Unraveling nonlinear effects of environment features on green view index using multiple data sources and explainable...

Area of Science:

• Urban ecology
• Environmental science
• Geographic information science

Background:

• Traditional methods struggle with the complex relationships between urban environmental factors and the Green View Index (GVI).
• Studying spatial heterogeneity and nonlinearity in urban greening is crucial for environmental sustainability and human well-being.
• Addressing challenges in interpretability and nonlinearity is essential for effective urban planning.

Purpose of the Study:

• To develop an interpretable spatial machine learning framework for analyzing urban greening.
• To investigate the nonlinear and heterogeneous relationships between environmental factors and GVI in Beijing.
• To provide quantitative insights for scientific urban greening resource allocation.

Main Methods:

• Utilized a novel framework combining Geographically Weighted Random Forest (GWRF) and SHapley Additive exPlanation (Shap) models.
• Integrated multi-source big data, including Baidu Street View and remote sensing imagery.
• Employed semantic segmentation and geographic data processing techniques for GVI analysis.

Main Results:

• GVI in Beijing exhibits significant spatial clustering, positive correlations, and distinct spatial variations.
• Green coverage rate positively correlates with GVI, while building density shows a strong negative correlation.
• The GWRF model significantly outperformed comparison models in predicting GVI, demonstrating excellent performance.
• Environmental and socioeconomic factors influence GVI non-linearly, with notable threshold effects.

Conclusions:

• The proposed interpretable spatial machine learning framework effectively captures nonlinear and heterogeneous relationships influencing GVI.
• Findings provide crucial quantitative insights into the impact of green coverage and urban density on GVI.
• Results support evidence-based decision-making for urban planners in optimizing green resource allocation and enhancing urban environments.