干旱区生态环境知识资源中心

Waterlogging and subsequent salinization and/or alkalization of soils is the major land degradation process operating in the irrigated areas of the arid and semi-arid regions of the world. Before taking up any prevention or reclamation measures for such lands, information on their nature, spatial extent, magnitude, distribution, and temporal behavior is a prerequisite. In the study reported here, an attempt has been made to evaluate the potential of high spatial resolution (5.8 m) Panchromatic (PAN) sensor data and the Linear Imaging Self-scanning Sensor (LISS-III) data from the Indian Remote Sensing Satellite (IRS-1C) for detection and delineation of salt-affected soils in a portion of the Indo-Gangetic alluvial plains of northern India. The approach involves the merging of LISS-III and PAN data through an Intensity, Hue, and Saturation (IHS) transformation, and a subsequent supervised classification using a per-pixel Gaussian maximum-likelihood classification algorithm. Results indicate a deterioration in the overall accuracy of salt-affected soils derived from LISS-III data as compared to IRS-1B LISS-II data owing to an improvement in the spatial resolution (23.5 in for LIS-III versus 36.5 m for LISS-II), leading to enhanced intra-class spectral variability. The PAN and Liss-III hybrid data without any transformation ranked last in terms of overall accuracy. Overall accuracy figures for LISS-11, LISS-III, and PAN and Liss-III hybrid data with the IHS transformation have been on the order of 89.6 percent, 85.9 percent, and 81.5 percent, respectively.