北京市城市功能类型对地表温度的季节性影响测度
Measuring the seasonal impact of Beijing urban functional types on land surface temperature
王楚 1王亮 2赵习枝 2欧尔格力3
作者信息
- 1. 兰州交通大学测绘与地理信息学院,兰州 730070;地理国情监测技术应用国家地方联合工程研究中心,兰州 730070;甘肃省地理国情监测工程实验室,兰州 730070;中国测绘科学研究院,北京 100036
- 2. 中国测绘科学研究院,北京 100036
- 3. 青海省地理空间和自然资源大数据中心,西宁 810001;青海省地理空间信息技术与应用重点实验室,西宁 810001
- 折叠
摘要
针对不同城市功能设施对于地表温度的影响方式和影响程度尚不明确的问题,该文以北京市为研究区,使用MODIS地表温度产品和兴趣点(POI)数据,首先合成北京市2022年4个季节的昼夜地表温度,再将POI数据进行清洗重分类表征城市功能设施类型信息,利用极限梯度提升(XGBoost)算法建立回归模型,其后通过可解释机器学习模型(SHAP)分析各城市功能设施类型在不同时间尺度与地表温度的非线性关系,探究了随着北京市5类设施实体聚集程度的变化,其对热环境影响能力的变化规律.研究结果表明,POI数据可以反映人类活动强度和类型,对于地表温度具有一定的解释能力,且不同城市功能设施类型和地表温度的关系存在季节性的变化,能够实现更加精细尺度的城市地表温度分析,服务城市空间布局规划.
Abstract
Given the unclear understanding of the influence mode and degree of different urban functional facilities on land surface temperature,the MODIS land surface temperature products and point of interest(POI)data were utilized to synthesize the diurnal land surface temperature in Beijing for the four seasons of 2022 in this paper.Subsequently,the POI data was cleaned and reclassified to characterize the type of urban functional facilities.By Utilizing the extreme gradient boosting(XGBoost)algorithm for regression modeling and employing Shapley additive explanations(SHAP)for interpreting the results,to examine the nonlinear relationship between different types of urban functional facilities and land surface temperature at various time scales.Furthermore,the variations in the impact capability of five types of facilities on thermal environment with changes in the degree of aggregation of facility entities in Beijing were explored.The results indicated that POI data could reflect the intensity and types of human activities and contribute to explaining land surface temperature.Additionally,the relationship between different types of urban functional facilities and land surface temperature exhibits seasonal changes,enabling a finer-scale analysis of urban land surface temperature and supporting urban spatial layout planning.
关键词
地表温度/POI/XGBoost/可解释机器学习模型Key words
land surface temperature/POI/XGBoost/SHAP model引用本文复制引用
基金项目
科技部重点研发项目(2022YFC3003604)
中国测绘科学研究院基本科研业务费项目(AR2204)
兰州交通大学优秀平台支持项目(201806)
青海省基础研究计划项目(2024-ZJ-927)
出版年
2024
NETL
NSTL
万方数据