Premium
Spatial Representativeness of PM 2.5 Concentrations Obtained Using Observations From Network Stations
Author(s) -
Shi Xiaoqin,
Zhao Chuanfeng,
Jiang Jonathan H.,
Wang Chunying,
Yang Xin,
Yung Yuk L.
Publication year - 2018
Publication title -
journal of geophysical research: atmospheres
Language(s) - English
Resource type - Journals
eISSN - 2169-8996
pISSN - 2169-897X
DOI - 10.1002/2017jd027913
Subject(s) - representativeness heuristic , haze , environmental science , spatial variability , meteorology , spatial correlation , particulates , image resolution , grid , spatial ecology , remote sensing , geography , computer science , statistics , mathematics , geodesy , ecology , artificial intelligence , biology
Haze has been a focused air pollution phenomenon in China, and its characterization is highly desired. Aerosol properties obtained from a single station are frequently used to represent the haze condition over a large domain, such as tens of kilometers, which could result in high uncertainties due to their spatial variation. Using a high‐resolution network observation over an urban city in North China from November 2015 to February 2016, this study examines the spatial representativeness of ground station observations of particulate matter with diameters less than 2.5 μm (PM 2.5 ). We developed a new method to determine the representative area of PM 2.5 measurements from limited stations. The key idea is to determine the PM 2.5 spatial representative area using its spatial variability and temporal correlation. We also determine stations with large representative area using two grid networks with different resolutions. Based on the high spatial resolution measurements, the representative area of PM 2.5 at one station can be determined from the grids with high correlations and small differences of PM 2.5 . The representative area for a single station in the study period ranges from 0.25 to 16.25 km 2 but is less than 3 km 2 for more than half of the stations. The representative area varies with locations, and observation at 10 optimal stations would have a good representativeness of those obtained from 169 stations for the 4 month time scale studied. Both evaluations with an empirical orthogonal function analysis and with independent data set corroborate the validity of the results found in this study.