基于地理加权回归的NO_2排放预测模型
|
摘要
为了探究影响NO_2排放的交通相关因素,以交通分析小区为基本单位,提取了美国洛杉矶城市的人口特征、道路网络特征、交通状况等数据,采用地理加权回归模型(GWR)分析各交通相关因素对NO_2排放的影响,从而建立交通小区NO_2排放预测模型。结果表明,交通小区路网密度、交通小区机动车吸引量、通勤时间在30～60 min的工作人数与NO_2的排放呈正相关,表明3种影响因素的增加会造成NO_2排放的增加;而交通小区在家工作的人数、慢行交通(步行、自行车等绿色出行方式)吸引量与NO_2的排放呈负相关,表明适当鼓励在家工作的新型办公方式、鼓励居民出行选择慢行交通,能有效减少交通小区NO_2的排放。
The paper is aimed at exploring the effects of the traffic-related factors on the nitrogen dioxide( NO_2) emissions at the level of the traffic analysis zone( TAZ). The influential factors on the nitrogen dioxide( NO_2) emissions have been found including the traffic conditions,the road network special features,the social-demographic features and the employment situation,as has been collected from Los Angeles,the USA. The said influential factors can be well matched with each TAZ in the software ArcGIS. Furthermore,we have also collected NO_2 emission data from the air quality monitoring stations in the city of Los Angeles,the USA. Taking into account the NO_2 emission data characterized as the discrete ones,they should be well matched to each TAZ with the Kriging interpolation method. Thus,seeing the interactions between the traffic analysis regions of each TAZ,this paper has made an analysis of the effects of the different influential factors on the NO_2 emission data by choosing the geographically weighted regression( GWR) status-in-situ. And,later,we have also chosen the conventional generalized linear regression model( GLM) to make the comparison of the related factors.Thus,correspondingly,the results of the above mentioned comparative study verify that the GWR adopted spatial heterogeneity can provide better fitness than that of the GLM. In addition,the visualization results have further helped us to validate that the GWR model can be used to include the varying relationships between the NO_2 emission data and the traffic-related factors at the different TAZs. Therefore,the results gained in the GWR have enabled us to conclude reasonably that the road dust density,the number of vehicles attracted by TAZ and the number of passengers during the commute time between 30-60 min may all have positive effects on the NO_2 emissions,suggesting that the NO_2 emission tends to increase with the increase of the above mentioned 3 factors. On the other hand,the number of people who are working at home and the volume of the slow traffic( e. g.walking,bicycle) attracted by TAZ should have negative effects on the NO_2 emissions. Hence,accordingly,it is reasonable.helpful and meaningful to encourage people to keep working at home and/or choose slow traffic. The results we have arrived at can help the urban planners to incorporate environmental considerations into their traffic and urban planning work.
The paper is aimed at exploring the effects of the traffic-related factors on the nitrogen dioxide( NO_2) emissions at the level of the traffic analysis zone( TAZ). The influential factors on the nitrogen dioxide( NO_2) emissions have been found including the traffic conditions,the road network special features,the social-demographic features and the employment situation,as has been collected from Los Angeles,the USA. The said influential factors can be well matched with each TAZ in the software ArcGIS. Furthermore,we have also collected NO_2 emission data from the air quality monitoring stations in the city of Los Angeles,the USA. Taking into account the NO_2 emission data characterized as the discrete ones,they should be well matched to each TAZ with the Kriging interpolation method. Thus,seeing the interactions between the traffic analysis regions of each TAZ,this paper has made an analysis of the effects of the different influential factors on the NO_2 emission data by choosing the geographically weighted regression( GWR) status-in-situ. And,later,we have also chosen the conventional generalized linear regression model( GLM) to make the comparison of the related factors.Thus,correspondingly,the results of the above mentioned comparative study verify that the GWR adopted spatial heterogeneity can provide better fitness than that of the GLM. In addition,the visualization results have further helped us to validate that the GWR model can be used to include the varying relationships between the NO_2 emission data and the traffic-related factors at the different TAZs. Therefore,the results gained in the GWR have enabled us to conclude reasonably that the road dust density,the number of vehicles attracted by TAZ and the number of passengers during the commute time between 30-60 min may all have positive effects on the NO_2 emissions,suggesting that the NO_2 emission tends to increase with the increase of the above mentioned 3 factors. On the other hand,the number of people who are working at home and the volume of the slow traffic( e. g.walking,bicycle) attracted by TAZ should have negative effects on the NO_2 emissions. Hence,accordingly,it is reasonable.helpful and meaningful to encourage people to keep working at home and/or choose slow traffic. The results we have arrived at can help the urban planners to incorporate environmental considerations into their traffic and urban planning work.
