Study On Prediction Of PM_2.5 Mass Concentration Of Combined Water Vapor Factor

As the calamity of the haze has gradually been exposed,it has now become the murderer that restricts the development of our local economy,endangers people's physical safety,and undermines China's sustainable development strategy.In the face of growing haze disasters,how to establish real-time and accurate haze monitoring and forecasting systems is a top priority.In recent years,with the continuous development of GPS meteorology,its comprehensive,all-weather,real-time,high-precision,and low-cost inversion of atmospheric vapor technology has not only promoted China's ability to predict heavy rainfall,but also caused haze disasters.The monitoring and forecasting provide new ideas.Therefore,this paper will further explore the application potential of GPS meteorology and neural network algorithm in real-time and effective monitoring and forecasting of haze and haze disasters,and exert their roles and advantages in environmental monitoring and forecasting,which will help prevent fogging.The recurrence of environmental disasters caused by earthworms has important research value for the prevention and treatment of haze disasters.For seven cities with severe smog pollution in China,the paper introduced the harm of smog and analyzed the factors of its formation,and pointed out the need to study how to reduce the harm of smog to humans,and made the following the study:1.The average monthly PM_2.5 concentration values and air quality index AQI in cities in2015 and 2016 were relatively small in summer and autumn,followed by spring,and winter was significantly higher than other seasons,showing significant seasonal changes.And it has a good corresponding relationship in different months,the correlation coefficient is concentrated between 0.771-0.999,which is highly correlated,so PM_2.5 quality concentration can be used as an important indicator to reflect environmental pollution;2.Using the CMONOC reference station data to obtain data on the total tropospheric zenith delay ZTD and the NOAA weather station data for the combined site,and inverting Saastamoinen model to obtain seven haze disasters in China in 2015 and 2016 in each of the four seasons.Precipitation PWV data were compared with the corresponding PM_2.5 mass concentration data.The results showed that the GPS-PWV and PM_2.5 data of each city under the change of time series showed the overall performance of the two.Moderate positive correlation;3.By comparing and analyzing the prediction values of different cities in 2015,2016 and2016 under different forecasting factors and models,it is found that the accuracy of the data predicted by the GA-BP neural network model is higher than that predicted by the BP neural network model.,More stable,as a whole,the relative error of the GA-BP neural network prediction value using the GPS-PWV data compared to simply using the atmospheric environmental factors and meteorological factors that affect haze as the forecasting factor is small,and its predicted value Both accuracy and reliability have improved;4.Due to more rain in summer and autumn,the correlation coefficient between GPS-PWV and PM_2.5 in each city in 2015 and 2016 was generally larger in March and December,and the correlation was stronger in June.September is relatively small,and the correlation is weak;the forecast performance of cities in 2015 and 2016 as a whole shows that the forecast values in March and December are compared with those in June and September,and their accuracy and reliability higher.Combining the GPS-PWV data with the main atmospheric environmental factors and meteorological factors of haze as predictors of GA-BP neural network model will help improve the accuracy and reliability of PM_2.5 mass concentration data prediction.