Retrieval of Soil Wetness over Indian Subcontinent Using DMSP SSM/I Data

Rajneesh Kumar  M.Tech (Civil Engineering)

Abstract 

     Soil wetness is an important parameter that provides information about the state of soils. This information can be used in the estimation of surface runoff and forecasting of floods. Advances in microwave remote sensing have shown the potential to provide information about soil wetness. Precise insitu measurement of soil wetness is difficult and each value is only representation of a small area. Soil wetness data for studies over large regions can be obtained through microwave remote sensing. The present study was initiated to explore the possible application of special sensor microwave imager (SSM/I) recorded brightness temperature data to estimate soil wetness index (SWI), and to use SWI, normalized difference vegetation index (NDVI) and precipitation data to forecast floods. SSM/I recorded brightness temperature data for land surfaces have been analyzed to calculate the SWI for the Indian subcontinent. The validity of SWI estimate was tested through (i) difference in estimated SWI values in monsoon and non-monsoon period and (ii) analysis of correlation between estimated SWI and precipitation. An attempt has been made to study the precipitation, SWI and NDVI patterns during the past floods of large areal extent.  

 The SWI estimates were found to be significantly higher during the monsoon period as compared to the non-monsoon period. A positive linear correlation exists between SWI estimates and precipitation. SWI estimates and precipitation data followed similar trends. SWI is highly sensitive to the flood. During the flood a sharp increase in SWI occurs. If the flood conditions remain persistent for a substantial period of time, SWI also remains high throughout the period. Increase in SWI is always followed by the increase in NDVI but with a time lag of one to two months period which is required for the vegetations to grow. During the floods, NDVI decreases. The microwave remote sensing data used for the present study has very poor spatial resolution. Because of the poor spatial resolution the results obtained are applicable only for floods of large areal extent.