引文
[1] Ministry of Ecological Environment of the People's Republic of China(中华人民共和国生态环境部). Annual report on prevention and control of motor vehicle pollution in China in 2015(2015年中国机动车污染防治年报)[EB/OL].[2016-01-19]. http://www. zhb. gov. cn/gkml/hbb/qt/201601/t20160119_326622. htm.
[2] YU L. Remote vehicle exhaust emission sensing for traffic simulation and optimization models[J]. Transportation Research Part D:Transport&Environment,1998,3(5):337-347.
[3] LI Wei(李伟). A study on the characteristics of on-road motor vehicle emissions in typical cities(典型城市道路机动车排放污染特征研究)[D]. Beijing:Tsinghua University,2002.
[4] KYOUNGHO A,RAKHA H,TRANI A,et al. Estimating vehicle fuel consumption and emissions based on instantaneous speed and acceleration levels[J]. Journal of Transportation Engineering,2002,128(2):182-190.
[5] MA Yintao(马因韬),LIU Qihan(刘启汉),LEI Guoqiang(雷国强),et al. Application of vehicular emission models and comparison of their adaptability[J]. Journal of Peking University:Natural Science Edition(北京大学学报:自然科学版),2008,44(2):308-316.
[6] ODURO S D,METIA S,DUC H,et al. Multivariate adaptive regression splines models for vehicular emission prediction[J]. Visualization in Engineering,2015,3(1):1-12.
[7] ZHANG Xizhen(张西振). The study of the influencing factors of vehicle emissions and microscopic emission models(汽车排放影响因素及微观排放模型研究)[D]. Xi’an:Chang'an University,2007:16-31.
[8] HU Qizhou(胡启洲),WU Peng(吴鹏),DENG Wei(邓卫),et al. Comprehensive measure model for air pollution due to vehicle emissions[J]. Journal of Southeast University:Natural Science Edition(东南大学学报:自然科学版),2016,46(4):884-887.
[9] LIU B,HU J,YAN F,et al. A novel optimal support vector machine ensemble model for NOxemissions prediction of a diesel engine[J]. Measurement,2016,92:183-192.
[10] DIRKS K N,MURRAY D J,HAY J E,et al. A semiempirical model for predicting the effect of changes in traffic flow patterns on carbon monoxide concentrations[J]. Atmosphere Environment,2003,37:2719-2724.
[11] BEYDOUN M,GULDMANN J M. Vehicle characteristics and emissions:logit and regression analyses of I/M data from Massachusetts,Maryland,and Illinois[J].Transportation Research Part D:Transport&Environment,2006,11(1):59-76.
[12] ZENG W,MIWA T,MORIKAWA T. Prediction of vehicle CO2,emission and its application to eco-routing navigation[J]. Transportation Research Part C:Emerging Technologies,2016,68:194-214.
[13] United States Environmental Protection Agency. Air quality summary statistics for the criteria pollutants by monitor[EB/OL].[2010]. https://www. epa. gov/outdoor-air-quality-data/download-daily-data.
[14] United States Census Bureau. Cartographic boundary shapefiles-traffic analysis zones[EB/OL].[2000]. https://www. census. gov/geo/maps-data/data/cbf/cbf_taz. html.
[15] American Association of State Highway and Transportation Officials(AASTO). 2006-2010 5-year American Community Survey(ACS)data[EB/OL].[2010]. https://ctpp.transportation.org/ctpp-data-setinformation/5-year-data/.
[16] Los Angeles County GIS Data Portal. LA county street&address file[EB/OL].[2014-06-16]. http://egis3. lacounty. gov/dataportal/2014/06/16/2011-lacounty-street-centerline-street-address-file/.
[17] California Department of Transportation. Caltrans traffic volumes 2010[EB/OL].[2010]. https://data. ca.gov/dataset/caltrans-traffic-volumes.
[18] California Department of Transportation. Traffic volume:annual average daily traffic(AADI)[EB/OL].[2010]. http://www. dot. ca. gov/trafficops/census/.
[19] United States Environmental Protection Agency. 2011National Emissions Inventory(NEI)data[EB/OL].[2011]. https://www. epa. gov/air-emissions-inventories/2011-national-emissions-inventory-nei-data.
[20] Unite State Department of Labor. Critical air pollutants from the facilities in different cities[EB/OL].[2017].https://www. bls. gov/.
[21] PU Ziyuan(蒲自源),JI Kui(纪魁),LI Zhibin(李志斌),et al. Spatial modeling and analysis on effects of price change on parking occupancy[J]. Journal of Southeast University:Natural Science Edition(东南大学学报:自然科学版),2017,47(1):193-197.
[22] WU Yuming(吴玉鸣),LI Jianxia(李建霞). Analysis of China's provincial industrial total factor productivity based on geographical weighted regression model[J].Economic Geography(经济地理),2006,26(5):748-752.
