哈尔滨街区PM_(2.5)分布与街区形态耦合机制及预测研究  

Study on the Coupling Mechanism and Prediction of PM_(2.5) Distribution and Block Morphology in Harbin Neighborhoods

作　　者：张军 戴春雨 崔鹏 殷青[2] 李亭亭 ZHANG Jun;DAI Chunyu;CUI Peng;YIN Qing;LI Tingting(School of landscape architecture,Northeast Forestry University,Harbin 150040,China;School of architecture,Harbin Institute of technology,Harbin 150001,China)

机构地区：[1]东北林业大学园林学院,哈尔滨150040 [2]哈尔滨工业大学建筑学院,哈尔滨150001

出　　处：《建筑科学》2022年第8期78-86,共9页Building Science

基　　金：国家自然科学基金项目“建筑能耗动态信息模拟和热工实验结合的寒地村镇低能耗住宅设计研究”(52078155)。

摘　　要：城市污染物扩散受城市空间形态影响。选取典型寒地城市哈尔滨城区内10.1 km^(2)区域进行长期实测,实测内容包括不同城市空间形态下的PM_(2.5)浓度与城市微气候数据,发现相同时间内不同测点之间PM_(2.5)浓度的最大差异可达69.03μg/m^(3)。结合地理信息系统(GIS)提取并筛选合理缓冲半径下的城市空间形态数据,开展PM_(2.5)浓度预测实验,建立梯度提升回归树(GBRT)预测模型,分别与随机森林(RF)、决策树(DT)和多元线性回归(MLR)模型进行对比,结果表明:GBRT模型精度最高,R~2达到0.981;对城市空间形态要素进行影响程度排序,影响程度最高的是建筑密度(57%)和平均建筑高度(49%)。研究成果在揭示PM_(2.5)分布与街区形态耦合机制的同时也肯定了GBRT模型用于街区尺度PM_(2.5)浓度预测的合理性。The diffusion of urban pollutants is affected by urban spatial morphology.In this study,a 10.1 km^(2) area of Harbin,a typical city in the cold region,was selected for long-term measurement.The measurement includes PM_(2.5) concentration data and urban microclimate data under different urban spatial morphology.It was found that the maximum difference in PM_(2.5) concentration was 69.03μg/m^(3).Using the geographic information system(GIS),the urban spatial morphology data with reasonable buffer radius was extracted and screened,PM_(2.5) concentration prediction experiment was conducted,and the gradient lifting regression tree(GBRT)PM_(2.5) concentration prediction model was also established.The model was then compared with random forest(RF),decision tree(DT),and multiple linear regression(MLR)models.The results show that the accuracy of the GBRT model is the highest,with R~2 reaching 0.981.The urban spatial morphology elements are ranked by the degree of influence,the highest is the building density(57%),followed by the average building height(49%).The research results not only reveal the coupling mechanism of PM_(2.5) distribution and block morphology but also confirm the rationality of the GBRT model for PM_(2.5) concentration prediction at the block scale.

关 键 词：城市空间形态 PM_(2.5)扩散 梯度提升回归树(GBRT)模型 严寒地区 

分 类 号：TU831[建筑科学—供热、供燃气、通风及空调工程]