Satellite remote sensing for evaluation of burned peat swamp forest

Mohammad Hasli Salleh Celestial, . and Mui-How P, . and Abang Norizan Abang Median, . and Mohd. Abdullah Abd. Ghani, . (2005) Satellite remote sensing for evaluation of burned peat swamp forest. [Proceedings Paper]

This paper discusses information extraction on burned peat swamp forest at Klias Forest Reserve Sabah Malaysia using satellite data. The study area is a part of the extensive peat swamp forest of Klias Peninsula but was severely damaged by forest fires especially during the 1997 El Nino event. Information related to the damage including burned area and its distribution as well as some forest variables can aid fire management. The burned area and distribution were evaluated using multi-temporal satellite images of landsat-TM 1997 and Landsat ETM 1999. The burned areas were detected using image differencing technique based on Normalized Burn Ratio NBRand Normalized Difference Vegetation Index NDVI. The NBR differencing with the accuracy of 86 and 71 for total accuracy and coefficient of kappa was superior to the NDVI differencing. Lower and upper estimates of the burned area were 262 ha and 300 ha respectively. This represents approximately 7- 8 of the forest reserve. We also examined the correlation between the 1999 satellite data including transformed indices and forest variables. Among the satellite variables NBR show relatively strong correlations with dark soil colour percent number of tree crowns and number of green tree crowns. These variables indicate directly and indirectly the degree of damage due to the fire and are important to rehabilitation and fire risks management.

Full text available from:

Abstract

This paper discusses information extraction on burned peat swamp forest at Klias Forest Reserve Sabah Malaysia using satellite data. The study area is a part of the extensive peat swamp forest of Klias Peninsula but was severely damaged by forest fires especially during the 1997 El Nino event. Information related to the damage including burned area and its distribution as well as some forest variables can aid fire management. The burned area and distribution were evaluated using multi-temporal satellite images of landsat-TM 1997 and Landsat ETM 1999. The burned areas were detected using image differencing technique based on Normalized Burn Ratio NBRand Normalized Difference Vegetation Index NDVI. The NBR differencing with the accuracy of 86 and 71 for total accuracy and coefficient of kappa was superior to the NDVI differencing. Lower and upper estimates of the burned area were 262 ha and 300 ha respectively. This represents approximately 7- 8 of the forest reserve. We also examined the correlation between the 1999 satellite data including transformed indices and forest variables. Among the satellite variables NBR show relatively strong correlations with dark soil colour percent number of tree crowns and number of green tree crowns. These variables indicate directly and indirectly the degree of damage due to the fire and are important to rehabilitation and fire risks management.

[error in script]

Item Type:	Proceedings Paper
Additional Information:	3 ills. 2 tables. 14 refs.
AGROVOC Term:	PEATLANDS
AGROVOC Term:	SWAMPS
AGROVOC Term:	WETLANDS
AGROVOC Term:	SATELLITES
AGROVOC Term:	REMOTE SENSING
AGROVOC Term:	MALAYSIA
Geographical Term:	MALAYSIA
Depositing User:	Ms. Norfaezah Khomsan
Last Modified:	24 Apr 2025 05:28
URI:	http://webagris.upm.edu.my/id/eprint/16786

Actions (login required)

View Item