Objective Green infrastructure (GI) is an important carrier for providing diverse ecosystem services and safeguarding the security and stability of regional ecosystems. Nevertheless, the trade-off relationship between various ecosystem services that GI can provide makes it difficult for GI planning decision to simultaneously maximize multiple service provisions. This research aims to construct a multi-objective spatial optimization model for GI planning with the goal of maximizing multiple ecosystem service provisions, and to provide succinct recommendations and technical guidelines for ecological spatial and GI planning within the framework of territorial spatial planning.

Methods Based on the NSGA-Ⅱ optimization algorithm, the GI planning optimization model for maximizing 3 ecosystem services (i.e., habitat quality, crop production, and runoff reduction) is constructed and applied to the central urban area of Wuhu, Anhui Province. The optimization model contains four parts: decision variables, constraints, objective functions, and the optimization algorithm. Decision variables determine the spatial location and type of GI, thus generating a variety of GI layout schemes, while constraints ensure that these schemes comply with certain requirements. Subsequently, the objective functions calculate the values for three critical ecosystem services, thereby representing the capacity of GI schemes to supply multiple ecosystem services. The InVEST − Crop Production Regression Model, InVEST − Habitat Quality Model, and InVEST − Urban Flood Risk Mitigation Model are used as objective functions for quantifying ecosystem services of crop production, habitat quality, and runoff reduction. The NSGA-Ⅱ algorithm can, through multiple iterations, generate GI schemes and obtain the objective function values of various ecosystem services, with the aim of maximizing multiple services and determining the optimal GI layout solution set. The research also analyzes the trade-offs and synergistic relationships among the three key ecosystem services through scatter distribution trends and curve fitting based on the optimization results. By comparing the optimal GI layout for each objective preference scheme, the optimal GI layout for each service trade-off, and the spatial layout differences of the algorithm for each GI type under the current scheme, the GI layout strategy for synergizing multiple ecosystem services in the research area is clarified.

Results The optimization model yields a corpus of 50 Pareto-optimal planning solutions for GI. The optimized GI planning solutions demonstrate substantially improved capacity to provide ecosystem services, contingent upon differential service preferences. The compromise scheme outperforms the current scheme in providing runoff reduction service and habitat quality service, the crop production yield and habitat quality index of which reach 71,658.67 mm/a and 0.2993 respectively, while failing to achieve an upgrade in providing crop production service over the current scheme. Synthesizing the optimization findings, the trade-offs and synergies between various ecosystem services provided by GI are delineated. Further, the research proposes the spatial layout characteristics of GI under each objective preference and for synergizing multiple services, which are as follows: Under the habitat quality preference, add forest and grass space in urban areas, and decentralize the placement of habitat patches; under the crop production preference, maintain the current agricultural pattern, and increase the proportion of riverside farmland; under the rainwater runoff reduction preference, increase the storage capacity of urban areas, and construct a flood regulating network; and in the case of compromising the three services, increase the proportion of multifunctional GI, and pay attention to the composite enhancement of the services.

Conclusion Utilizing the NSGA-Ⅱ algorithm and the InVEST model, this research innovatively develops a synergistic optimization model for GI spatial layouts. The results illuminate the trade-offs and synergies between services, which may guide the spatial planning of GI under diverse service preferences. The research demonstrates that multi-objective optimization can significantly aid in GI planning toward enhancing ecosystem service supply and informing spatial strategies. It also highlights specific GI layout characteristics for various service preference schemes, bridging the gap between ecosystem service theory and practical GI planning, and laying a foundation for future research on efficient service provision and service synergy understanding. However, this research has some limitations, particularly in its simplified modeling process due to the generality of the InVEST model and the uncertainty in data precision, potentially leading to skewed results. Future work requires refined models and high-precision data for validation. Additionally, the GI classification based on land use, layout rules, and constraints entail the enhancement of practical implementation and a deeper understanding of the ecosystem service provision mechanisms.