Mapping and monitoring of food legumes and dryland cereal production systems

Mapping and monitoring of the agricultural production systems on a regular interval provide important spatial matrix on the status, trend, and options for effective intervention at multiple scales. The recent advances in agro-geoinformatics big-data enriched with increasing open-access protocols become an integral part of solving the food security equation. This paper demonstrates use of an integrated earth observation system (EOS) for mapping and monitoring major agricultural production systems. The approach uses multi-temporal and multi-scale remote sensing data coupled with in-situ observation to map the legume and cereal production systems. The support vector machine (SVM) classification was found to be the best with overall classification accuracy of 82%. The in-situ data on crop grain and straw yields were measured using nested sampling approach. The best fit equation of yield values were regressed with remote sensing indices (NDVI and EVI). The significant correlation (R²) value of cereal and lentil crop were 0.74 and 6.9 at p<0.01 respectively. The R² value between observed yield and predicted yield was 0.80 and 0.97 in cereal and lentil crops respectively. The predicted yield based on remote sensing data varies from 3,303 to 5,710 kg ha⁻¹ and mean yield is 3,840 kg ha⁻¹. The productivity of the cereal crop was varies from 4228 kg ha⁻¹ to 4598 kg ha⁻¹ while lentil crop was between 304 to 1,500 kg ha⁻¹. The huge inter and intra field variably was observed through the study areas. Such information yielded vital information about yield gaps exists within and across the fields. Study is in progress to develop systematic and semi-automated algorithms to map and monitor the agricultural production on regular interval to quantify the changes in the cropping pattern, rotation, production and impacts of the technological interventions and ex-ante - nalysis.