WUDAPT方法的局地气候区分类流程优化及地表热环境特征分析——以西安主城区为例  

Optimization of local climate zone classification process and analysis of surface thermal environment characteristics by WUDAPT method:Taking the main city of Xi’an as an example

作　　者：刘衍 雷宸骁 杨柳[1] 李奇 赵龙[3] LIU Yan;LEI Chenxiao;YANG Liu;LI Qi;ZHAO Long

机构地区：[1]西安建筑科技大学建筑学院 [2]华南理工大学建筑学院 [3]西安建筑科技大学设计研究总院

出　　处：《西部人居环境学刊》2024年第5期117-125,共9页Journal of Human Settlements in West China

基　　金：国家自然科学基金面上项目(52078407)。

摘　　要：局地气候区(Local Climate Zone,LCZ)是依据城市地表特征进行分区的城市气候定量分析方法。现有WUDAPT分类流程(World Urban Database and Access Portal ToolsM,WUDAPT)由于人为样本类别识别偏差造成建成类型分类精度明显低于自然类型,本文以西安主城区为例对这一分类流程进行优化,优化流程采用GIS计算和归一化差异指数计算达到预先识别建筑形态特征与自然覆盖属性的目的,结果用于提取不同LCZ样本类型。采用大气校正法反演得到研究区内地表温度数据,与LCZ分类地图叠置分析得到不同LCZ地表热环境差异与规律。从LCZ分类结果来看,优化的分类流程制作的LCZ地图总体分类精度、建成类型分类精度和自然覆盖类型分类精度分别为79.76%、79.97%和79.41%,相比常规流程下的精度分别提升17.4%、19.4%和9.5%,Kappa系数为0.78,一致性达到显著水平。从地表温度分析结果来看,建成类型LCZ地表温度受建筑密度、高度、功能等共同影响,建筑高度相近时密度越大地表温度越高,而建筑密度一定高度越高地表温度越低,LCZ10受人为热影响而地表温度最高。自然覆盖类型地表温度整体低于建成类型,但仍表现出显著的类间差异,如无植被覆盖的硬化地面与裸地均高于植被覆盖类型,低矮植被与稀疏林地均高于稠密林地,水体地表温度最低。从局地气候区分类视角合理规划不同LCZ用地类型可达到改善城市地表热环境的目的。Local Climate Zone(LCZ)is a quantitative analysis method for u rban climate based on urban surface characteristics.The principle of LCZ is to delineate the response ability of urban surface to thermal environment based on urban surface characteristics,and establish the connection between urban surface properties and urban climate issues.It is widely used in the research of urban climate related issues.At present,in the local climate region classification method,The WUDAPT classification process(World Urban Database and Access Portal ToolsM,WUDAPT)relies on its easily accessible remote sensing data(30 m resolution Landsat8 multispectral image)and Google earth platform.LCZ classification mapping is realized with the help of random forest classifier in the open source tool SAGA-GIS,which provides a clearer workflow for remote sensing image classification and reduces the requirements of users’remote sensing professional background,making this method a mainstream method for the study of urban local climate area.In the standard WUDAPT classification process,the classification of the training sample area and the verification sample area refer to Google Earth images and judge the building form feature parameters(mainly building height and density)and natural land cover information through artificial visual interpretation.However,due to the lack of data support,the method of delineating LCZ training areas by visual interpretation is more difficult to apply to urban built-up areas with highly heterogeneous spatial forms,and the classification accuracy of built-up types is often significantly lower than that of natural types due to the bias of human sample classification.In this paper,the main urban area of Xi’an was taken as an example to optimize the classification process.GIS calculation and normalized difference index calculation were used to identify the architectural form characteristics and natural cover attributes in advance,and the identification results were used to extract different LCZ sample types.The data

关 键 词：局地气候区分类 GIS计算 归一化差异指数 分类精度 地表温度 

分 类 号：TU984[建筑科学—城市规划与设计]