顾及地理环境各向异性的PM_(2.5)浓度时空预测模型  

A Spatiotemporal Prediction Model of PM_(2.5)Concentration Incorporating Geographic Environmental Anisotropy

作　　者：王军杰 王达[1] 丁晨[1] 唐建波[1,2] 石岩 杨学习[1,2] WANG Junjie;WANG Da;DING Chen;TANG Jianbo;SHI Yan;YANG Xuexi(Department of Geo-Informatics,Central South University,Changsha 410083,China;Hunan Geospatial Information Engineering and Technology Research Center,Changsha 410018,China)

机构地区：[1]中南大学地理信息系,长沙410083 [2]湖南省地理空间信息工程技术研究中心,长沙410018

出　　处：《地球信息科学学报》2024年第9期2106-2122,共17页Journal of Geo-information Science

基　　金：国家自然科学基金面上项目(42271485);湖南省自然科学基金项目(2022JJ40585、2024JJ8343);中南大学中央高校基本科研业务费专项资金资助(2024ZZTS0638)。

摘　　要：PM_(2.5)浓度预测对于空气污染的防控和治理具有重要意义。传统图卷积神经网络(Graph Convolutional Network,GCN)等时空预测模型主要通过监测站点间的欧氏距离来度量PM_(2.5)分布的空间相关性,未顾及地形和风向等因素对大气污染物传输过程的各向异性影响,导致地形复杂区域内的预测结果精度偏低。为此,本文提出了一种顾及地理环境各向异性的PM_(2.5)浓度时空图卷积网络预测模型。首先,考虑地理环境的各向异性特征,利用地形和风向等对站点间PM_(2.5)传播的各向异性影响构建GCN的边,将站点PM_(2.5)浓度、土地利用和其他气象因子建模为GCN的节点特征。其次,通过GCN提取站点PM_(2.5)浓度的空间特征。最后,通过门控循环单元(Gate Recurrent Unit,GRU)建模站点PM_(2.5)浓度的时间特征并进行预测。本文以山地型省份贵州省2017年逐小时PM_(2.5)浓度记录进行实验,并采用一系列时空预测基线模型(GTWR、STSVR和GCN+GRU)与本文模型进行对比。实验结果表明:①本文模型的PM_(2.5)浓度预测结果的RMSE、MAE别为10.047、6.848,相比基线模型平均下降了11.29%和12.16%;R2为0.883,比基线平均提升了3.72%;②通过不同地形对站点PM_(2.5)浓度相关性影响分析,论证了山脉沟谷等地形会显著影响站点间的PM_(2.5)浓度的相关性,进而影响站点PM_(2.5)浓度预测结果;③充分考虑地形和风向对PM_(2.5)传播产生的各向异性影响,能够显著提升存在山脉和沟谷地形区域内PM_(2.5)预测精度。PM_(2.5)concentration prediction plays a pivotal role in the prevention and control of air pollution.Traditional forecasting models,such as the Graph Convolutional Network(GCN)and other spatial-temporal prediction models,measure the spatial correlation of PM_(2.5)distribution primarily by monitoring the Euclidean distance between monitoring stations.However,these models often fail to account for the anisotropy effects of terrain,wind direction,and other factors that significantly influence the transport process of air pollutants.This oversight can result in lower accuracy of prediction results,especially in areas with complex terrain.This paper proposes a novel spatiotemporal convolutional network prediction model for PM_(2.5)concentration that takes into account the anisotropy of the geographical environment.The model first constructs the edges of the GCN by incorporating the anisotropic effects of terrain and wind direction on PM_(2.5)propagation between stations.It then models the station PM_(2.5)concentration,land use,and other meteorological factors as node characteristics of the GCN.The model employs GCN to extract the spatial characteristics of PM_(2.5)concentration and subsequently uses a Gate Recurrent Unit(GRU)to model and predict the temporal characteristics of PM_(2.5)concentration at the station.The model's performance was evaluated using hourly PM_(2.5)concentration records from the mountainous Guizhou province in 2017.It was compared against several spatiotemporal prediction baseline models,including Geographically and Temporally Weighted Regression(GTWR),Spatio-Temporal Support Vector Regression(STSVR),and a combined GCN+GRU model.The experimental results demonstrate that the model proposed in this paper significantly outperforms the baseline models.The Root Mean Square Error(RMSE)and Mean Absolute Error(MAE)ratios of PM_(2.5)concentration predicted by the model are 10.047 and 6.848,respectively,which represent decreases of 11.29%and 12.16%compared to the baseline models.The R-squared value of 0.8

关 键 词：PM_(2.5)浓度预测 时空图卷积网络 空间相关性 地理协变量 地形距离 风向 各向异性 地理环境特征 

分 类 号：X513[环境科学与工程—环境工程]