Agricultural Drought Modeling Using Remote Sensing

Abstract

Maharashtra is one of the worst drought affected states in India, and as a result the agricultural productions has kept low. In the present study Standardized Precipitation Index (SPI) and Normalized Difference Vegetation Index (NDVI) obtained from MODIS data (MOD13Q1) and sunspot number which represents the solar activity has been considered for drought modeling in that part of Krishna Basin which lies in Maharashtra. Two multiple linear regression models for predicting agricultural drought are developed for Kharif season, based on the data for a period of 13 years from 2000 to 2012. In the first model, previous NDVI and SPI values and in the second model Sunspot data along with NDVI and SPI values are used as variables. It was found that the coefficient of determination of the second model has improved over the first model, which suggests the high significance of solar activity in the occurrence of drought. The correlation between NDVI and SPI has been utilized in the study. Highly significant correlations were obtained between current NDVI and SPI of various time lags in the assured rainfall zone and the scarcity Zone. Crop yield model was developed from the predicted NDVI for the major crops and was validated with the actual yield. It was found that the predicted NDVI of both the fortnights of July is highly correlated with the yield of major crops in assured rainfall zone and scarcity zone. Index Terms-Agricultural drought, crop yield prediction model, MODIS NDVI, sunspot numbers.