基于最大稳定过程的北京市PM2.5监测站空间网络设计

Spatial Network Design for PM2.5 Monitoring Stations in Beijing Urban Area Based on Max-Stable Processes

本文提出了一种新的PM2.5监测站最优化空间网络设计方法,该方法是基于GPD模型和最大稳定过程的。模型拟合分为两个部分,第一个部分是边际分布,即GPD模型估计的部分;第二部分是最大稳定过程模型部分。我们把本文提出的方法应用于北京市主城区的17个PM2.5监测站2013年12月5日至2017年8月5日的数据中,并利用ACLIC选择Schlather模型作为最大稳定过程模型进行分析。最终,在由网格方法得到的231个备选点中,选择了熵值最大的10个备选点作为新监测站的位置。

In our paper, we propose a new method for optimal spatial design of PM2.5 monitoring stations network, which is based on GPD model and Max-Stable Processes. There are two parts in model fitting, one is the marginal part, namely estimating parameters of GPD models, the other is fitting max-stable processes correspondingly. Then we apply our method to the data of PM2.5 of 17 monitoring stations in Beijing urban area from December 5, 2013 to August 5, 2017. After model fitting we select Schlather model as the most suitable max-stable process to do further analysis, utilizing ACLIC. Finally,we choose 10 candidates with top 10 entropy values among 231 candidates generated by grid method as the positions of monitoring stations to be added.