Remote sensing leaf area index (LAI) data assimilation with crop model for yield predictions in rice

Crop yield estimation has gained prominent importance due to its vital significance for policymakers and decision-makers in enacting schemes, ensuring food security, and assessing crop insurance losses due to biotic and abiotic stress. Precise and timely crop yield estimates at regional, national and international levels is essential for making policy to overcome food security worldwide and helping farmers for crop insurance through insurance premium pricing by the companies. Rice is considered the major staple food which is having highest area and production in India. Telangana contributes to 4.49 % of rice area (1.9 million ha) and 5.54 % of production (6.25 million tons) with a productivity of 3176 kg ha-1. Several studies revealed that remote sensing technology had resulted in higher accuracy in crop growth monitoring with added advantage of high revisit frequency and precision. On the other hand, crop simulation models were also been recognized to assess the effects of different scenarios like climate change, drought, stress etc., on crop yield under varied climatic conditions. LAI is main criterion for evaluating the grain yield as it shows good correlation with the grain yield. There are lack of studies on comparing the ceptometer LAI to any crop model simulated LAI and also yields estimation at local level though they were done at a broad level like state or district. Hence this research was focused on rice yield estimation at the field level in the Karimnagar district of Telangana during 2021 and 2022 by employing the leaf area index (LAI) as the primary criterion for integrating remote sensing technology and crop simulation models. Optimization of crop cutting experiments were performed based on the criterion encompassing a wide range of potential combinations, further four villages each in Kharif and Rabi were selected for study and 15 fields were selected in each village for study. Ground data visits were planned according to the satellite passing dates and during the visits LAI readings in each field were collected using the LP-80 ceptometer. Supervised classification was performed using the ERDAS imagine. It has been noted that most of the area in the district was occupied by rice in both the seasons. Accuracy showed that overall accuracy of 94.23% and 88.5% was recorded, while kappa coefficient of 0.89 and 0.85 was resulted in kharif and rabi season respectively. On an average, kharif and rabi rice grain yields were 5324 kg ha-1 and 6436 kg ha-1 respectively in selected villages. The average simulated rice grain yield in kharif and rabi were 5339 kg ha-1 and 6858 kg ha-1 respectively with DSSAT model which considered sentinel-2 satellite for estimation of LAI. The R2 values of above 0.72 in kharif and above 0.85 in rabi, D index of 0.70 in both the seasons in all the villages showed the model is accurate for predicting yields. In both the seasons, correlation of above 0.8 was observed between observed rice grain yield with the quantity of nitrogen applied, whereas above 0.77 was noted between ceptometer measured and model simulated LAI. However LAI showed a good R2 of above 0.75 with the grain yield. Due to its strong correlation with LAI of above 0.80, the Normalized Difference Vegetation Index (NDVI) was selected as the critical element for integration with the model. Hence, it can be noted that NDVI is one among the important parameter which can be used to integrate with LAI for grain yield estimation. By utilizing the linear equation generated between the NDVI and model LAI a spatial LAI map was generated for the Karimnagar district. Further the linear equation developed between the model LAI and model grain yield, spatial yield map was generated. From the spatial yield map, it can be concluded that most of the areas fall under the rice grain yield range of 5700 to 6000 kg ha-1 in kharif, while in rabi in the range of 6500 to 7000 kg ha-1. These spatial mean yields for kharif and rabi were 5300 kg ha-1 and 6458 kg ha-1 which were then compared with the Telangana government statistics and it has been noted that a deviation of less than 10 %. Therefore, this study’s findings show that assimilating remote sensing data with crop models enhances the precision of rice yield prediction for insurance companies and policy- and decision-makers.