An Integrated Geospatial Framework for Assessing Agricultural Suitability Using Multi-Source Environmental Criteria

Abstract:

Achieving sustainable farming in dry and semi-dry areas depends on accurately identifying optimal land zones by evaluating environmental factors. This research presents a combined approach merging object-oriented image processing with pixel-level modeling to evaluate farming land suitability in Iran's Anjir Plain. By employing Landsat-9 satellite data, 525 spatial units were created via multi-scale segmentation, ensuring boundaries matched natural terrain characteristics. Eleven environmental criteria related to water, climate, and terrain were normalized and combined using the Ordered Weighted Averaging (OWA) method to generate a suitability score (0–1) for each unit.

Findings revealed that altitude (r = 0.8), soil composition and bedrock (r = 0.7), and nearness to groundwater sources (r = 0.7) had the strongest positive correlation with farming suitability, whereas local climate conditions and temperature variables exhibited adverse effects. Merely 7 spatial units, spanning about 8,670 hectares, surpassed the high-suitability benchmark of 0.9, marking them as prime areas for agricultural expansion. Validation using 40 randomly sampled field data points yielded 90% accuracy and an 80% Kappa score. The combined method improved both classification precision and spatial consistency, ensuring better conformity with actual land divisions.

These outcomes highlight the efficacy of blending object-based analysis and OWA for pinpointing feasible agricultural lands, especially in water-limited settings. The framework serves as a practical planning aid for optimizing land use and managing farming resources in arid regions.