Mapping Spatial Variability of Soil Salinity Using Remote Sensing Data and Geostatistical Analysis: A Case of Shadegan, Khuzestan

Extended abstract 1- Introduction Soil salinity is one of the most important desertification parameters in many parts of the world. Thus, preparing soil salinity maps in macro scales is necessary. Water and soil salinity as one of the contributing parameters in desertification, cause soil and vegetation degradation. Soil salinization represents many negative effects on the earth systems such as water and wind erosion, increasing dust storms, removing vegetation, reducing production capacity of the soils, etc. Most of the saline soils are located in the regions with hot and dry climates like Iran. One of the ways to combat desertification phenomenon is understanding effective factors in intensification. On the other hand, soil salinity measurement in laboratory is costly and time-consuming, especially in the large-scale regions. Spatial interpolation methods and satellite images interpretation can be used to map soil salinity with high accuracy in both temporal and spatial resolution. Remote sensing and geostatistics can play an important role in identifying the phenomena, mapping, time monitoring changes, controlling, modifying and finally managing soils salinity. The purpose of this study is soil salinity zonation and its trend investigation using remote sensing data and geostatistical techniques in Shadegan area.   2- Methodology Geostatistical techniques are generally used for spatial changes and they are useful for soil salinity investigation and results can be more valuable when they are coupled to remote sensing data. Interpolation methods can be done by many GIS programs and also remote sensing can be a useful tool for collecting the earth data of a broad area in a short time. These applications are more useful for impassable, dangerous and wide areas. Electromagnetic wave reflections are different in various lands and this is the basic principle of using satellite images for landscape interpretation. For mapping the soil salinity in the study area, 54 soil samples were used which have been sampled in 2006 using interpolation methods with the maximum likelihood of mapping the soil salinity. Some descriptive statistical analyses (e.g. mean, mode, variance, standard deviation, kurtosis, skewness) and the normality of the data were conducted using SPSS software. The interpolation methods including deterministic methods and geostatistical methods were used for mapping the soil salinity in ArcGIS software. In this study, deterministic statistical methods such as inverse distance weighted, global polynomial, radial basis functions and geostatistical methods (ordinary kriging and simple kriging) for soil salinity mapping were evaluated for the region. For this goal, the remote sensing data (bands, salinity indices and principal component analysis) were used for the satellite image of ETM+ from the nearest time to the sampling time, namely 2006. For studying the correlation between brightness of the pixel values and soil samples, the regression with the highest correlation with the sample points, was selected as the suitable method between the fit method and soil samples to establish the regression equation. Finally, regarding to the sampling points, supervised classification was used. Then, regression obtained equation was used for salinity map in 1990 and 2015 for soil salinity trend analysis investigation. 3- Results The results showed that the simple Kriging interpolation has the higher accuracy than the other methods for mapping the soil salinity. Among the geostatistical methods, simple kriging and ordinary kriging are similar in terms of accuracy, but the simple Kriging with spherical semivariogram model, compared to the other methods of soil salinity zonation is more appropriate in the study area. Study of the methods of salinity map showed that in 2006, the PCA123 method has the highest correlation with the sampling point compared to the real map of soil salinity. Trend analysis of soil salinity in 1990, 2006 and 2015 showed that the area of average and high salinity are reduced but the area of very high salinity‌ increased sharply from 1990 to 2015. On the other hand, the area with medium and high salinity classes has decreased and closed to zero in 2015, but the extreme salinity class has increased about 2.5 times more. 4- Discussion & Conclusions It can be concluded that satellite images, remote sensing data and geostatistical techniques are reliable tools for soil salinity studies. Increased soil salinity in the study area shows that the intense salinity of the southern part to the northern parts has slowly been moved. Abadan and Mahshahr in Khuzestan province are considered coastal areas except Shadegan city located in the southern part. In the coastal areas with low slope, transition of salt from the sea to the coastal area is acceptable but for other regions, other reasons are needed. Finally, it is suggested to use these methods and techniques for soil salinity investigation for the similar areas.