摘要
针对资源型城市热环境成因机制的复杂性以及高温挑战日益加剧问题,该文选择4个典型资源型城市为研究区,选取自然因素和人文因素作为影响因子,构建了基于麻雀搜索算法(SSA)优化的XGBoost回归模型,并结合SHAP解释机制量化各驱动因子对城市热环境的影响。研究发现,所选因子对热环境的作用效果和贡献程度因城市的阶段性发展特征存在显著差异,这与城市化过程中地表覆被类型的空间差异具有密切关联;SHAP可解释性分析进一步揭示了各变量对热环境的具体影响,展现了模型在解释变量作用机制上的可靠性和透明性;SSA能够有效的对模型进行优化,构建的SSA-XGBoost模型的R^(2)均在0.9以上,表现出良好的稳定性和回归能力。该模型更精确地分析了资源型城市热环境非线性因素影响,为典型资源城市的建设和管理提供参考。
Considering the complexity of the mechanisms behind the thermal environment in resourcebased cities and the increasingly severe heat crisis,this study selects four typical resource-based cities as the study areas.Natural factors and construction factors were chosen as influencing variables to build an SSA-optimized XGBoost regression model,and the SHAP interpretability mechanism was employed to quantify the impact of various driving factors on the urban thermal environment.The study found that the effects and contributions of the selected factors to the thermal environment show significant differences due to the phase-specific development characteristics of each city,which are closely related to the spatial differences in land cover types during urbanization.SHAP interpretability analysis further revealed the specific impacts of each variable on the thermal environment,demonstrating the reliability and transparency of the model in explaining variable mechanisms.SSA effectively optimized the model,and the constructed SSA-XGBoost model achieved R^(2)values above O.9,indicating good stability and regression performance.This model more accurately analyzes the impact of nonlinear factors on the thermal environment of resource-based cities,providing a more scientific theoretical basis for the construction and management of typical resource-based cities.
作者
范强
刘凯泽
张兵
FAN Qiang;LIU Kaize;ZHANG Bing(School of Geomatics,Liaoning Technical University,Fuxin,Liaoning 122300,China)
出处
《测绘科学》
北大核心
2025年第8期80-91,共12页
Science of Surveying and Mapping
基金
国家自然科学基金项目(42204031)。
