GNSS Tropospheric Delay Prediction And Analysis Of Water Vapor Spatial-temporal Characteristics During Typhoons

Zenith tropospheric delay(ZTD)is a key factor affecting the precise positioning of Global Navigation Satellite System(GNSS).With the rapid development of artificial intelligence in recent years,more and more intelligent algorithms are used for time series analysis.Typhoon is an important weather phenomenon.During typhoon,it is often accompanied by a large range of extreme rainfall,causing serious natural disasters,serious economic losses and life hazards.The important components of the atmosphere during water vapor play an important role in meteorology.With the continuous development of GNSS technology,Precipitable Water Vapor(PWV)based on GNSS inversion has been widely used in the analysis of extreme climate events in the north,and PWV is obtained by tropospheric wet delay calculation.Therefore,the prediction of tropospheric delay was studied,and the spatial and temporal characteristic distribution of GNSS water vapor and rainfall during typhoon was analyzed.It is of great practical significance for GNSS to accurately locate,to establish regional tropospheric delay model,to forecast meteorological disasters such as extreme rainfall and typhoon warning.In order to improve the prediction accuracy of ZTD,a Gaussian Process Regression(GPR)prediction model based on phase space reconstruction is proposed.In view of the chaotic characteristics of ZTD time series,using the ZTD data provided by the International Global Navigation Satellite System Service(IGS)station.Firstly,the embedded dimension was determined by the Cao method,and then the phase space reconstruction of ZTD data was carried out to explore the accuracy and accuracy of the Gaussian Process(GP)model in predicting tropospheric delays at12 stations of different latitude levels in the southern and northern hemispheres.In order to verify the effectiveness of the model,the prediction results were compared with the original data and the Back Propagation(BP)neural network model,and the influence of different time on the prediction accuracy of ZTD was further explored.Finally,the influence of longitude and altitude on the prediction accuracy of ZTD was analyzed.The results show that the Root Mean Square Error(RMSE)of GP model reaches mm level,the correlation between GP model and theoretical value reaches0.997,and the prediction accuracy index is obviously better than BP neural network model.The prediction accuracy of GP model in the southern hemisphere is higher than that in the northern hemisphere,and RMSE in the high latitude area is less than3.6 mm,which is more suitable for the tropospheric delay prediction in the high latitude area.In the time domain of the study,the prediction accuracy of GP model at night is higher than that in the day at most sites,the longitude has no obvious influence on the prediction accuracy of ZTD,and the altitude is proportional to the prediction accuracy of ZTD.GP model has better practicability and feasibility for the prediction of tropospheric delay.In order to analyze the spatial and temporal characteristic distribution of PWV and rainfall during typhoon,this paper explores the variation characteristics of PWV and rainfall during typhoon “Nanmadol” from different regions,space,time and frequency domains.Firstly,Taiwan was divided into 5 regions,and the distribution characteristics of average PWV and cumulative rainfall of 77 stations in different regions were explored.Secondly,the variation of the cumulative rainfall and the average PWV with distance and local time(from 12:00 on August 27,2011 to 12:00on August 31,2011)of typhoon “Nanmadol”in 6h was calculated.Furthermore,five GNSS stations in different regions of Taiwan were selected,including ANKN station in the north,SGUN station in the west,YUSN station in the middle,DSIN station in the east,and HENC station in the south.In order to explore the occurrence mechanism of typhoons,5 typhoons from 2008 to 2011,including “Nanmadol”,were selected to analyze the correlation between PWV and wind speed and pressure.Finally,based on the typhoon megi,which also made landfall in southern Taiwan,the spatiotemporal variation characteristics of the two were further verified.The results are as follows:(1)According to the characteristics of PWV and rainfall in different regions of Taiwan,during typhoon “Nanmadol”,the distribution of PWV and rainfall in the southwest of Taiwan was significantly different,the average PWV in the southwest was the largest,the accumulated rainfall in the south was the largest,and the PWV decreased significantly with the increase of altitude.(2)From the typhoon during PWV and the analyses of rainfall on the space and time,there is significant increase or decrease trend PWV and rainfall,and both the change of the trajectory and the movement of the typhoon track is roughly same,PWV significant changes with the typhoon activity,the typhoon approaching Taiwan,PWV levels will be increased dramatically in a short time,and after a few hours with heavy rainfall and the continuous heavy rainfall.(3)According to the analysis of PWV and rainfall in the time-frequency domain during typhoon,there is a positive correlation between rainfall and PWV on the whole,with the correlation coefficient mainly between 0.8 and 1,and it has passed the 95% significance test.(4)From the analysis of wind speed,air pressure and other typhoons,it is found that PWV is negatively correlated with air pressure,and the difference of wind speed and air pressure of typhoons plays an important role in water vapor transport during typhoons.During typhoon“Megi”,there was a strong positive correlation between PWV and rainfall,with the correlation coefficient reaching 0.7.The PWV first changed dramatically,followed by heavy rainfall,and then continued to affect the change of PWV,which was basically consistent with the analysis results of typhoon “Nanmadol”.