Rice Paddies Mapping and Yield Estimating using Satellite Images and Remote Sensing Techniques (Case study: Kunduz province, Afghanistan)

Investigating the area under cultivation and yield estimation of agricultural products like rice; can greatly ensure food security, analyze the status of agricultural products and, finaly, sustainable development of developing countries. This study used Sentinel-2 satellite images, to estimate the area under cultivation and the yield of rice paddies in Kunduz province, Afghanistan in the 2020 crop year. Using the time series of NDVI index, the phenology stages of rice plants were obtained and the phenology parameters (SoS and EoS) were extracted using the maximum resolution method. Then, object-oriented classification method based on phenology was used to identify and determine the under-cultivated area of rice fields. In this method, three types of data of reflectivity of reflective bands, NDVI vegetation index and phenology parameters were used as auxiliary data. Yield estimation was done using the experimental method of regression analysis between remote sensing plant indices and the data obtained from ground harvesting. Also, the experimental method based on the regression analysis of ground data and distance measurement with the coefficient of determination of 0.86 and the Pearson correlation coefficient of 0.92 showed its high accuracy in estimating the yield of rice fields. The accuracy of the estimated performance in this research was evaluated by comparing the actual performance (field harvest data) in 27 control points. For this purpose, Pearson's correlation test was used. This test showed that there is a positive and very strong relationship between actual performance and estimated performance (P=0.000, N=27 and R2=0.929).