Mapping rice residues burning and generated pollutants using Sentinel-2 data over northern part of India

Abstract

The study investigates the potential of Sentinel-2 imageries in spatio-temporal assessment of Rice residue burnt areas and generated pollutants in Kurukshetra, districts of Haryana, India. Datasets spanning from pre-burning (Sep.) to post-burning (Nov.) for the year 2018 were taken and variation in color and spectra of burnt and un-burnt surfaces were assessed. Two indices widely accepted for burned area assessment namely Normalized Difference Vegetation Index (NDVI) and Normalized Burnt Ratio (NBR) have been compared as numeric indicator for burnt surfaces. Dynamic thresholds developed using image statistics (over ground surveyed locations near to the Visible Infrared Imaging Radiometer Suite (VIIRS) fire events) was applied on time series indices for burnt area mapping. New plug-in model addable in ArcGIS software has been developed for the automatic burned area extraction. Short wave infrared composite (SIC) showed better visualization of active fire events. NBR was found better in spectral discrimination of burned and unburned surface with over all accuracy 91.38% as compared to NDVI (accuracy = 75.86%). Comparison of Sentinel-2B burnt area product for the year 2018 and 2019 with MODIS burnt area showed higher coverage under burning while with AWiFS it showed lower. Nearly 24.5% of total shown area (120800 ha) i.e. 30974.9 ha were found under stubble burning which contributed 138.69 Gg/yr air pollutants for the year 2018. The main reasons for the stubble burning obtained both from current study and literature were mechanized agriculture practices, poor storage facilities, farmers intension to get more profit from produces, easily available irrigation facilities, shift in rice cropping, and less time for next cropping.