Comparative analysis of different vegetation indices with respect to atmospheric particulate pollution using sentinel data

Abstract Several remote sensing based indices have been developed for studying vegetation using varied combinations of spectral bands. However, for regions like Gautam Buddha Nagar (GBN) district where the atmospheric condition is highly influenced by particulate matter loading and human-induced pollution, such indices produce faulty results because of the effects of scattering and absorption in the atmosphere. To reduce these impacts on vegetation indices, atmospheric correction becomes essential. The general objective of this study is to evaluate two non-atmospherically corrected vegetation indices viz. Normalized Difference Vegetation Index (NDVI); Soil-Adjusted Vegetation Index (SAVI) and two atmospherically corrected vegetation indices viz. Enhanced Vegetation Index (EVI); Atmospherically Resistant Vegetation Index (ARVI) with particular reference to GBN district which has high atmospheric aerosol presence. The spatio-temporal distribution of Aerosol Optical Density (AOD) and four vegetation indices for three months March (winter), June (summer) and October (post-monsoon) of year 2018 are analysed statistically. The mean AOD in all three seasons was observed to be above 0.05 and near to 1 indicating hazy conditions in the study area. Nine statistical image quality measures were employed such as Peak Signal to Noise Ratio (PSNR), entropy, Root Mean Square Error (RMSE), Relative Average Spectral Error (RASE), Standard Deviation (SD), Correlation Coefficient (CC), ERGAS, Mean Absolute Deviation (MAD) and Mean Absolute Percentage Error (MAPE), respectively to determine the best suitable index to study vegetation in highly polluted areas. The results indicated that the ARVI represents more enhanced vegetation information with respect to quantifying temporal variation of vegetation in all three seasons, especially in areas with high atmospheric particulate pollution.