Application of phenology-based algorithm and linear regression model for estimating rice cultivated areas and yield using remote sensing data in Bansloi River Basin, Eastern India

Abstract

Delineation of paddy (Oryza sativa L.) cultivating areas using modern technologies like remote sensing and geographical information system is an important way for managing food security, agricultural production and water resource. This paper aims to identify the Aman paddy cultivating areas and estimate yield of Aman paddy using phenological algorithm based on different vegetation indices extracting from Landsat OLI images in Bansloi River basin, Eastern India. Different vegetation indices (viz. NDVI, EVI, LSWI and RGVI) were used to identify paddy cultivating areas and yield. Phenological phases of paddy growth were recognized considering daily air temperature, rainfall and field survey data. After running the linear regression between different vegetation indices and paddy yield collected during the field survey, the paddy yield was predicted for the identified paddy cultivated areas. The estimated paddy cultivated areas were validated using user, producer, overall accuracy and kohen's kappa test. On the other hand, predicted paddy yield was validated through RMSE and ME statistical techniques. All the validating methods are showing good accuracy of estimated paddy cultivated areas and predicted yield. The recognized total rice cultivating area, average yield and total production was 107173 ha, 2.51 tons/ha and 269004.23 tons respectively from 2015 to 2017. The estimation of paddy yield before harvest can provide valuable information to a planner about export, import and storage of food for the future.