基于GWR模型的天山北坡经济带PM2.5浓度反演及时空特征分析  

Inversion of PM Concentration in the Economic Zone of Northern Tianshan Mountain Slope Based on the GWR Model and the Temporal and Spatial Characteristics

作　　者：易唯 杨东[1] 李茜荣 YI Wei;YANG Dong;LI Xirong(College of Geography and Environmental Science,Northwest Normal University,Lanzhou 730070,China)

机构地区：[1]西北师范大学地理与环境科学学院,兰州730070

出　　处：《地球与环境》2021年第1期51-58,共8页Earth and Environment

基　　金：国家自然科学基金项目(41262001)。

摘　　要：构建PM2.5浓度与相关因子的关系模型已成为获取干旱区经济带连续变化PM2.5浓度数据的有效手段之一。本文以天山北坡经济带为研究对象,基于PM2.5浓度监测数据、中分辨率成像光谱仪(moderate-resolution imaging spectroradiometer,MODIS)的气溶胶光学厚度数据(aerosol optical depth,AOD)和气象数据,利用地理加权回归模型(geographically weighted regression,GWR)反演了研究区2018年3~11月PM2.5浓度,进而分析其时空变化特征。结果表明:(1)相比多元回归模型(multiple linear regression,MLR),GWR模型在天山北坡经济带的PM2.5浓度反演效果更优,决定系数R2、平均绝对误差MAE和均方根误差RMSE分别为0.897、4.569μg/m3和5.627μg/m3,明显优于MLR模型的0.819、5.825μg/m3和7.731μg/m3;(2)天山北坡经济带PM2.5浓度在月尺度上呈"凹字型"变化特征,其中11月最高,达到59.50μg/m3,3月和10月次之,9月最低,仅为17.92μg/m3;在季节尺度上表现出秋季(9~11月)>春季(3~5月)>夏季(6~8月)的变化特征,其中春季呈波动下降趋势,夏季总体维持在较低水平,秋季呈急剧上升趋势;(3)在空间分布上,研究区PM2.5浓度呈现出"东高西低"的特征,峰值出现在乌鲁木齐附近,说明经济带东部地区PM2.5污染相对严重,尤其是乌鲁木齐。Constructing a model for the relationship between PM2.5 concentration and related factors has been regarded as one effective means to obtain the continuously PM2.5 concentration data in the economic zone of arid regions.In this study,the Economic Zone in Northern Tianshan Mountain Slope was taken as the research object.Based on the data of monitored PM2.5 concentration,aerosol optical depth( AOD) acquired by the moderate-resolution imaging spectroradiometer( MODIS) and the meteorological data,the geographically weighted regression model( GWR) was used to invert the PM2.5 concentration in the study area from March to November in 2018,furthermore,the spatiotemporal variation characteristics was analyzed.The results showed that:( 1) compared with the multiple linear regression model( MLR),the inversion effect of PM2.5 concentration by using the GWR model in the Economic Zone of Northern Tianshan Mountain Slope illustrated a better result,and the coefficient of determination R2,mean absolute error( MAE) and the root mean square error( RMSE) were 0.897,4.569 μg/m3 and 5.627 μg/m3,respectively,which were remarkably better than 0.819,5.825μg/m3 and 7.731 μg/m3 of MLR model;( 2) the PM2.5 concentration in the Economic Zone of Northern Tianshan Mountain Slope indicated a concave type for the monthly time-scale,with the highest in November( 59.50 μg/m3),then followed by March and October,and the lowest in September( 17.92 μg/m3).For different seasons,it decreases in the order of autumn( September-November) >spring( March-May) >summer( June-August),and a characteristic of fluctuation with downward trend in spring,stable at a low level in summer,sharply increase in autumn,was observed;( 3) in terms of spatial distribution,the PM2.5 concentration in the study area displayed the characteristics of"higher in the east and lower in the west",and the peak value appeared near Urumqi,indicating that the PM2.5 pollution in the eastern part of the economic belt was relatively serious,especially in Urumqi.

关 键 词：地理加权回归模型 天山北坡经济带 PM2.5浓度反演 时空特征 

分 类 号：X51[环境科学与工程—环境工程]