Authors: Yingying Mei*, Department of Sociology, Wuhan University, Jiayi Li, School of Remote Sensing and Information Engineering, Wuhan University, Xin Huang, School of Remote Sensing and Information Engineering, Wuhan University
Topics: China, Environment
Keywords: surface ozone, Bayesian maximum entropy, hard data, soft data
Session Type: Paper
Start / End Time: 5:00 PM / 6:40 PM
Room: 8217, Park Tower Suites, Marriott, Lobby Level
Presentation File: No File Uploaded
In recent years, surface ozone pollution in China has become an increasingly prominent problem. Up to now, the spatial resolution of the current studies to predict the spatiotemporal distribution of ozone concentration is relatively rough (i.e., 10km or even coarser) .
As for spatiotemporal prediction of daily 1km surface ozone concentration across China, we first develop a generalized linear model (GLM) between average 8-hr daily ozone concentrations for each observed station-day and relative explanatory data based on remote sensing high spatial-temporal processing (temperature, precipitation, relative humidity, road density, latitude/longitude and day of year). Then we construct a Bayesian maximum entropy (BME) model incorporating both O3 monitoring station data and the GLM outputs.
Measurements in 2015 are tested. China is divided into several 5 degree *5 degree grid units and 5 sample units are selected from different climate zone. For all eligible measured ozone concentration data of each sample unite, 70% station-days are selected as “hard” data. “Soft” data were ozone levels and estimated errors of the remaining 30% station-days (testing samples) estimated from the GLM model for 1km*1km grid cells. It is clearly shown that the sums-of-squares ranges from 0.48 to 0.61 which indicate the method is suitable for China daily 1km surface ozone concentration estimation. Further research will focus on predicting nationwide daily surface ozone concentration on a 1km*1km grid.