29/08/2013
1
ESTIMATING THE SPATIAL DISTRIBUTION OF THE ABOVE GROUND BIOMASS DERIVED FROM TERRESTRIAL-BASED FOREST INVENTORY
I Nengah Surati Jaya Fauziah Dwi Hayati
Putu Arimbawa & M. Buce Saleh
INTRODUCTION
METHODS
RESULTS
CONCLUSION
Fac. of Forestry, IPB
INTRODUCTION 1:
International concern
on GHG issues:
C stock of Indonesia, India & Myanmar for (70%) of the Asian tropical forest
Global Concerns
Defor Degrad
Conserv.
SFM
Increase C. stock .
Need •C St. measurement
Support •SFM, Research, MRV
29/08/2013
2
INTRODUCTION 2
One
• Many techniques forest biomass terrestrial or remote sensing technique
Two
• Periodically performed forest inventory, IHMB by each forest concession holders (IUPHHK), in natural or plant. Forests
Three
• spatial information of forest biomass is an important for supporting SFM
Four
• Very few researches were found related to biomass estimation using terrestrial data. Almost no research for biomass estimation using IHMB data in Indonesia
ONE
• (1) What is the characteristics of spatial distribution for each forest ecosystem type?
TWO
• (2) Is the spatial distribution pattern of old forest biomass similar to the newly logged over forests or disturbed logged-over forest or degraded forest due to fires?
INTRODUCTION: RESEARCH QUESTION & OBJECTIVE
OBJECTIVE
1. Identify the most accurate INT. METHOD for estimating the spatial distribution of forest biomass
in the dry forest.
(add) 2. Determine the spatial distribution patterns of forest stand biomass dry land
29/08/2013
3
SITES • SITE 1 LABANAN
2O0’N;117005’)
•
• SITE 2 LAMANDAU TSI: 1°45’S 111°37’30E
1050 m 1000 m
1450 m 1450 m
899 m 1000 m
1345 m 1345 m
1
2
(1) site 1, PT Inhutani I, in Labanan District, Berau
Regency, East Kalimantan Prov.
(2) site 2, PT TSI , Lamandau District, Kota
Waringin Barat regency, C. Kalimantan Prov.
Methods Approaches Interpola
tion
method
Deterministic
Splines
Thiessen P.
IDW
Geostatistic Ordinary Kriging
Universal Kriging
Combination
METHODS
29/08/2013
4
ANALYSIS
4. MODEL EVALUATION: ACTUAL VS PREDICTED COMP.
MEAN DEV (MD); RMSE; Chi-sq; BIAS
n
i
p
i
p
i
i
d
d
1
/1
/1
)]()(var[2
1)(),( 000 xZxZhxx i
1. IDW
2. KRIGING
LIN
Gaus
Sper
Exp
Cir
3. Eq AGB IHMB AGB all (Jaya, et al, 2012)
AGBAll = 7.355 + 1.003*AGBIHMB
RESULTS ERROR VALUE & SCORE FOR SITE 1
• LABANAN SITE, BERAU EAST KALIMANTAN
,0000
,005000
,010000
,015000
,020000
,025000
,030000
,0
20,0
40,0
60,0
80,0
100,0
120,0
P05 P1 P2 P3 P4 P5 P6 SP CR EX GS LN
valu
e o
f er
ror
Validation value for each interpolation method
Chi-sq
RMSE (%)
BIAS (%)
MD (%)
AD
interpolation methods
-
2,00
4,00
6,00
8,00
10,00
12,00
P05 P1 P2 P3 P4 P5 P6 SP CR EX GS LN
Sco
res
Scores for each interpolation method
Chi-sqRMSEAD(SA)BIAS
Interpolation method
29/08/2013
5
ERROR VALUE & SCORE FOR SITE 2
• LAMANDAU – CENTRAL KALIMANTAN
-,050000
,0000
,050000
,10000
,150000
,20000
,250000
,0
10,0
20,0
30,0
40,0
50,0
60,0
P05 P1 P2 P3 P4 P5 P6 K_S K_C K_E K_G K_L
valu
e o
f er
ror
Validation value for each interpolation method in PT TSI
Chi-sq
RMSE (%)
BIAS (%)
MD (%)
AD
Interpolation method va
lue
of
AD
-1,90
,10
2,10
4,10
6,10
8,10
10,10
P05 P1 P2 P3 P4 P5 P6 K_S K_C K_E K_G K_L
Sco
re
Score value for each interpolation method in PT TSI
Chi-sq
RMSE (%)
AD
BIAS (%)
MD (%)
Average
Interpolation method
SPATIAL DISTRIBUTION SITE 1
• IDW, P=2
• ORD KRIG - EXPONENTIAL
SITE 2
• IDW, p=2
• ORD KRIG - EXPONENTIAL
29/08/2013
6
The IDW give more accurate estimation than Kriging
For Labanan (with 1450 m x 1450 m: the best IDW with p= 2, having MD = 15% >> Kriging MD=21%
For Lamandau ( dist 1345 m x 1345 m: the best is IDW p=2, provide MD = 17% >> The Kriging MD = 27%.
The IHMB data is quite useful for estimating biomass with low error
CONCLUSION
COMPARTMENT OF INHUTANI I
29/08/2013
7
TSI – ALOS PALSAR
TSI – biomass measurement
Sample plots