A Geomorphic Process-Based Multi-Objective Optimization Framework for Near-Natural Landform Reconstruction in Mining Areas

Landform reconstruction of abandoned mine lands is a pivotal component of ecological restoration, where the central challenge lies in systematically balancing long-term ecological stability with immediate economic costs. Conventional near-natural restoration methods predominantly focus on mimicking static geomorphic parameters derived from reference landscapes. While providing valuable quantitative targets, this approach often overlooks the underlying geomorphic evolution processes and lacks a systematic framework for navigating the inherent trade-offs between conflicting objectives. This study aims to transcend these limitations by proposing a novel paradigm for landform reconstruction based on geomorphic process simulation and multi-objective optimization. The framework is designed to concurrently minimize two competing objectives: (1) the potential for water erosion, represented by the topographic factor (LS-factor) of the Revised Universal Soil Loss Equation (RUSLE), and (2) the engineering cost, represented by the total earthwork volume. The powerful Non-dominated Sorting Genetic Algorithm II (NSGA-II) is employed to computationally evolve and explore the solution space of potential landform designs. A case study was conducted using a high-resolution Digital Elevation Model (DEM) from a typical abandoned mine site in a loess hilly region. The optimization successfully generated a Pareto optimal front, which explicitly quantifies the trade-off relationship between ecological stability and economic cost. The results demonstrate that the framework can provide decision-makers with a diverse portfolio of optimal design schemes, ranging from ‘cost-priority’ to ‘ecology-priority,’ and quantitatively illustrates the best achievable erosion control levels for different budgetary constraints. Meanwhile, plan subsequent revegetation and management strategies accordingly. Through detailed morphological and statistical analysis of representative schemes, it was verified that the process-driven optimization spontaneously converges towards forming geomorphic features that are remarkably similar to the natural reference area. This research not only provides a scientific and efficient decision-support tool for landform reconstruction but also promotes a paradigm shift in mine ecological restoration from ‘form mimicking’ to ‘process-oriented, intelligent optimization,’ offering a new application of geographical information science to solve complex environmental management problems.