Microwave Remote Sensing for Land Cover Identification

Abstract

This study considers the potential application of digital JERS-1 SAR data (L-band, 23.5 cm Wavelength) particularly for land cover monitoring. The SAR Data used in this study were acquired on 5 January 1998, 3 April 1998 and 17 May 1998 respectively. These three images available were from the dry season to the beginning of the rainy season. The study area selected is in the province of Khon Kaen, Northeast Thailand, which is covered by agricultural land, mainly paddy fields and upland crops. Multi-dates SAR data were firstly geocoded to UTM projection, then the data image were scaled from 16 bits to 8 bits. Some filtering and enhancement methods had to be applied in order to reduce speckle noise and to contrast the image. Composite color images were produced for visual interpretation and field surveys. After investigation of the ground truth, representative areas of each land cover type were identified and allocated to the images. The SAR data of training areas were statistically analyzed and were supervise clarified using Maximum Likelihood. The preliminary results show that general land cover categories could be well indentified, such as water bodies, villages and some lineament features (airfield, road, canal). Separability matrices indicate the confusion between paddy and upland crops, forest and riparian forest. Therefore SAR data is useful for land cover characterization and for hydrology applications, but a specific methodology for the SAR system needs to be developed.