Sensitivity Analysis of the Soil Criterion in Forest Road Planning

Introduction: Decision-making models based on Multicriteria Evaluation (MCE) often face uncertainties. Sensitivity analysis helps quantify these effects by assessing how changes in input maps influence the results. Soil maps, particularly soil erodibility (Factor K), are crucial for environmental analysis but are commonly classified into discrete classes, which poses a challenge for MCE models. This study aimed to create a continuous Factor K map using geostatistical techniques (K2) combined with fuzzy logic, compared to discrete classes (K1), and to evaluate its impact on MCE results. Sensitivity analysis compared the two methods of Factor K spatialization to identify critical areas for forest road allocation, considering slope, distance from accumulated water flow, and aspect. K2 demonstrated higher accuracy (94.5%) than K1 (93.1%), showing better representativeness. It is concluded that geostatistical methods improve Factor K maps, aligning with MCE requirements and enhancing the representation of soil characteristics. Objective: This study aims to investigate the impact of geostatistical and fuzzy logic methods on soil erodibility mapping, with the aim of improving the accuracy of forest road planning through multicriteria evaluation. Theoretical Framework: The study is grounded in Multicriteria Evaluation (MCE), Sensitivity Analysis (SA), Fuzzy Logic, and Geostatistics (particularly kriging). These frameworks provide a solid basis for understanding the role of soil erodibility (Factor K) in environmental planning and decision-making for forest road allocation. Method: The methodology involved comparing two MCE models for forest road planning using different soil erodibility mapping approaches (Boolean vs. kriging). Soil samples were collected from 105 points, analyzed for texture, and used to generate Factor K maps. The models incorporated slope, aspect, and distance from accumulated water flow and were validated using statistical analyses. Results and Discussion: The geostatistical model (K2) demonstrated higher accuracy than the class-based model (K1). Continuous mapping improved the representation of soil erodibility, enhancing spatial decision-making. The discussion highlights the benefits of geostatistics while acknowledging sampling density as a limitation. Research Implications: The study reinforces the importance of accurate soil mapping for improved forest road planning, with implications for environmental management, erosion control, and spatial decision support. Originality/Value: This study presents an innovative method combining geostatistics and fuzzy logic to enhance soil erodibility mapping within MCE models. The approach improved spatial representation and increased model accuracy from 93.1% to 94.5%, demonstrating practical value for forest road planning and environmental modeling.