The land use impact evaluation on soil properties through remote sensing in an arid area in southern Iran

In this study, to examine the images of Google Earth Engine, soil samples were randomly taken in different parts of the Roudan basin. PH, water soil, SOC, clay, and Sand experiments on samples were done. The use of GIS and GEE applications is suitable based on the results obtained for the review and evaluation of different parameters of land cover/land use. This study confirmed the use of GEE techniques in the analysis of land use changes in the Roudan basin by using different components of GIS. Quantitative data production about the parameters of soil and land use was possible. Changes in soil organic carbon content cause apparent specific gravity, and also show statistically significant differences between land uses. Increasing the amount of soil organic carbon from 0.5 to 3% in all soil texture classes increases the usable water capacity by more than 2 times; Of course, for pastures and road uses, the value of this index is limited. The results thus show that the amount of soil moisture is highest in farmland and garden land. However,  the depletion of soil organic carbon in agricultural land can be attributed to the erosion of soil structure caused by repeated cultivation and the removal of crop leftovers. Soil organic carbon is the result of the deposition of vegetation residues. The organic carbon levels decrease with continuous farming on agricultural lands (Dor et al, 2023). The decline in organic carbon levels could result in soil deterioration and reduced fertility in the long run. Overall, the LULC situation showed that the Garden and water areas' soil samples have the highest correlation with images. The use of ground truth datasets ensures the reliability of the classification results, with overall accuracies of 95% and a Kappa coefficient of 0.93. This agrees with findings by Ghassemi et al (2022b), Rosina et al(2018) and, Congalton and Green (2019). The Poor Rangeland achieved the highest producer’s accuracy (98.4%), while Water areas had the lowest user accuracy (87.5%) due to cross-class confusion (Foody, 2020). So Woldeamlak and Solomon (2013) reported that removing vegetation reduces the recycling of organic carbon in the soil. Soil organic carbon is the result of the deposition of vegetation residues. Findings Doa et al (2011) suggested that land use changes can influence soil properties. Changes from a garden or forest to rangeland or bare soil type remove the addition of litter that decreases the nutrient content of soils and increases rates of erosion and loss of soil organic matter, these results agree with Rawat and Kumar (2015). Google Earth Engine presents a new platform that can be integrated into a Traditional Digital Soil Mapping process. It is precisely designed to deal with vast datasets, vital in DSM, for mapping soil characteristics. One main advantage of using GEE is that there are lots of rasters that can be easily assessed and used to make both data gathering and application of an algorithm much easier, hence increasing computational efficiency to a huge degree. Whereas GEE is under constant development and has its shortcomings, it is not a strictly digital soil mapping software. Rather, it is a tool for mapping the wider environment, and its potential for the soil science and environmental community is vast.

Soil is an essential component of terrestrial ecosystems and plays a crucial role in the cycling of energy and materials between the atmosphere and the biosphere. The mapping of soil parameters such as soil organic carbon (SOC) and soil moisture has advanced through the rapid growth of Earth observation data (e.g., Sentinel) collection and the advent of appropriate tools such as the Google Earth Engine (GEE). This study aimed to assess soil parameters derived from GEE imagery and examine their relationship with topsoil properties across various land-use types in the Roudan Basin, located in Hormozgan Province, southern Iran. To this goal, to measure characteristics such as sand and clay content, pH, SOC, and soil moisture (SM) in the laboratory for 205 topsoil samples (0-10 cm) collected from various land uses. to assess the accuracy of the land use maps in the region using Copernicus satellite imagery, which has a spatial resolution of 100 meters. Statistical methods Correlation coefficient (r), mean absolute error (MAE), and root-mean-square error (RMSE), were employed to analyze soil samples to GEE image outputs. Results showed estimation accuracy varied by property and land use, so that soil moisture and clay content were reliably estimated with values r = 0.92 and r = 0.89 in farmland, whereas pH and sand estimations were less accurate with values r = 0.33 and r = 0.55 in farmland. Spatial analysis using GEE demonstrated that farmland and gardens retained the highest soil moisture, attributed to high clay content and low sand presence, which enhanced water retention and reduced drainage risk. Correlation analysis revealed that SOC is positively associated with clay content (r = 0.82) and negatively with sand content and moisture (r = -0.89), emphasizing textural influences on soil carbon dynamics. The result of land-use classification showed high accuracy (95%, Kappa = 0.93), with poor rangeland demonstrating the least classification error. However, water areas exhibited higher misclassification rates, attributed to seasonal fluctuations.

Keywords: Land Use, soil Property, soil organic carbon, Google Earth Engine, Roudan Watershed