Iranian Remote Sensing and GIS Society / Shahid Beheshti University Iranian Journal of Remote Sensing and GIS 2008-5966 13 4 2022 02 20 Evaluation of the Efficiency of Spectral Data and Indices Derived from OLI and TIRS Sensors in Estimating Soil Salinity in Arid Regions of Southern Ilam Province Evaluation of the Efficiency of Spectral Data and Indices Derived from OLI and TIRS Sensors in Estimating Soil Salinity in Arid Regions of Southern Ilam Province 51 66 101573 10.52547/gisj.13.4.51 FA Hamidreza Matinfar Associate Prof. of Soil Science, College of Agriculture, Lorestan University, Lorestan Foziyeh Kohani Ph.D. Student of Soil Science, College of Agriculture, Lorestan University, Lorestan Ali Akbar Asilian Mahabadi Graduate Master of Soil Science, College of Agriculture, Ilam University, Ilam Journal Article 2020 04 21 Soil salinity is one of the most important environmental problems, and the identification and zoning of saline soils is difficult due to the need for sampling and laboratory analysis, as well as having temporal and spatial variability. In recent years, the use of satellite imagery has always been of interest to experts due to its ease of use and ability to detect phenomena. Remote sensing information greatly aids the study of soil salinity and can be helpful in identifying salinity values. In this study, 220 soil samples were collected from Meymeh area of Dehloran, south of Ilam province, according to the type of study and physiographic types and soil units. Then, pH and EC values were measured using standard methods. Soil salinity values were evaluated using correlations between EC electrical conductivity values obtained from ground data and variables obtained from Landsat 8 satellite images including bands, salinity indices, vegetation indices and soil indices. Finally, the soil surface salinity estimation model was obtained using stepwise regression method. This method involves the automatic selection of independent variables, and with the availability of statistical software packages, it is possible to do so even in models with hundreds of variables. In previous studies, indicators and bands have been used separately and in a limited way, but in this study, an attempt has been made to use a combination of different indicators more widely, and finally to achieve the best relationship by eliminating the indicators that have the least impact on soil salinity estimation. Using significant level analysis and correlation between the output of models and ground data, the best model with a value of (R2 = 0.882) was selected and a soil salinity map was prepared based on it. In the study area, the highest area belonged to non-saline class which comprises 75% of the total study area and about 1% of the soils belong to the saline class. Soil salinity is one of the most important environmental problems, and the identification and zoning of saline soils is difficult due to the need for sampling and laboratory analysis, as well as having temporal and spatial variability. In recent years, the use of satellite imagery has always been of interest to experts due to its ease of use and ability to detect phenomena. Remote sensing information greatly aids the study of soil salinity and can be helpful in identifying salinity values. In this study, 220 soil samples were collected from Meymeh area of Dehloran, south of Ilam province, according to the type of study and physiographic types and soil units. Then, pH and EC values were measured using standard methods. Soil salinity values were evaluated using correlations between EC electrical conductivity values obtained from ground data and variables obtained from Landsat 8 satellite images including bands, salinity indices, vegetation indices and soil indices. Finally, the soil surface salinity estimation model was obtained using stepwise regression method. This method involves the automatic selection of independent variables, and with the availability of statistical software packages, it is possible to do so even in models with hundreds of variables. In previous studies, indicators and bands have been used separately and in a limited way, but in this study, an attempt has been made to use a combination of different indicators more widely, and finally to achieve the best relationship by eliminating the indicators that have the least impact on soil salinity estimation. Using significant level analysis and correlation between the output of models and ground data, the best model with a value of (R2 = 0.882) was selected and a soil salinity map was prepared based on it. In the study area, the highest area belonged to non-saline class which comprises 75% of the total study area and about 1% of the soils belong to the saline class.

https://gisj.sbu.ac.ir/article_101573_d2a63359aabd363d1e098a99e4665c19.pdf