An analysis to interpret the heterogeneous resident's willingness to pay for green roofs to improve the understanding of decision heterogeneity

Urban waterlogging has become a frequent and threatening issue in recent years due to rapid urbanization and extreme weather conditions, resulting in economic losses and health hazards. In this context, green roofs (GRs) emerge as a sustainable and innovative solution to mitigate these issues by absorbing rainfall, reducing runoff, and enhancing urban biodiversity. Despite the apparent benefits, the adoption of GRs remains limited, largely due to a lack of quantitative understanding of the factors that influence urban residents' GR adoption willingness.

This study aims to fill this knowledge gap via a survey approach, and distribute and collect survey responses from 999 residents in Shenzhen, a rapidly developing coastal city in China. The survey is designed to capture a range of variables that may influence residents' decision-making regarding GR adoption, including demographic information, housing characteristics, waterlogging experiences, roof utilization preference, knowledge of and preference for GR, and willingness to adopt GR. The GR adoption willingness is collected assuming two policy scenarios, one with government subsidy and the other without. By leveraging a machine learning model for data analysis, the study identifies five key predictors that commonly influence GR adoption willingness with and without subsidy: recognition of the advantages of GRs (GR_advantage), whether a resident lives on the top floor (Top_floor), the degree of concern about GRs (GR_concern), the duration of waterlogging experienced in and around the community (WL_time), and the individual's level of education (Education). Interestingly, the study also reveals that GR adoption willingness is affected differently under scenarios with and without policy incentives. In the absence of subsidies, the property fee (Pro_fee) is a significant factor; conversely, when policy incentives are present, age and house ownership (House_own) emerge as influential factors.

The complexity of these influencing factors is further evaluated using the SHAP (SHapley Additive explanation) model, which provides a nuanced interpretation of how these factors interact and exert nonlinear impacts on residents' willingness to adopt GRs. The insights derived from this analysis are critical for policymakers and urban planners who are looking to promote GRs as part of an integrated urban water management strategy. Specifically, a combination of long-term subsidies and one-time subsidies can be combined to motivate residential adoption. Recognizing the general unfamiliarity with GRs and related policies among residents, relevant outreach and education programs are essential. In addition, targeted subsidy levels could be helpful in stimulating more GR adoptions. An important consideration in this targeting process is the frequency of waterlogging events, which has been shown to significantly influence residents' willingness to pay for GRs.