Modeling of land use/land cover change and effects on soil properties: A case study of Adıyaman, Besni

In this study, a tripartite approach has been aimed in understanding the relationships between land capability and land use. These are: I.) Revealing the results of land use, II.) Evaluating the ability of the land to fulfill its functions, and III.) Estimating the suitability of the land for potential land use types. As a land degradation process, gully erosion (GE) is one of the indicators of long-term land degradation that reflects the effects of environmental variables on the landscape. In the first stage of the thesis, many factors have been included in the model in Adıyaman Besni district and the areas affected by gully erosion have been determined by using Geographic Object Based Image Analysis (GEOBIA) and Random Forest (RF) supervised machine learning. Accuracy assessment; Precision has been determined by Recall, F1-Score and Kappa index. These values have been found to be 0.90, 0.95, 0.92, and 0.88, respectively. A hybrid (MLP-OK) approach, integrating the prediction of multi-layer perceptron artificial neural network (MLP) and ordinary kriging (OK) residues, has been adopted to produce a soil organic carbon (SOC) spatial prediction map in order to evaluate the ability of the land to fulfill its functions. The prediction accuracy of SOC content has been compared with MLP, MLP-OK, OK, Inverse Distance Weighting (IDW), and Empirical Bayesian Kriging (EBK). Compared to other methods, MLP-OK produced lower prediction errors (ME: -0.035, MAE: 0.138, and RMSE: 0.175) for validation points, and the SOC content was more accurately estimated. In the estimation of SOC content, relative improvements of approximately 58%, 63%, and 70% have been observed compared to other methods. In the final stage of the study, the suitability of agricultural sustainable land use types for wheat and perennial horticultural crops has been assessed. A total of 132 soil samples have been taken from Besni district where the study has been conducted and clay, sand, silt, soil organic matter, salinity, pH, lime, cation exchange capacity, field capacity and wilting point have been determined. Depending on the long-term climatic characteristics, the length of growing season period has been evaluated, and the annual soil loss has been calculated with the RUSLE model. Fuzzy continuous classification has been made using the Mamdani Fuzzy Inference System. Sustainable land use types for wheat; For the fuzzy soil layer, land units have been found to be moderately suitable (S2, 0.87%), marginally suitable (S3, 72.2%), and currently not-suitable (N1, 26.9%). When evaluated together with the fuzzy erosion layer and the fuzzy climate layer, it has been determined that the sustainability categories S1, S2, S3, N1, and N2 covered 1.36%, 3.8%, 47.6%, 18.6%, and 28.6% of the area. According to the fuzzy soil layer results for perennial horticultural crops, 38.1% of the area is S1, 57.8% S2 and 4.02% S3. However, when all three fuzzy layers are assessed together, the distributions of S1, S2, S3 and N1 classes are 2.72%, 5.94%, 48.86% and 42.48%, respectively. The most important results are: (i) the land assessment, which only takes soil characteristics into account, does not give sufficient importance to sustainability.; (ii) soil erosion as a result of improper land use types threatens sustainable agricultural production; (iii) SOC content is successfully predicted by the hybrid spatial model; (iv) marginal agricultural lands occupy quite a large area in the study area. As a result, the study's established a land degradation sensitive sustainable land suitability assessment framework for the extensive land use types in Besni. The results highlight the agricultural land degradation process while offering potential solutions to reverse soil degradation.