A multistage multiobjective optimization model for land use change based on the green economy with spatial constraint parameters determined by artificial neural networks

This study aims to formulate a Multiobjective Multistage optimization model for land use change problems within a green economy framework. Land use allocation is formulated as a sequential decision-making process, where the decision variables denote the area allocated to each land use type at every stage. A dynamic programming structure is constructed by defining stages as the sequential determination of land allocations, states as the remaining land capacity, and a transition function that updates the state based on decisions made at the previous stage, thereby establishing a recursive relationship consistent with Bellman’s principle of optimality to obtain optimal solutions. The model features two objective functions, namely the maximization of economic benefits as measured by Gross Regional Domestic Product (GRDP) contributions and the minimization of net carbon emissions calculated from land use carbon stock coefficients. This formulation explicitly incorporates green economy principles via a trade-off that reconciles economic growth with environmental sustainability. The proposed approach combines Artificial Neural Networks (ANN) to estimate spatial transition probabilities from historical data and environmental predictor variables, which are subsequently incorporated as spatial constraint parameters in the optimization model, and the Non-dominated Sorting Dynamic Programming (NSDP) algorithm to solve the Multiobjective Multistage problem and generate a set of Pareto optimal solutions. This study has produced a set of Pareto optimal solutions for land allocation through a dynamic programming model that evaluates economic and environmental trade-off at each decision stage, thereby providing a data-driven mathematical framework to support sustainable land-use planning within the context of a green economy.