Using Sequential Gaussian Simulation to Assess the Spatial Uncertainty of PM2.5 in China

Abstract

Based on the observed PM2.5 concentration data in 2016, ordinary kriging (OK) and sequential Gaussian simulation (SGS) were used to map spatial distribution of PM2.5 in China, and SGS can model not only single, but also multi-location uncertainties, which assess the uncertainty of the PM2.5 spatial distribution. A smoothing effect was produced when using OK technique in mapping of PM2.5, however relatively discrete and fluctuant map was obtained by the SGS. Their results of spatial distribution show that east and west regions have higher PM2.5 concentration, middle regions have lower concentration in China. Based on the SGS realization, the probability that PM2.5 concentration at single location was higher than the defined threshold (10μg/m3) was big for the whole study area. The minimum value was 0.77. When the defined threshold changed to 35 μg/m3, the extent of higher probability was shrunk, the bigger value (0.8-1) existed in Xinjiang and North China. The probability which PM2.5 concentrations were higher than the defined threshold in several locations at the same time was also called joint probability. Given the critical probabilities (pm=1 and> 0.98), joint probability of PM2.5 in area a being higher than 10μg/m3 respectively is 0.85 and 0.5; while joint probability of PM2.5in area a being higher than 35μg/m3 respectively is 0. 65 and 0.14. The probability map can be very helpful for controlling and making environmental management decision of PM2.5 pollution.