Optimization of green infrastructure networks based on potential green roof integration in a high-density urban area—A case study of Beijing, China

Green infrastructure network (GIN) optimization is an effective measure to reduce the landscape fragmentation caused by rapid urbanization. However, there are few targeted and practical studies of GINs in high-density urban areas with a prominent contradiction between ecological construction and land scarcity, leading to insufficient feasibility of most optimization paths as they avoid practical contradictions (scarcity of land, high cost, etc.). As an effective way to economically increase green infrastructure, green roofs have been demonstrated to provide habitats and stepping stones to increase landscape connectivity for high-mobility organisms. However, few studies have applied green roofs to GIN optimization. To address this question, a new approach to optimize GINs was proposed from the perspective of integrating potential green roofs (PGRs). A complete and feasible workflow was also established to rapidly, accurately, and cost-effectively extract PGRs, scientifically evaluate the comprehensive landscape connectivity accounting for PGR isolation factors, and practically optimize GINs according to the spatial differentiation of PGRs with high landscape connectivity. This was done by integrating high-spatial-resolution remote sensing, machine learning, morphological spatial pattern analysis, landscape index method, and a minimum cumulative resistance model. A case study in a typical high-density urban area within the Beijing Fifth Ring Road, China demonstrated the applicability and implications of the workflow. The results clearly showed that the study area had a high potential for green roof retrofitting, PGRs with high landscape connectivity could effectively improve the GINs, and the spatial differentiation characteristics of the PGR network optimization benefits provided the scientific guidance for developing targeted ecological strategies. The new approach effectively improves the scientificity and implementability of GINs. It also provides a strong reference for landscape planning and ecological construction in other high-density urban areas facing the contradiction between ecological construction and land scarcity.