EVALUATION OF PRE-HARVEST PRODUCTION FORECASTING OF MUSTARD CROP IN MAJOR PRODUCING STATES OF INDIA, UNDER FASAL PROJECT
Keywords: FASAL, Rapeseed-mustard, Maximum Likelihood Classifier, NDVI, Semi-Physical Model, RMSE and t-test
Abstract. Rapeseed-mustard (Brassica spp.) is the major rabi oilseed crop of India. India is fourth largest contributor of oilseeds and Rapeseed-mustard contributing to around 11% of world’s total production and about 28.6% in total oilseeds production of the country. More than 85% Rapeseed-mustard production comes from 5 States viz. Rajasthan [48%], Haryana [12%], MP [10%], UP [9%] and West Bengal [7%]. In the previous few years, remote sensing technique has been progressively more considered for evolving as an alternative, standardized, possibly cheaper and faster technology for crop acreage estimation. Furthermore, satellite remote sensing data have strong advantages in comparison with other monitoring techniques because it provides timely, synoptic and latest information of crop at various stages over large scales. Therefore, under FASAL project, cloud free crop season’s images of different satellites (Sentinel-2, Resourcesat-2 and Landsat-8) were used and mustard crop was discriminated using Maximum Likelihood Classifier (MLC). Yield was estimated using different methods such as remote sensing derived NDVI, Agrometeorological yield model and Semi-Physical Model. The RMSE values for state level were found to be 4–17%, 8–19% and 13–23% for area, yield and production, respectively. The correlation coefficient (r) between DES and FASAL estimates were close to 0.9 in all the cases. The results of t-test at 5% level of significance inferred that FASAL and DES results were not significantly different. These results show that RS and weather-based techniques can be effectively used for pre-harvest acreage, yield and production estimation of mustard crop at district, state and national level.