本期目录
张亮林,潘竟虎,赖建波,魏石梅,王云,张大弘.基于GWR降尺度的京津冀地区PM2.5质量浓度空间分布估算[J].环境科学学报,2019,39(3):832-842
基于GWR降尺度的京津冀地区PM2.5质量浓度空间分布估算
- Estimation of PM2.5 mass concentrations in Beijing-Tianjin-Hebei region based on geographically weighted regression and spatial downscaling method
- 基金项目：国家自然科学基金（No.41661025）；西北师范大学青年教师科研能力提升计划（No.NWNU-LKQN-16-7）
- 作者
- 单位
- 张亮林
- 西北师范大学地理与环境科学学院, 兰州 730070
- 潘竟虎
- 西北师范大学地理与环境科学学院, 兰州 730070
- 赖建波
- 西北师范大学地理与环境科学学院, 兰州 730070
- 魏石梅
- 西北师范大学地理与环境科学学院, 兰州 730070
- 王云
- 西北师范大学地理与环境科学学院, 兰州 730070
- 张大弘
- 西北师范大学地理与环境科学学院, 兰州 730070
- 摘要：卫星遥感估算PM2.5质量浓度研究已较为成熟，但精度还未取得突破性进展.本文利用2017年京津冀地区气溶胶光学厚度（AOD）遥感数据、戈尔德地球观测系统的GEOF气象格网数据以及地面环境监测站PM2.5数据，采用地理加权回归空间降尺度方法，估算京津冀地区的逐月PM2.5质量浓度.基于3种不同的残差插值修正，修正后的PM2.5估算结果均很理想，其中，基于自然邻近残差插值修正后的模型估算结果最优.经验证，在95%的置信水平下，其相关系数r达到0.951，决定系数R2为0.904，调整后的R2为0.903，平均预测误差MPE为7.307 μg·m-3，均方根误差RMSE为11.62 μg·m-3，相对预测误差RPE为18.35%，说明该模型能客观估算京津冀地区2017年PM2.5质量浓度.2017年PM2.5呈现出南高北低的空间分布特征，南北高低值区域界线与保定市和沧州市的市级行政界线具有较高的一致性.经变异系数分析发现PM2.5在2017年内的稳定性程度与PM2.5质量浓度空间分布呈反向性，即PM2.5质量浓度高的区域稳定性低，年内的变化程度剧烈，而PM2.5质量浓度低的区域稳定性强，年内变化程度弱.
- Abstract：Although the study of estimating PM2.5 mass concentrations by satellite remote sensing have been mature, the accuracy of estimating has not made breakthrough progress. In this study, based on the Aerosol Optical Depth (AOD) remote sensing data, the GEOF meteorological grid data of Golden Earth Observation System and PM2.5 data from ground environment monitoring station of Beijing-Tianjin-Hebei region in 2017, the monthly PM2.5 mass concentrations in Beijing-Tianjin-Hebei region were estimated by using geographically weighted regression spatial downscaling method. The PM2.5 estimations results of three different residual interpolation modifications are all ideal, among which the result based on the model of natural neighborhood residual interpolation modification is the best. After verification, at 95% confidence level, the correlation coefficient is 0.951, the determination coefficient R2 is 0.904, the adjusted R2 is 0.903, the average prediction error MPE is 7.307 μg·m-3, the root mean square error RMSE is 11.62 μg·m-3, and the relative prediction error RPE is 18.35%. It shows that the model could objectively estimate PM2.5 mass concentrations in Beijing-Tianjin-Hebei in 2017. The annual average PM2.5 in 2017 showed the spatial distribution characteristics of high in the south and low in the north. The regional boundaries of PM2.5 concentrations between North and South were in good agreement with the municipal administrative boundaries of Baoding and Cangzhou City. In addition, the analysis of coefficient of variation indicated that the stability of PM2.5 in 2017 was inverse to the spatial distribution of PM2.5 mass concentrations. Namely, the region with high PM2.5 mass concentrations had low stability and the degree of changes during the year was intense, while the region with low PM2.5 mass concentrations had strong stability and the degree of changes during the year was weak.
摘要点击次数： 1064 全文下载次数： 851
