Research On Near-Surface No₂ Estimation Optimization And Integrated Mapping Supported By GOMR-2B/OMI/TROPOMI

Nitrogen dioxide（NO₂）is an important pollutant gas in the tropospheric atmosphere.Near-surface NO₂（NS-NO₂）is closely related to human health.While top-down approaches have been widely applied to estimate NS-NO₂ using Ozone Monitoring Instrument（OMI）tropospheric NO₂ vertical column density（VCD）,there still exist low accuracy and underestimation in high concentration areas when only use OMI data.In addition,due to the lack of OMI satellite data,the estimated results based this data has low spatial coverage.In response,this study using satellite data as comparison variables,meteorological station data and other auxiliary data as public variables,based the random forest algorithm,designed four sets of experiments（combined with Global Ozone Monitoring Experiment（GOME-2B）and OMI,only OMI or GOME-2B is used,no satellite data is used）to estimate the NS-NO₂ concentration in mainland China to explore the impact of satellite data at different transit times on the estimation results.The results are as follows:（1）The model estimation results combining GOME-2B and OMI have the best fit with ground observations,the verification results:R²=0.80,RMSE=9.0μg/m³;The underestimation in the high-concentration area and the overestimation in the low-concentration area have been significantly reduced;（2）The NS-NO₂ has obvious seasonal changes,with the highest concentration in winter,the lowest in summer;（3）The NS-NO₂ concentration in 2014-2018 dropped significantly,the proportion of the population of the areas where the NO₂ concentration not up to the standard decreased from 20.42%to 10.42%,and there are obvious provincial differences.In addition,this paper also uses TROPOspheric Monitoring Instrument（TROPOMI）and OMI data,combined with GOME-2B data and other auxiliary data,based on the RF and XGBoost models to estimate daily NS-NO₂ concentration in central and eastern China,respectively.After comparative and analysis of the results,a data integration method was developed by combining the advantages of the two data sets,and then a set of integrated data sets with high temporal-spatial resolution and high coverage were obtained.The results are as follows:（1）The XGBoost model based on the TROPOMI dataset has the best result,the verification result:R²=0.87,RMSE=6.3μg/m³;（2）The space coverage of estimated result based on the TROPOMI dataset（0.856）is higher than based on the OMI dataset（0.346）.At ground stations,the two sets of estimation results have the same accuracy,in high-concentration areas without stations,the estimation results based on TROPOMI are underestimated;（3）The integrated data combined with the two sets of data has higher spatial coverage and accuracy.