Study On Temporal And Spatial Variation Monitoring Of Atmospheric Water Vapor And Short-term Impending Precipitation Prediction Based On BDS-3

Atmospheric water vapour plays an important role in the Earth’s atmosphere,it is inseparable from the formation of various precipitation such as rainfall and snowfall,and deeply affects the working life of human society,so it is particularly important to accurately and efficiently detect atmospheric water vapour.At present,it is mainly used as an auxiliary tool to traditional atmospheric water vapour detection methods.After a three-step development strategy,on July 31,2020,the Beidou-3 system will be fully completed,marking the official provision of services to the world.In this thesis,we use the observation data of Beidou-3 provided by Shandong continuously operating reference stations(SDCORS)as the basis for atmospheric water vapour inversion and obtain atmospheric water vapour distribution information.The main research contents of this thesis are as follows:(1)The current status of domestic and international research on GNSS atmospheric water vapour inversion and precipitation forecasting is analysed,the development history and unique advantages of China’s Bei Dou satellite navigation system are introduced,and the advantages of GNSS inversion of water vapour over traditional methods are discussed.The principles of the specific methods involved in the research of this thesis are elaborated in conjunction with Eq.(2)Atmospheric water vapour(PWV)inversion is carried out using GAMIT software on Bei Dou-3 observations,and the PWV inversion performed by Bei Dou-3is compared and analysed with that provided by meteorological sounding stations and ERA5 reanalysis data.The results show that Bei Dou-PWV is in good agreement with sounding-PWV and ERA5-PWV,and the root mean square error,mean deviation and correlation coefficient can meet the needs of meteorological research.(3)Combined with the actual precipitation information,the spatial and temporal variations of PWV before and after precipitation were analysed in detail from the perspective of a single CORS station and 50 CORS stations across the province,respectively.The results show that the high value areas of PWV overlap highly with the precipitation generation areas,the changes of PWV are highly correlated with the precipitation generation,and PWV is indicative of the generation of precipitation.The seasonal distribution characteristics of PWV were analysed,and the results showed that PWV has obvious seasonal differences.(4)Combining the spatial and temporal variation characteristics of PWV,the PWV maximum,PWV variation rate and PWV variation value were selected as forecast factors,and the corresponding one-factor precipitation forecast models were constructed using data from June to August for two years from 2020 to 2021,and the accuracy of the models was verified using data from 2022.The results show that the applicability of the three single-factor models varies in different months,and the correct and wrong forecast rates are both high.(5)In response to the high misreporting rate of the single-factor model,a multi-factor precipitation forecasting model incorporating three single factors was constructed,divided into an equal-weighted multi-factor model and a non-equal-weighted multi-factor model.Compared with the single-factor model,the equal-weight multi-factor model significantly reduces the false alarm rate and improves the accuracy of the model while ensuring the correct forecast rate.(6)Using a back propagation neural network method,a 6-factor precipitation forecasting model including annual cumulative day,time series,PWV,PWV maximum,PWV variation value and PWV variation rate was constructed using the PWV variation feature as the main forecasting factor,and the accuracy of the model was slightly improved compared with that of the equal weight multi-factor model.Afterwards,an 8-factor precipitation forecast model was constructed by adding two meteorological elements,namely temperature and barometric pressure,to the 6-factor model.Compared with other models,the 8-factor model has a high correct forecast rate and a low misreporting rate,which has a strong practicality.