JWSS - Journal of Water and Soil Science

Soil depth is a major soil characteristic commonly used in distributed hydrological modeling in order to present watershed subsurface attributes. It strongly affects water infiltration and accordingly runoff generation, subsurface moisture storage, vertical and lateral moisture movement, saturation thickness and plant root depth in the soil. The objective of this study is to develop a statistical model that predicts the spatial pattern of soil depth over the watershed from topographic and land cover variables derived from DEM and satellite image, respectively. A 10 m resolution DEM was prepared using 1:25000 topographic maps. Landsat8 imagery, OLI sensor (May 06, 2015) was used to derive different land cover attributes. Soil depth, topographic curvature, land use and vegetation characteristics were surveyed at 426 profiles within the four sub-watersheds. Box Cox transformations were used to normalize the measured soil depth and each explanatory variable. Random Forest prediction model was used to predict soil depth using the explanatory variables. The model was run using 336 data points in the calibration dataset with all 31 explanatory variables (18 variables from DEM and 13 variables from remote sensing image), and soil depth as the response of the model. Prediction errors were computed for validation data set. Testing dataset was done with the model soil depth values at testing locations (93 points). The Nash-Sutcliffe Efficiency coefficient (NSE) for testing data set was 0.689. The results showed that land use, Specific Catchment Area (SCA), NDVI, Aspect, Slope and PCA1 are the most important explanatory variables in predicting soil depth.

The soil quality is defined as the ability of soil to function as an essential part of the human habitat. In this study, the effects of land use change (conversion of forest lands into agricultural lands) on the soil physical quality were studied in the Gilan-e-Gharb region. For this study, soil samples were collected from surface and subsurface layers of both land uses, and the peak and shoulder slope positions, in Miandar and Vidjanan catchments. Soil physical properties such as soil texture and particle size distribution, soil hydraulic conductivity, bulk density, mean weight diameter of aggregates, water holding capacity, and the soil organic carbon content were measured. The results showed that land use change of the forest to agricultural lands resulted in a sharp decline in the soil organic matter (52%) and an increase in silt and sand percentage and soil bulk density. Also, deforestation decreased the mean weight diameter of aggregates (from 0.39 to 0.14 mm in Miandar) and clay percent.  It caused a reduction in the total porosity followed by a decrease of soil water holding capacity, and a decrease in the saturated hydraulic conductivity (from 10.34 to 1.86 cm/h), as well. In general, the results proved that the land use change from forest to agriculture severely decreased soil physical quality and its productivity.

The effect of broad-leaved forest trees (Alnus glotinusa, Ulmus glabra, Popolus caspica and Parrotia persica) and their canopy position on soil C and N storage and mineraization in the plain forest areas of Noor was investigated. Soil samples were taken from two positions (near and away from the main stem) with the microplots of 30×30×15 cm. Litter (C and N), soil physical (bulk density, texture and water content), chemical (pH, EC, organic C, total N and available Ca), biochemical and biological (N mineralization and microbial respiration) characteristics were measured at the laboratory. Carbon mineralization rate (CMR) was calculated using the equation [incubation time period (hour) ×soil volume (gr) / CO₂ amount (mol C)]. Soil C and N storage (ton/ha) was calculated by C and N contents, bulk density, and the soil sampling depth. The results showed that there was no significant difference between the C storage under the studied tree spcies, whereas N storage presented significantly greater amounts, under Alnus glotinusa (0.79 ton/ha) rather than Ulmus glabra, Popolus caspica and Parrotia persica (0.69, 0.45 and 0.21 ton/ha, respectively). The higher values of soil C (0.001 mol C/kg) and N (0.3 ml N/kg) mineralization were significantly recorded under Alnus glotinusa instead of tree species. Soil C and N storage and mineralization process were not affected by the sampling positions. According to the results, soil C and N storage and mineralization were influenced by litter quality and soil chemistry.

