Agrometeorological Approach for Sugarcane Yield Estimation at Regional Scale Using Satellite Remote Sensing

Sugarcane is the major sugar producing crop of India and the prediction of sugarcane yield has immense importance for sugar industries. Satellite remote sensing has the potential to provide periodical information on the morphology throughout the growing season of sugarcane over a large area at various spatial scales and thus has the capability to be used for spatial and regional scale yield prediction. Agrometeorological indicators like crop evapotranspiration (ET), rainfall and crop response factors and the crop health indicators like spectral vegetation index from satellite data influence the crop water and drymatter content and hence can be used as the yield predictors. This study was carried out to generate these agrometeorological and health indicators from the available satellite image and develop yield models based on these indicators for estimating sugarcane yield at a regional scale. The study was conducted over 22 sugarcane growing districts of Uttar Pradesh, India. The bivariate regression model based on ET and Normalized Difference Vegetation Index representing the water relation and the spectral behavior of sugarcane crop was found to be the best (r = 0.79) having RMSE of 7.9% in terms of consistency and performances as compared to univariate models that account only spectral behaviour or only water relations. Moreover, the moisture availability in terms of ET when used along with the crop response factor based on the relative yield proved to be a better yield predictor than the ET alone. This study brought out the options to choose the model depending on the availability and accessibility of input data products from space-based optical and thermal remote sensing.