Rice yield estimation using spectral and meteorological parameters

Abstract

With an aim to develop yield models in terms of spectral aod meteorological parameters for rice crop, field experiments are conducted in the wet seasons of 1984 and 1985 at the Mala Rice Research Station, Nawagam, Gujarat. Fertiliser, varietal and dates of planting treatments were included so as to induce variability in yield and to arrive at a model that can be used to predict yield under a wide range of growing conditions. Spectral data like radiance values in dilferent wavelength bands namely, visible and near infrared (NIR) and biometric data like Ieaf area Index, wet biomass and dry biomass are collected at weekly intervals over these plots. Regression analysis was done to study the relations existing between spectra1 and biometric parameters. A linear relation between leaf area index and NIR/red radiance ratio was observed. Linear relations were also found to exist between spectral parameters measured and grain yield. A linear regression relationship incorporating the radiance ratio aud a multilinear regression relationship incorporating integrated radiance ratio and accumulated temperature degree days were tried to estimate yield. The relationships developed for 1984 were tested for the year 1985. The relationship incorporating both the spectraI and meteorologica1 variables fared better.