Integrating S1A microwave remote sensing and DSSAT CROPGRO simulation model for groundnut area and yield estimation

This study sought to corroborate microwave remote sensing and simulation models to efficiently delineate groundnut cultivation area and to estimate the yield by integration. Near real-time information on crop acreage and yield estimation is essential for making policy decisions. S1A SAR data were downloaded for entire crop growth period of groundnut during Kharif monsoon seasons (June – October) of 2019 and 2020 and were processed using MAPSCAPE RIICE software to extract groundnut cultivated area in the study districts of Tamil Nadu. Spectral dB curve groundnut generated using multi-date Sentinel 1 A SAR data showed a minimum at sowing, reached a peak at the pod development stage and decreased after that towards maturity. Groundnut area map was generated with a classification accuracy of 85.2 and 84.8 per cent with a kappa coefficient of 0.70, and total groundnut area of 104343 and 116199 ha was mapped during Kharif monsoon season 2019 and 2020, respectively. The mean agreement of 75.01 and 84.94 per cent was observed between DSSAT model simulated LAI and observed LAI at thirty monitoring locations in the study area during Kharif monsoon season 2019 and 2020, respectively, whereas agreement for yield was 82.11 and 83.70 per cent with RMSE of less than 20 per cent. Spatial distribution of groundnut LAI and yield was estimated by assimilating dB from satellite image and from DSSAT model, respectively. The estimated mean spatial LAI was 2.81 and 3.52, whereas mean spatial pod yield was 2124 and 2195 Kg ha⁻¹ during Kharif monsoon season 2019 and 2020, respectively with RMSE of less than 20 per cent and R² for integrating satellite products and simulation model for spatial estimates during both the year was >0.70, it shows the fitness of products towards increased accuracy of estimation.