A Decomposition-Based Stochastic Multilevel Binary Optimization Model for Agricultural Land Allocation Under Uncertainty

Crop cultivation planning is vital for optimizing agricultural productivity and sustainable land use under farming uncertainties. This study developed a decomposition-based stochastic multilevel binary optimization model for agricultural plot management. Using land and crops as the division standard, the complex problem of agricultural land management was broken down into manageable sub-modules, which were efficiently solved using a greedy algorithm. In order to verify the actual effectiveness of the model, this study conducted an empirical analysis based on the production practice scenario in the mountainous areas of North China from 2023 to 2026. The performance of the model was verified through dimensions such as agricultural income accounting, the assessment of planting dispersion, and the optimization of legume crop rotation patterns. The stability of the system was also tested using sensitivity tests for multiple variables. To further evaluate the performance of the model, we compared it with two single-factor benchmark models that only considered uncertainty or only considered the land constraints. The results showed that in the multi-year and multi-income scenarios, our comprehensive model was significantly better than the two benchmark models in terms of optimization performance, which proves the necessity of considering land constraints and uncertainty at the same time.