Land use/cover maps are the basic inputs for most of the environmental simulation models; hence, the accuracy of the maps derived from the classification of the satellite images reduces the uncertainty in modeling. The aim of this study was to assess the accuracy of the maps produced by machine learning based on classification methods (Random Forest and Support Vector Machine) and to compare them with a common classification method (Maximum Likelihood). For this purpose, the image of the OLI sensor of Landsat 8 for the study area (Sattarkhan Dam’s basin in the Eastern Azerbaijan) was used after the initial corrections. Five land uses including urban, irrigated and rain-fed agriculture, range and water body were considered. For conducting the supervised classification, ground truth data were used in two sets of educational (70% of the total) and test (30%) data. Accuracy indexes were used and the McNemar test was employed to show the significant statistical difference between the performances of the methods. The results indicates that the overall accuracy of Support Vector Machine, Random Forest, and Maximum Likelihood methods was 96.6, 90.8, and 90.8 %, respectively; also the Kappa coefficient for these methods was 0.93, 0.81 and 0.83, respectively. The existence of a significant statistical difference at the 95% confidence between the performances of the Support Vector Machine algorithm and the other two algorithms was confirmed by the McNemar test.

Digital soil mapping plays an important role in upgrading the knowledge of soil survey in line with the advances in the spatial data of infrastructure development. The main aim of this study was to provide a digital map of the soil family classes using the random forest (RF) models and boosting regression tree (BRT) in a semi-arid region of Ilam province. Environmental covariates were extracted from a digital elevation model with 30 m spatial resolution, using the SAGAGIS7.3 software. In this study area, 46 soil profiles were dug and sampled; after physico-chemical analysis, the soils were classified based on key to soil taxonomy (2014). In the studied area, three orders were recognized: Mollisols, Inceptisols, and Entisols. Based on the results of the environmental covariate data mining with variance inflation factor (VIF), some parameters including DEM, standard height and terrain ruggedness index were the most important variables. The best spatial prediction of soil classes belonged to Fine, carbonatic, thermic, Typic Haploxerolls. Also, the results showed that RF and BRT models had an overall accuracy and of 0.80, 0.64 and Kappa index 0.70, 0.55, respectively. Therefore, the RF method could serve as a reliable and accurate method to provide a reasonable prediction with a low sampling density.

In the future, the risk of land subsidence due to water resources shortage crisis and improper water resources management will become more and more dangerous. It is necessary to assess and identify areas susceptible to subsidence risk and take necessary actions to reduce risks related to land subsidence. In this study, first, the risk of land subsidence was identified and evaluated using a radar interferometry method called LiCSBAS. Then, the spatial relationship between the occurrence of land subsidence hazard and effective factors such as ground elevation, slope, slope aspect, lithology, land use, groundwater decline, distance from rivers, distance from faults, topographic moisture index, and arc curvature was investigated using the random forest (RF) model. In the end, the land subsidence hazard sensitivity map was prepared after calibrating the random forest algorithm. The analysis of LiCSBAS interferometric time series data from 2015 to 2022 showed that the center of the Marvdasht-Kharameh plain and adjacent agricultural areas are continuously subsiding and the mean deformation rate map showed a subsidence rate of 11.6 centimeters per year. The results of determining the spatial relationship between subsidence occurrence and effective factors confirmed the positive impact of distance from rivers, urban and agricultural land uses, depth of bedrock (aquifer thickness), groundwater decline, and alluvial and fine-grained formations on this phenomenon. Also, the results of subsidence modeling using the random forest algorithm showed that factors such as bedrock depth, groundwater decline, land use, and geology have the greatest impact on the potential for subsidence occurrence in the study area. Also, based on the results, about 3 to 4 percent of the areas are in the very high and extremely high-risk classes of land subsidence, especially in the center and suburbs of Mervdasht. Therefore, water resources management and control and developing a systematic program to reduce subsidence risk and aquifer recharge conservation in Merudasht-Kharameh Plain is essential.

change, accurately predicting wheat production is essential for developing precision agriculture. Remote sensing enables the indirect prediction of crop production before harvest. This research investigates the application of the random forest method and support vector regression for simulating wheat production across ten selected farms in Qazvin Plain from 2019 to 2020, employing NDVI, MSAVI, and EVI vegetation indices. Sentinel 2 satellite data was utilized for the vegetation indices. Production data for the ten wheat fields was obtained from the Agricultural Jihad Organization of Qazvin Province. Evaluation of support vector regression and random forest to assess both the observed and simulated wheat production data was conducted using R2, MBE, RMSE, and MAE statistics. To explore the simulation of wheat production using vegetation indices, seven methods were defined: methods 1 to 3 examine each index separately; methods 4 to 6 focus on binary combinations of the indices; and method 7 considers the combined effects of all three indices. The support vector regression model provided good estimates of wheat production in all methods, except methods one and four, in the test phase, with a coefficient of determination of more than 0.98 and a low RMSE. The random forest model showed significant results in all methods except methods two and six during the test phase, achieving a 95% probability (P-value=0.00) with a coefficient of determination greater than 0.8. Overall, this research highlights the importance and potential of machine learning techniques for timely crop production prediction as a strong foundation for regional food security.

