Cotton yield estimation using agrometeorological model and satellite-derived spectral profile

Abstract

Timely and accurate estimates of district-level cotton yield are essential for management decisions related to the cotton economy of a country. Surendranagar is a major cotton growing district of Gujarat, India. The yield of cotton for Surendranagar district was estimated using (i) a linear time series trend and actual evapotranspiration (AET) based model for irrigated cotton and an empirical model between the AET and the yield for unirrigated cotton; and (ii) the NDVI (Normalized Difference Vegetation Index) spectral profile area, using data from the LISS I sensor on-board the Indian Remote Sensing (IRS) satellite. AET was estimated using a simple agrometeorological soil moisture balance model for 11 years (1983-1993). The cotton lint yield estimates from the above two different approaches had a 1.2% relative deviation between them and differed by -14.4 and -13.1%, respectively, from the yield estimates given by the Department of Agriculture, Gujarat.