

0.235

0.174

Table 3.

## Plantation forest

## Model

## Coefficient

## Standard error of coefficient

## Adjusted R2

## Standard error of estimate

## Loglikelihood

## AIC

AGB=aDb

a

0.159

0.03

0.945

0.32

-324.28

654.55

## Â

b

2.290

0.07

AGB=aDbHc

a

0.29

0.07

0.96

0.29

-328.43

662.85

## Â

b

3.06

0.20

## Â

c

-0.95

0.23

AGB=aÏD2H

a

0.47

0.02

## -

## -

-327.82

661.63

AGB=aDbÏc

a

1.940

0.261

0.980

0.193

-311.54

629.07

## Â

b

1.872

0.057

## Â

c

3.175

0.300

## Development of general models

## Conclusion

## ANNEX

Basuki, T. M., van Laake, P. E., Skidmore, A. K., & Hussin, Y. A. (2009). Allometric equations for estimating the above-ground biomass in tropical lowland Diptercarp forests. Forest Ecology and Management 257 , 1684-1694.

Brown, S. (1997). A primer for estimating biomass and biomass change of tropical forests. Rome: FAO Forestry Paper.

Brown, S., Gillespie, A. J., & Lugo, A. E. (1989). Biomass estimation methods for tropical forests with applications to forest inventory data. Forest Science, Vol 35 , 881-902.

Chave, J., Andalo, C., Brown, S., Cairns, M. A., Chambers, J. Q., Eamus, D., et al. (2005). Tree allometry and improved estimation of carbon stocks and balance in tropical forests. Oecologia 145 , 87-99.

Eka Widyasari H., N. A. (2010). Pendugaan biomass dan potensi karbon terikat di atas permukaan tanah pada hutan gambut merang bekas terbakar di Sumatera Selatan ( Tesis Pascasarjana). Bogor: Institut Pertanian Bogor.

Fahrizal. (2002). Pendugaan biomassa pohon bagian bawah tanah di hutan primer dan hutan bekas tebangan (Studi kasus di hutan Tri Dharma IPB, Dusun Aro Jambi) ( Skripsi Sarjana IPB). Bogor: Institut Pertanian Bogor.

FAO . (2000, November 2). Retrieved March 12, 2012, from Forestry Department: Forest resources Assessment: http://www.fao.org/

Fehrmann, L., Lehtonen, A., Kleinn, C., & Tomppo, E. (2008). Comparison of linear and mixed-effect regression models and a k-nearest neighbour approach for estimation of single-tree biomass. Canadian journal forest research 38 , 1-9.

Gibbs, H. K., Brown, S., Niles, J. O., & Foley, J. A. (2007). Monitoring and estimating tropical forest carbon stocks: making REDD a reality. Environmental Research Letter 2 , 13.

Gillespie, A. J., Brown, S., & Lugo, A. E. (1992). Tropical forest biomass estimation from truncated stand tables. Forest Ecology and Management, 48 , 69-87.

Grace, J. (2004). Understanding and managing the global carbon cycle. Journal of Ecology, Vol 92, No.2 , 189-202.

Johnsen, K., Samuelson, L., Teskey, R., McNulty, S., & Fox, T. (2001). Process models as tools in forestry research and management. Forest Science 47 , 1.

Kementerian Kehutanan Republik Indonesia. (2010). Retrieved March 1, 2012, from http://www.dephut.go.id/

Ketterings, Q. M., Coe, R., van Noordwijk, M., Ambagau, Y., & Palm, C. A. (2001). Reducing uncertainty in the use of allometric biomass equations for predicting above-ground tree biomass in mixed secondary forests. Forest Ecology and Management 146 , 199-209.

LaBarbera, M. (1989). Analyzing body sizes as a factor in ecology and evolution. Annual Reviews of Ecology and Systematics, Vol 20 , 97-117.

Laumonier, Y., Uryu, Y., Stuwe, M., Budiman, A., Setiabudi, B., & Hadian, O. (2010). Eco-floristic sectors and deforestation threats in Sumatra: identifying new conservation area network priorities for ecosystem-based land use planning. Biodiversity Conservation 19 , 1153-1174.

Levang, D. P., Yoza, I. B., & Tasman, D. A. (1999). In the shadow of rubber. Jakarta: Institut de recherche pour le developpement.

Makila, M., Saarnisto, M., & Kankainen, T. (2001). Aapa Mires as a carbon sink and source during the Holocene. Journal of Ecology, Vol.89, No.4 , 589-599.

Miller, D. M. (1984). Reducing transformation bias in curve fitting. The American Statistician, Vol.38 , 124-126.

Moutinho, P., & Schawrtzman, S. (2005). Tropical deforestation and climate change. Washington DC: Amazon Institute for Environmental Research.

Navar, J. (2010). Measurement and assessment methods of forest aboveground biomass: A literature review and the challanges ahead. Mexico: CIIDIR-IPN Unidad Durango.

Novita, N. (2010). Potensi karbon terikat di atas permukaan tanah pada hutan gambut bekas tebangan di Merang, Sumatera Selatan (Tesis Pascasarjana). Bogor: Institut Pertanian Bogor.

Partohardjono, S., Pasaribu, D., & Fagi, A. M. (2005). The forest margins of Sumatra, Indonesia. Bogor: The National Perspective, ASB, CGIAR.

Ponce-Hernandez, P., Koohafkan, P., & Antoine, J. (2004). Assessing carbon stocks and modelling win-win scenarios of carbon sequestration through land-use changes. Rome: Food and Agriculture Organization Of The United Nations.

Rahayu, S; Ketterings et al. 2001. (n.d.). World Agroforestry. Retrieved August 20, 2012, from World Agroforestry: http://www.worldagroforestry.org/sea/Products/AFDbases/wd/Index.htm

Somogyi, Z., Cienciala, E., Makipaa, R., Muukkonen, P., Lechtonen, A., & Weiss, P. (2007). Indirect methods of large-scae forest biomass estiomation . Eur J Forest Res 126 , 197-207.

Somogyi, Z., Cienciala, E., Makipaa, R., Muukkonen, P., Lehtonen, A., & Weiss, P. (2007). Indirect methods of large-sclae forest biomass estimation. Euro Journal Forest Resource 126 , 197-207.

Stolle, F., Chomitz, K. M., Lambin, E. F., & Tomich, T. P. (2003). Land use and vegetation fires in Jambi Province, Sumatra, Indonesia. Forest Ecology and Management 179 , 277-292.

UNFCCC. (2008). Report of the Conference of the parties on its thirteenth session. Bonn: UNFCCC.

Wulder, M. A., White, J. C., Fournier, R. A., Luther, J. E., & Magnussen, S. (2008). Spatially explicit large are biomass estimation: three approaches using forest inventory and remotely sensed imagery in a GIS. Sensors 8 , 529-560.

Yuniawati, Budiaman, A., & Elias. (2011). Estimationg biomass and carbon mass potency of wood plantation of Acacia crassicarpa growing on peat land site (A case study on fiber wood plantation area at Pelalawan, Riau Province). Penelitian Hasil Hutan Vol. 29 No.4 , 343-355.

Zianis, D. (2008). Predicting mean aboveground forest biomass and its associated variance . Forest Ecology and Management 256 , 1400-1407.