High-resolution satellite remote sensing for aboveground biomass estimation of tropical rainforests and oil palm plantations in Sabah

Alexius Korom (2017) High-resolution satellite remote sensing for aboveground biomass estimation of tropical rainforests and oil palm plantations in Sabah. Doctoral thesis, Universiti Malaysia Sabah.

	Text 24 PAGES.pdf Download (832kB)
	Text FULLTEXT.pdf Restricted to Registered users only Download (16MB)

Abstract

The lack of research in the transition of reporting method in Reducing Emissions from Deforestation and Forest Degradation-plus (REDD+) program from using a low-resolution to high-resolution satellite image (HRSI) has led to this study. The approach of using geographical object-based image analysis (GEOBIA) towards the AGB quantification method is still not completely explored. The potential uses of HRSI in estimating the aboveground biomass (AGB) for two major land covers, oil palm plantations and tropical rainforests, in the middle of Sabah, East Malaysia was examined in this study. Field data collections to determine the AGB in oil palm plantation and forest reserve area were obtained through stratified random sampling method. Using HRSI, the effective spectral bands and vegetation indices were identified for segmentation and classification of objects where three kinds of useful information were extracted, that are spectral, geometrical and textural properties for modelling purposes. The AGB of oil palm plantation was estimated based on crown using a WorldView-2 image. A total of 222 samples of fieldcollected age-based data, AGB's regression towards the crown variables ( crown diameter, crown area and crown perimeter) revealed the exponential nonlinear functions (R2 of 0.80 rv 0.85), which fulfil the lack of an allometric equation. Watershed technique was used to segment the oil palm crown at 4.8 % and 10.6 % of omission and commission errors. Due to overlaps in the oil palm's crown, agebased crown difference correction was implemented unto the detected crown. Among the crown variables, crown diameter was found to be the best in estimating the AGB for mature oil palm which improved from 62.9 Mgha-1 (root-mean-square error (RMSE) 34.2 Mgha-1) for detected crown to 122.5 Mgha-1 (RMSE 16.4 Mgha-1) for corrected crown; relative RMSE was 4.1 times lower after the correction. For crown area and crown perimeter, the relative RMSEs are both 1.8 times lower after the correction. On the other hand, AGB of the logged-over forest reserves were estimated using IKONOS-2 image based on two approaches; forest degradation classification and crown-based approach. Forest degradation classification approach utilised the spectral and textural information of forest surface roughness where else, a crown-based approach used the geometrical information from the crown. Forest degradation reduces AGB and alters forest canopy structure implicitly but related. Modelling by restricting the complexities of forest surface roughness into respective forest degradation classes (very-degraded, degraded and intact forests) was implemented. Regression analyses confirmed the best independent variables for modelling the AGB are textural and spectral properties for each forest type. The estimated AGBs for very-degraded, degraded and intact forests are 73.6, 148.4 and 270.7 Mgha-1 with RMSEs of 15.9, 38.9 and 18.7 Mgha-1 respectively. A comparison with the classical non-classified forest approach (AGB = 126.1 Mgha-1, RMSE = 64.1 Mgha-1) approved the advantage of forest degradation classification approach. Meanwhile, the approach to use geometrical information from crown has underestimated the AGB by 19.4 % (AGB = 115.8 Mgha-1, RMSE = 87.9 Mgha-1) from field-based AGB. In using optical high-resolution satellite remote sensing data, forest degradation classification approach had greatly improved the widely reported saturation problem in AGB estimation, which normally occurred at high AGB density. As a conclusion, this study had extensively examined the use of crown shape and forest texture to estimate the AGB of oil palm plantation and tropical rainforest; two major competing land use in tropics.

Item Type:	Thesis (Doctoral)
Keyword:	Biomass, Tropical rainforests, Oil palm
Subjects:	S Agriculture > SD Forestry > SD1-669.5 Forestry
Department:	FACULTY > Faculty of Science and Natural Resources
Depositing User:	DG MASNIAH AHMAD -
Date Deposited:	29 Apr 2024 10:36
Last Modified:	29 Apr 2024 10:36
URI:	https://eprints.ums.edu.my/id/eprint/38554

Actions (login required)

View Item