Using Technology at Gram Panchayath Level
We implemented technologies based on the Terrestrial Observation and Prediction System (TOPS) for estimating GP level yields of rice (Oryza sativa) in 25 districts across 6 states of India. We used a combination of Synthetic Aperture Radar and optical data to map rice extent. We adopted a light use efficiency (LUE) model to estimate crop yields.
To run the model, we generated daily 2-km grids of input weather conditions using a machine learning algorithm that incorporated station observations, satellite data, and reanalysis model outputs. Required crop biophysical estimates of Leaf Area Index (LAI) and the Fraction of intercepted Photosynthetically Active Radiation (FPAR) were derived using daily cloud-screened MODIS 250-m data from Terra and Aqua satellites and a modified MOD15 LAI/FPAR backup algorithm. We used the crop extent maps, daily climate, gap-filled FPAR and the LUE model to estimate above-ground biomass, which was accumulated over the growing season and converted to crop yields using a harvest index. Extensive field work (2250 field surveys and 4000 CCEs) was conducted in support of the validation effort. In each of the 25 districts, we conducted 90 ground surveys in support of crop mapping and 160 CCEs to verify modelled yields.
⦁ Map the rice extent in each of the 25 districts.
⦁ Estimate rice yields in each of the 25 districts.
⦁ Assess the accuracy of crop extent and crop yields using field observations.
⦁ Summarize crop extent and crop yields by village in all 25 districts.
We implemented technologies based on the Terrestrial Observation and Prediction System (TOPS) for estimating GP level crop yields of rice (Oryza sativa) in 25 districts in 6 states. We used a combination of Synthetic Aperture Radar and optical data to map crop extent, validating the resulting maps against field surveys. We adopted a light use efficiency (LUE) model adapted from MODIS (Moderate Resolution Imaging Spectroradiometer)- algorithm (MOD17-GPP/NPP) to estimate crop yields. To run the model, we generated daily 2-km grids of input weather conditions using a machine learning algorithm that incorporated station observations, satellite data, and reanalysis model outputs. We used the crop extent maps, daily climate, gap filled FPAR in the LUE model to estimate above-ground biomass, which was accumulated over the growing season and converted to crop yields using a crop-specific harvest index.