In the present study, the Road Sediment Delivery Model (SEDMODL) and Geographic Information System (GIS) were utilized to estimate the average annual sedimentation caused by the forest road network in the oak forests in the west of Iran, Chaharmahal and Bakhtiari Provinces (Nazi forest road with a length of 5171 meters). Sedimentation from the study forest road network was estimated based on three basic factors in the model. Also, 30 erosion benchmarks were installed to measure the erosion and sedimentation rate at different distances from the road and in different parts of the study road and changes were measured during a year. The results showed that the average soil erosion at different distances from the Nazi road based on erosion benchmarks is 5.7 mm per year. In addition, the estimated erosion and sedimentation rate of the entire study road network based on the SEDMODL model is 2875 and 570 tons per year per kilometer, respectively. Model evaluation using erosion benchmarks showed that SEDMODL is a suitable model for estimating soil erosion on forest roads in the west of Iran (R²=0.78 and RMSE=0.73). It should be noted that statistical analysis of erosion hot-spot analysis showed that 39 percent of forest roads in Nazi showed very high erosion. Based on the results of the present study, it is suggested that conservative, protective, and road maintenance measures in areas with high erosion risk should be prioritized by decision-makers.

Soil loss and extensive degradation caused by gully erosion have always caused serious damage. Because direct field measurement and monitoring of gully erosion are costly and time-consuming, it is very difficult to determine the amount of soil loss caused by gully erosion. The present research was conducted to calculate the volume of soil loss due to gully erosion using machine learning models in the Abgendi watershed of Kohgiluyeh and Boyar Ahmad province based on field studies. Machine learning models include random forest, support vector machine, artificial neural network, and adaptive neural fuzzy inference system. The location of 68 gullies in the area was recorded. Hence, initially, digital layers of factors affecting the expansion of gullies, including topography, pedology, lithology, and hydrology, were prepared as independent variables to model soil loss caused by gullies. Then, representative gullies were selected in the studied watershed, and the volume of soil loss due to gully erosion was directly measured in the field as a dependent variable. The measured gullies were randomly divided into two training and validation groups. The results of the models were evaluated using root mean square error (RMSE) and R2, and the models were compared. According to the results, gully erosion in the Abgendi watershed of Kohgiluyeh and Boyar Ahmad province is increasing every year. Also, the amount of erosion and soil loss will increase when the amount of rainfall and the frequency of intense rainfall (≥5mm) are high. Among the machine learning models used in the present research, the random forest (RF) model was selected as the best model to predict soil loss generated by gully erosion.

By utilizing land surface temperature (LST), valuable insights can be gained regarding the impact of land use on energy balance processes. Therefore, this study aimed to investigate the trend of LST changes due to land use changes in the Gorab rural district. Four land use types, including water bodies, bare land, Agricultural area, and forest, were determined from 2013 to 2024 for the maximum likelihood classification (MLC) and support vector machine (SVM) models. The surveys showed that the area of water in the dry period decreased from 0.9 km2 in 2013 to 0.4 km2 in 2024, a decrease of 0.5 km2. In contrast, the area of forest areas increased from 136.1 km2 in the dry period of 2013 to 147.2 km2 in 2024. The Kappa coefficient values for the SVM and MLC models during the wet season of 2021 were 53.94 and 68.7, respectively. Based on this, it was found that the MLC model has higher accuracy. To match spectral indices with LST values, NDVI, NDSI, and NDWI were calculated. Land use changes during the 2013-2024 period affected land surface temperatures, causing fluctuations from 11.5°C to 21.18°C in the wet season and from 13.81°C to 31.45°C in the dry season. The highest LST values were associated with barren land, while water bodies and vegetation cover had the lowest LST values. Among the spectral indices, the highest positive correlation was observed with NDWI, with a value of 0.64 in 2024. The highest negative correlation, -0.66, was observed with NDVI in the same year. Over the 11 years, the area of forest cover increased by 8.15%, while agricultural land decreased by 33.5%. The most significant change occurred in agricultural lands, which declined in area from 35.5 km² to 23.6 km².