Combination of Spectral Indices of OLI and TIRS Sensor and Magnetic Induction Data to Estimate Spatial Variation of Soil Salinity

Soil salinity and alkalinity are believed to be among the most important soil degradation processes, specifically in arid and semi-arid regions. identification and monitoring of salinity is also necessary for land and environmental management. continuous salinity monitoring is traditionally costly and time-consumingwhile using satellite data and combining them with magnetic induction data could be an alternative to the traditional method. the present study aimed to evaluate spectral indicators and identify saline soils and spatial changes via the data of the 2em38. the study area was conducted in ghahavand plain located in hamedan province. in this study, landsat 8 satellite data were used. the image was georeferenced with the image-to-map method with more than 12 control points and accuracy of fewer than 0.4 pixels. atmospheric correction was performed through the black body method. soil sampling of 37 points was performed and 86 points were read using an electromagnetic induction device. the soil samples were transferred to the laboratory and passed through a 2-mm sieve. the sample analysis was performed according to standard methods. using protomorphic units based on visual interpretation of oli 543 false-color composite image and field observations, a total of nine homogeneous units were identified in the region using these units as training regions for supervised classification. the results revealed that the detection of soil salinity in the visible spectrum (blue, green, and red bands) is feasible. bands 5, 6, and 7 could be useful for differentiating salty white crust lands from salty gray crust lands. in the reflective bands, the white and smooth crust lands exhibited the highest reflectance. the results of classification accuracy showed that the highest total accuracy was 90.0 and the kappa coefficient was 80.45 when bands 2, 3, 4, 5, 6, 7, 10, and 11 were used and shallow and abandoned plowed lands had the lowest accuracy. furthermore, the final model of salinity estimation implied that si6 and si11 indicators and electromagnetic induction vertical measurements (emv) were the most suitable variables for estimating salinity spatial changes.