|
|
|
Evaluation of MODIS Derived Normalized Difference Water Index for Agricultural Drought Assessment
Dr V.K. Sehgal
Senior Scientist
Indian Agricultural Research Institute,
India
Email: sehgal@iari.res.in; vksehgal@yahoo.com
A. Chakraborty
PhD Scholar
Indian Agricultural Research Institute
Dr R.N. Sahoo
Scientist (SS)
Indian Agricultural Research Institute
rnsahoo@iari.res.in
Globally, drought is the biggest natural disaster and is also of common recurrence in various parts of the India affecting agricultural production. Early detection is very important for mitigation of drought, which requires measures of its quantification at regular time interval in real-time. Satellite remote sensing derived vegetation indices viz. NDVI (Normalized Difference Vegetation Index) are regularly used for regional estimation and early warning of agricultural drought. The NDVI suffers from two shortcomings for agricultural drought assessment, (a) there is a considerable time lag between the rainfall deficit and NDVI response and (b) discrepancy between the level of penetration of red and infrared bands in crop canopy as red band don’t penetrate beyond 2-3 leaves layer. With the availability of images from MODIS sensor aboard TERRA AM and PM, it is now possible to generate Normalized Difference Water Index (NDWI) which uses both the bands in near infrared region (0.86 and 1.24 or 0.86 and 1.64 and thus has advantages over NDVI. The NDWI is very sensitive to liquid water content of vegetation canopy thus making it a very suitable index for mapping and monitoring of agricultural drought. In the present study, an attempt has been made to evaluate the sensitivity of MODIS derived NDVI and NDWI to crop water stress, response of NDVI and NDWI to current rainfall and potential of multi-date NDWI images in delineating drought affected areas to monitor the progress of the agricultural drought. The study was conducted for the western part of India which was severly affected by drought in the year 2002 using MODIS 09A1 8 days composite data product of surface reflectance from June to October. The NDVI and NDWI maps for the kharif season of year 2002 (drought year) and 2003 (normal monsoon) were generated. The study used daily rainfall product, which is generated by NOAA Climate Prediction Centre (CPC) at a grid size of 0.1ºX0.1º. Daily Rainfall products were aggregated to 8 day cumulative and were then brought to the resolution of MODIS data by oversampling. It was observed that NDWI has a strong relation with current rainfall than current NDVI. The spatial comparison of changes in NDVI and NDWI between the two years for each 8 day period showed that NDWI was better than NDVI in picking up the intensity and extent of drought. Thus, NDWI based techniques are more promising in assessing agricultural drought at regional scales.
|
|
|