Assessment of Microw Ave Remote Sensing Technology (Nasa's Airsar-Topsar Data) for Forest Type Classification on Tioman Island, Pahang, Malaysia

Active microwave remote sensing is able to provide information about land surface and forest canopy that would otherwise be unobtainable in regions where cloud cover and darkness prevail. The general objective of this study is to assess the capability and applicability of NASA's airborne SAR...

Full description

Saved in:
Bibliographic Details
Main Author: Sebastian, Chew
Format: Thesis
Language:English
English
Published: 2000
Subjects:
Online Access:http://psasir.upm.edu.my/id/eprint/10001/1/FH_2000_1_IR.pdf
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Active microwave remote sensing is able to provide information about land surface and forest canopy that would otherwise be unobtainable in regions where cloud cover and darkness prevail. The general objective of this study is to assess the capability and applicability of NASA's airborne SAR (AirSAR) data to classify and map tropical forests utilizing the Environment for Visualizing Images (ENVI) image processing software. The specific objectives are to test the applicability of TOPSAR in classifying forest types of Tioman Island applying standard classification method, generate a digital topographic map, generate a DEM of the study site and produce a forest-type map of Tioman Island. The island is approximately 13,354 hectares. AirSAR's Topographic SAR (TOPSAR) data of the entire island comprising of two strips were acquired on a 3rd December 1996 flight mission. Prior to analysis, the image had to be despeckled to remove noise. Five adaptive filters were used and the Gamma filter with an 11x11 window produced the best visual results after initially applying image contrast stretching. Preliminary ground survey revealed that the island has at least five main vegetation types identified as primary forest, beach forest, secondary forest, coconut palm plantation and mangrove forest. Two methods of interpretation were applied. In the first method, visual interpretation was initially applied where distinct different tones and texture were designated as a "Region of Interest" (ROI) for signature extraction. 16 ROIs were created to represent four vegetation covers and polarization signatures for each ROI were generated. Extracted polarization signatures showed no specific signature or pattern for a particular known forest type. The second method, unsupervised classification, initially yielded 10 classification classes. However, only two land cover classes were readily distinguished and these could be classified as primary and secondary forests. An additional classification obtained is cleared land or developed land. It is therefore suggested that fully Polarimetric SAR data be used together with the TOPSAR.