Modeling Kharif Crop Yield Using NDVI and Artificial Neural Network in Raipur, Chhattisgarh, India

The estimation of crop yield is a key component in the planning and management of various crops during different seasons by farmers, agricultural industrialists, and policymakers. The traditional techniques for estimation of crop yield are cumbersome and laborious which requires field surveys. The present study aims to identify the method through which the estimation of the crop yield could be possible with minimum effort without field survey using satellite-based meteorological-climatic factors during the cultivation period of the specific crop. For this purpose, first, the relationship between the crop yield and meteorological and agricultural indices, namely, the Standard Precipitation Index (SPI) and Normalized Difference Vegetation Index (NDVI) is assessed using the correlation analysis and then using a popular Artificial Neural Network (ANN) technique, the crop yield for Kharif crop in Raipur, Chhattisgarh, India is estimated. For the estimation of NDVI for water and vegetation cover during the Kharif season, LANDSAT 5, 7, and 8 images for thirty years were utilized. The ANN model performance in estimating Kharif crop yield is assessed using the four standard statistical measures. The results obtained from the study suggest that Kharif crop yield is mostly related to the NDVI values representing the vegetation cover and the ANN technique comes out to be one of the potential methodologies for crop yield estimation.