Remote Sensing Estimation Of PM_2.5/1 Concentrations Over Eastern China From GOCI

Over the past 40 years of reform and opening up,with the acceleration of economic development and urbanization,various air pollutant emissions have increased,and air quality problems have become increasingly serious.Frequent smog weather in recent years has seriously affected people’s healthy life.Particles with aerodynamic diameter less than or equal to 2.5 microns(PM_2.5,also known as fine particles)are the main pollutants causing haze.More seriously,particulate matter with an aerodynamic diameter of 1 micron or less（PM₁）is more harmful than PM_2.5 because it has a smaller diameter and is more accessible to the respiratory circulatory system.The Aerosound Optical Depth（AOD）data and mathematical models provide an effective method for estimating the concentration of a wide range of particulate matter(PM,PM_2.5 and PM₁).Due to restrictions on high time resolution AOD products,there have rarely been studied for high spatial and temporal resolution assessment of near-surface PM in China.Based on the above background,this paper uses a Geostationary Ocean Color Imager（GOCI）AOD with a time resolution of 1 hour and a spatial resolution of 6 km as the main feature parameters,and establishes a spatio-temporal weighted regression model for estimating surface PM_2.5 over eastern China and PM₁ over Yangtze River Delta.The PM_2.5 concentration in eastern China and the PM₁ concentration in the Yangtze River Delta from January to December,2015 provide basic data and scientific basis intelligence for air quality management and health research in eastern China.The main results are as follows:（1）Construct a data set to evaluate GOCI satellite AOD products.The five representative AERONET sites closest to eastern China were selected,and 928 data were matched with GOCI.The Pearson correlation coefficient R² was 0.85,and the root mean square error RMSE was 0.16.This indicates that the quality of GOCI/AOD data is high and can be used for hourly PM estimation studies.（2）Develop a daytime hourly（0800-1500）space-time weighted regression model and evaluate the model accuracy.We use the Geographically and Temporally Weighted Regression（GTWR）model to correlate satellite AOD with near-surface PM.Linear search and grid search methods are used to determine the optimal model hyperparameters,which are evaluated by 10-fold cross-validation method.The results show that the estimated PM value of the satellite is significantly consistent with the measured PM value.The cross-validation results of the PM_2.5 validation set(R²=0.86,RMSE=17.31μg·m^-3,N=240,275)and the PM₁ validation set(R²=0.78,RMSE=14.51μg·m^-3,N=29,368)are at the leading level of high spatial-temporal resolution PM estimation.（3）Assessment of high temporal resolution near-surface PM_2.5 concentration in eastern China and PM₁ concentration in the Yangtze River Delta,and preliminary exploration of the quantity and spatial relationship between PM_2.5 and PM₁ in the Yangtze River Delta.The average annual concentration of PM_2.5 in eastern China and PM₁ in the Yangtze River Delta were 39.22±7.99μg·m^-33 and 41.50±6.76μg·m^-3,respectively.The trend of PM_2.5 estimated by GOCI is that it starts from 0800 first,reaches the peak at 1000,then starts to decrease,then the minimum at 1500.PM₁ and PM_2.5 have obvious spatial correlation.PM₁ is a part of PM_2.5,accounting for about80%,and the proportion is increasing according to the seasons of spring,summer,autumn and winter.（4）A Case analysis of air pollution incidents in the Bohai Rim region.The GOCI satellite remote sensor clearly records the pollution incidents in the Bohai Rim region on March 21,2015.Pollutants are transported from northwest to southeast from 08 AM to 15 PM,and the concentration of pollutants achieve highest PM_2.5 value of276.90μg·m^-3 with an average of 65.32μg·m^-3.The invaded area is over 300,000 square kilometers,which means that about 150 million people are affected by the air pollution of the dusty weather.