Characterization of soil salinity and its impact on wheat crop using space-borne hyperspectral data

Abstract

Facing the risk of soil salinization worldwide, there has been a growing interest in identifying rapid and inexpensive tools for soil salinity assessment. Remote sensing has shown great advantages in the field in recent decades. In present research, Hyperion Hyperspectral remote sensing data (EO-1) was used for characterization and mapping of salt-affected soils, to generate crop inventory map and to evaluate soil salinity impact on wheat crop growth in part of Mathura district of Uttar Pradesh representing Indo-Gangetic plain. Narrow bands can discriminate critical spectral differentials in detail and can assess the salinity hazard over crop. A detailed field survey was carried out in the study area in order to identify the salt-affected soils and to collect soil samples, groundwater table depth, chlorophyll content, LAI to characterize impact of soil salinity over crop. Various wheat crop spectra were collected for calculation of narrow band indices to discriminate various stress conditions. Spectral angle mapper (SAM) was used to generate crop inventory map with various types of crops. The same technique (SAM) was used to map various categories of salt affected soils represented by spectral endmembers of normal, slightly, moderately and highly salt-affected soils. The results showed that various severity classes of salt-affected soils could be reliably mapped using spectral angle mapper (SAM) analysis with an overall accuracy of 74.24%. Empirical relationships developed between crop & soil parameters and vegetation indices using SMLR could show its possibility with an R2 of 0.52 and 0.41 to predict LAI and CCI, respectively. Validation results showed the RMSE of 0.8 and 5.2 to predict LAI and CCI. Partial least square regression (PLSR) statistical model (using spectroradiometer derived narrow band indices) was developed to assess different stress level with varying crop and soil parameters.