A statistical approach to map saline and fluoride (F) affected areas using optical and microwave remote sensing data simulations: A case study in Palacode Taluk, Tamil Nadu, India

Abstract

This research presents an approach for the indirect assessment of soil contamination by soil salinity and Fluoride (F) in varying environmental conditions, including both dry and wet seasons. The research was conducted in Dharmapuri District, utilizing Landsat 8 and Sentinel-1 data for analysis. The region exhibits a diverse soil regime, encompassing both dry-state and irrigated agriculture. The Modified Salinity Index (MSI) and Plant Senescence Reflectance Index (PSRI) were extracted from Landsat 8 imagery to assess contamination in dry conditions. Simultaneously, the Simplified Hallikainen's model and the Inverse of Dual-Polarization Soil Vegetation Index (DPSVI−1) were obtained from Synthetic Aperture Radar (SAR) data to investigate contamination in wet conditions. The correlation coefficient (CC) analysis was performed between optical data simulations (MSI and PSRI) and synthetic aperture radar (SAR) data simulations (Simplified Hallikainen's model and DPSVI−1), yielded R2 values of 0.90 and 0.92 in the dry and wet seasons, respectively, indicating a high degree of accuracy and reliability. From the correlation coefficient (CC) analysis, approximately 23% (173 km2) of the area exhibited negative correlation was identified as F affected regions and 54% (378 km2) of the areas exhibited positive correlation was identified as saline-affected regions in dry season. Similarly, approximately 41% (288 km2) of the area was identified as F affected regions and 46% (322 km2) of the areas was identified as saline-affected regions in wet season. The findings were verified through ground measurements of soil electrical conductivity (EC) and F using a confusion matrix. The results indicate an overall accuracy of 83% and 84%, accompanied by kappa coefficient measures of 0.79 and 0.83, for salinity and F, respectively, during the dry season. Likewise, in the wet season, there is a recorded overall accuracy of 86% and 88%, with corresponding kappa coefficients of 0.81 and 0.85 for salinity and F, respectively. This implies that the proposed methodology has a higher potential for mapping saline and F-affected areas in both wet and dry conditions. Hence, the methodology outlined in this study provides crucial information for local authorities, farmers, and policymakers, enabling informed decision-making and proactive measures.