[23] FAN Yongdong(范永东). Review of cross validation methods in model selection(模型选择中的交叉验证方法综述)[D]. Shanxi:Shanxi University,2013:15-18.
[2] YU L. Remote vehicle exhaust emission sensing for traffic simulation and optimization models[J]. Transportation Research Part D:Transport&Environment,1998,3(5):337-347.
[3] LI Wei(李伟). A study on the characteristics of on-road motor vehicle emissions in typical cities(典型城市道路机动车排放污染特征研究)[D]. Beijing:Tsinghua University,2002.
[4] KYOUNGHO A,RAKHA H,TRANI A,et al. Estimating vehicle fuel consumption and emissions based on instantaneous speed and acceleration levels[J]. Journal of Transportation Engineering,2002,128(2):182-190.
[5] MA Yintao(马因韬),LIU Qihan(刘启汉),LEI Guoqiang(雷国强),et al. Application of vehicular emission models and comparison of their adaptability[J]. Journal of Peking University:Natural Science Edition(北京大学学报:自然科学版),2008,44(2):308-316.
[6] ODURO S D,METIA S,DUC H,et al. Multivariate adaptive regression splines models for vehicular emission prediction[J]. Visualization in Engineering,2015,3(1):1-12.
[7] ZHANG Xizhen(张西振). The study of the influencing factors of vehicle emissions and microscopic emission models(汽车排放影响因素及微观排放模型研究)[D]. Xi’an:Chang'an University,2007:16-31.
[8] HU Qizhou(胡启洲),WU Peng(吴鹏),DENG Wei(邓卫),et al. Comprehensive measure model for air pollution due to vehicle emissions[J]. Journal of Southeast University:Natural Science Edition(东南大学学报:自然科学版),2016,46(4):884-887.
[9] LIU B,HU J,YAN F,et al. A novel optimal support vector machine ensemble model for NOxemissions prediction of a diesel engine[J]. Measurement,2016,92:183-192.
[10] DIRKS K N,MURRAY D J,HAY J E,et al. A semiempirical model for predicting the effect of changes in traffic flow patterns on carbon monoxide concentrations[J]. Atmosphere Environment,2003,37:2719-2724.
[11] BEYDOUN M,GULDMANN J M. Vehicle characteristics and emissions:logit and regression analyses of I/M data from Massachusetts,Maryland,and Illinois[J].Transportation Research Part D:Transport&Environment,2006,11(1):59-76.
[12] ZENG W,MIWA T,MORIKAWA T. Prediction of vehicle CO2,emission and its application to eco-routing navigation[J]. Transportation Research Part C:Emerging Technologies,2016,68:194-214.
[13] United States Environmental Protection Agency. Air quality summary statistics for the criteria pollutants by monitor[EB/OL].[2010]. https://www. epa. gov/outdoor-air-quality-data/download-daily-data.
[14] United States Census Bureau. Cartographic boundary shapefiles-traffic analysis zones[EB/OL].[2000]. https://www. census. gov/geo/maps-data/data/cbf/cbf_taz. html.
[15] American Association of State Highway and Transportation Officials(AASTO). 2006-2010 5-year American Community Survey(ACS)data[EB/OL].[2010]. https://ctpp.transportation.org/ctpp-data-setinformation/5-year-data/.
[16] Los Angeles County GIS Data Portal. LA county street&address file[EB/OL].[2014-06-16]. http://egis3. lacounty. gov/dataportal/2014/06/16/2011-lacounty-street-centerline-street-address-file/.
[17] California Department of Transportation. Caltrans traffic volumes 2010[EB/OL].[2010]. https://data. ca.gov/dataset/caltrans-traffic-volumes.
[18] California Department of Transportation. Traffic volume:annual average daily traffic(AADI)[EB/OL].[2010]. http://www. dot. ca. gov/trafficops/census/.
[19] United States Environmental Protection Agency. 2011National Emissions Inventory(NEI)data[EB/OL].[2011]. https://www. epa. gov/air-emissions-inventories/2011-national-emissions-inventory-nei-data.
[20] Unite State Department of Labor. Critical air pollutants from the facilities in different cities[EB/OL].[2017].https://www. bls. gov/.
[21] PU Ziyuan(蒲自源),JI Kui(纪魁),LI Zhibin(李志斌),et al. Spatial modeling and analysis on effects of price change on parking occupancy[J]. Journal of Southeast University:Natural Science Edition(东南大学学报:自然科学版),2017,47(1):193-197.
[22] WU Yuming(吴玉鸣),LI Jianxia(李建霞). Analysis of China's provincial industrial total factor productivity based on geographical weighted regression model[J].Economic Geography(经济地理),2006,26(5):748-752.
[23] FAN Yongdong(范永东). Review of cross validation methods in model selection(模型选择中的交叉验证方法综述)[D]. Shanxi:Shanxi University,2013:15-18.