Land use change and soil organic carbon stock (SOCS) depletion over time is one of the predominant worldwide environmental problems related to global warming and the need to secure food production for an increasing world population. In our research, satellite images from 1988 and 2018 were analyzed for a 177.48 km2 region in Kurdistan Province, Iran. Across the study area. 186 disturbed and undisturbed soil samples were collected at two depths (0–20 cm and 20–50 cm). Bulk density (BD), soil organic carbon (SOC), rock fragments (RockF) and SOCS were measured. Random forest was used to model the spatial variability of SOCS. Land use was mapped with supervised classification and maximum likelihood approaches. The Kappa index and overall accuracy of the supervised classification and maximum likelihood land use maps varied between 83% and 88% and 78% and 85%, respectively. The area of forest and high-quality rangeland covered 5286 ha in 1988 and decreased by almost 30% by 2018. Most of the decrease was due to the establishment of cropland and orchards, and due to overgrazing of high-quality rangeland. As expected, the results of the analysis of variance showed that mean values of SOCS for the high-quality rangeland and forest were significantly higher compared to other land use classes. Thus, transformation of land with natural vegetation like forest and high-quality rangeland led to a loss of 15,494 Mg C in the topsoil, 15,475 Mg C in the subsoil and 15,489 Mg C−1 in total. We concluded that the predominant causes of natural vegetation degradation in the study area were mostly due to the increasing need for food, anthropogenic activities such as cultivation and over grazing, lack of government landuse legislation and the results of this study are useful for land use monitoring, decision making, natural vegetation planning and other areas of research and development in Kurdistan province.

No abstract

Soil depth is a major soil characteristic, which is commonly used in distributed hydrological modelling in order to present watershed subsurface attributes. This study aims at developing a statistical model for predicting the spatial pattern of soil depth over the mountainous watershed from environmental variables derived from a digital elevation model (DEM) and remote sensing data. Among the explanatory variables used in the models, seven are derived from a 10 m resolution DEM, namely specific catchment area, wetness index, aspect, slope, plan curvature, elevation and sediment transport index. Three variables landuse, NDVI and pca1 are derived from Landsat8 imagery, and are used for predicting soil depth by the models. Soil attributes, soil moisture, topographic curvature, training samples for each landuse and major vegetation types are considered at 429 profiles within four subwatersheds. Random forests (RF), support vector machine (SVM) and artificial neural network (ANN) are used to predict soil depth using the explanatory variables. The models are run using 336 data points in the calibration dataset with all 31 explanatory variables, and soil depth as the response of the models. Mean decrease permutation accuracy is performed on Variable selection. Testing dataset is done with the model soil depth values at testing locations (93 points) using different efficiency criteria. Prediction error is computed for both the calibration and testing datasets. Results show that the variables landuse, specific surface area, slope, pca1, NDVI and aspect are the most important explanatory variables in predicting soil depth. RF and SVM models are appropriate for the mountainous watershed areas that have been limited in the depth of the soil and ANN model is more suitable for watershed with the fields of agricultural and deep soil depth.

Despite the importance of the prediction of land susceptibility to gully erosion, there is a lack of research studies adopting the deep‐learning approach. This study aimed to predict gully susceptibility hotspots using hybridized deep‐learning models and evaluate their efficiency. Field records of gully occurrences in a gully‐prone region, the Talwar watershed (6468 km2), eastern Kurdistan province, Iran, were used to generate a gully inventory dataset. A total of 14 geomorphometric, environmental, and topo‐hydrological gully drivers were selected as predictor variables. The hybridized models were developed using convolutional neural network (NNC) and metaheuristic procedures, including the gray wolf optimizer (GWO) and the imperialist competitive algorithm (ICA). The validity of the resulting outputs was investigated based on the area under the receiver operating characteristic (ROC) curve. Results revealed that the NNC‐GWO had the highest efficiency in the validation step (AUC = 97.2%), whereas the NNC‐ICA was the second‐best model (AUC = 95.1%). The standalone NNC model showed the lowest accuracy (AUC = 91.2%) in predicting gully susceptibility hotspots compared to NNC‐GWO and NNC‐ICA. Thus, both hybridized models had better predictive performance for identifying gully susceptibility in comparison with the standalone NNC model. Furthermore, according to the NNC‐GWO model, about 0.2% (1294.8 ha) and 0.05% (235.2 ha) of the study area were identified as high and very high gully susceptibility classes. In addition, the application of the standalone NNC led to an overestimation of the susceptibility degree for gully initiation. This study supports researchers efforts to increase the model's performance when working in the land degradation domain.