The accuracy and feasibility of computing the zenith tropospheric delays (ZTDs) from data of the European Center for Medium-Range Weather Forecasts (ECMWF) and the United States National Centers for Environmental Prediction (NCEP) are studied. The ZTDs are calculated from ECMWF/NCEP pressure-level data by integration and from the surface data with the Saastamoinen model method and then compared with the solutions measured from 28 global positioning system (GPS) stations of the Crustal Movement Observation Network of China (CMONOC) for 1 year. The results are as follows: (1) the error of the integration method is 1-3 cm less than that of the Saastamoinen model method. The agreement between the ECMWF ZTD and GPS ZTD is better than that between NCEP ZTD and GPS ZTD; (2) the bias and root mean square difference (RMSD), especially the latter, have a seasonal variation, and the RMSD decreases with increasing altitude while the variation with latitude is not obvious; and (3) when using the full horizontal resolution of 0.5°9 0.5°of the ECMWF meteorological data in place of a reduced 2.5°9 2.5°grid, the mean RMSD between GPS and ECMWF ZTD decreases by 4.5 mm. These results illuminated the accuracy and feasibility of computing the tropospheric delays and establishing the ZTD prediction model over China for navigation and positioning with ECMWF and NCEP data.Keywords GPS Á Zenith tropospheric delay (ZTD) Á CMONOC Á ECMWF Á NCEP Abbreviations CMONOCThe crustal movement observation network of China CDAS Climate data assimilation system ECMWF

The 2015 St. Patrick's Day geomagnetic storm caused numerous disturbances of the ionosphere, particularly, plasma irregularities, large‐scale traveling ionospheric disturbances, and equatorial ionization anomaly enhancement. This study for the first time quantifies the global‐scale impacts of the ionospheric disturbances on Global Positioning System (GPS) precise point positioning (PPP) solutions during this extreme space weather event by taking advantage of 5,500 + GNSS stations installed worldwide. The overall impact was more severe at high latitudes, while PPP degradation at low latitudes was associated with different types of ionospheric disturbances. Specifically, our results show that kinematic PPP solutions degraded following an intensified auroral particle precipitation during the storm's main phase (06–23 UT) when up to ~70% of the high‐latitude stations experienced degraded position solutions in the multimeter range at 16–18 UT. Around magnetic noon and midnight, the storm‐induced plasma irregularities caused notable PPP errors (>10 m) at high latitudes. Interhemispheric differences were observed with a more severe impact seen in the Southern Hemisphere, where PPP outage lasted for ~12 hr during the second main phase (12–23 UT). At low latitudes, post sunset equatorial plasma irregularities were suppressed across most longitudes, but large PPP errors (>2 m) associated with storm‐induced plasma bubbles were registered at the Indian sector at 14–18 UT. The storm‐induced equatorial ionization anomaly enhancement and large‐scale traveling ionospheric disturbances were responsible for the low‐latitude PPP degradation at dayside sectors. This study fills the research gap between physical and practical aspects of severe ionospheric storm effects.

The vertical structure of water vapor in atmosphere is one of the initial information of numerical weather forecast model. Because of the strong variation of water vapor in atmosphere and limited spatio-temporal solutions of traditional observation technique, the initial water vapor field of numerical weather forecast model can not accurately be described. At present, using GPS slant observations to study water vapor profile is very popular in the world. Using slant water vapor(SWV) observations from Shanghai GPS network,we diagnose the three-dimensional(3D) water vapor structure over Shanghai area firstly in China. In water vapor tomography, Gauss weighted function is used as horizontal constraint, the output of numerical forecast is used as apriori information, and boundary condition is also considered. For the problem without exact apriori weights for observations, estimation of variance components is introduced firstly in water vapor tomography to determine posteriori weights. Robust estimation is chosen for reducing the effect of blunders on solutions. For the descending characteristic of water vapor with height increasing, non-equal weights are used along vertical direction. Comparisons between tomography results and the profile provided by numerical model (MM5) show that the forecasted moisture fields of MM5 can be improved obviously by GPS slant water vapor. Using GPS slant observations to study 3D structure of atmosphere in near real-time is very important for improving initial water vapor field of short-term weather forecast and enhancing the accuracy of numerical weather forecast.Keywords: Shanghai GPS network, GPS slant water vapor, tomography, 3D structure of water vapor, robust estimation of variance component.The vertical profile of water vapor is very important for revising and improving the initial moisture field of mid-scale numerical weather model. At present, the three-dimensional (3D) information of water vapor is provided mainly by radiosonde. However, the distances between any two radiosonde stations exceed 200-300km and the observing time interval is 12 h, which is not enough for observing storm system smaller than 100km and can not satisfy real time short-term weather forecast. The accuracy of water vapor retrieved from meteorology satellite is yet too low. While PWV (Precipitable Water Vapor) with high accuracy and resolutions can be retrieved from ground-based GPS network in real time [1] . However, PWV is the integrated amount of water vapor over each station which can not reflect the 3D structure of atmosphere. Commonly, it is considered difficult to retrieve vertical profile of water vapor from ground-based GPS which is the choke of ground-based GPS meteorology and limits the further application of GPS technique in meteorology. In fact, GPS slant observations include the vertical information of water vapor. Using GPS slant observations from a dense network, 3D structure of water vapor can be retrieved by appropriate tomography technique. In these years, Flores et al. [2] and Lubomir ...

Slant delay data obtained from global positioning system (GPS) observations carry valuable meteorological information. The spatial distribution of the water vapour can be reconstructed from such slant delays. To estimate the quality of the GPS slant delays two validation studies were carried out. One study was based on the observations of a water vapour radiometer, a second on the analysis fields of a numerical weather model which were used to compute the corresponding GPS delays. Both studies yielded a high correlation between the available slant delays at higher elevation angles but showed deficiencies at low elevations. The mean bias between the GPS zenith delays and the radiometer data is 1.18 mm with a RMS of 6.0 mm. The corresponding bias and RMS of the GPS vs. model comparison are 3.3 mm and 2.9 mm. ZusammenfassungDie Laufzeitverzögerung der GPS-Signale in der Atmosphäre kann wertvolle meteorologische Informationen liefern. Insbesondere ist es möglich, die räumliche Wasserdampfverteilung in der Troposphäre aus den Laufzeitdaten zu rekonstruieren. Hierzu muß jedoch zunächst die Qualität der GPS-Daten durch ValidierungsStudien ermittelt werden. Hier werden zwei Studien beschrieben: In der ersten werden die GPS-Daten mit denen eines Wasserdampf-Radiometers verglichen, in der zweiten werden die Analysen eines numerischen Wettermodells genutzt, um die GPS-Signal-Laufzeiten zu berechnen und mit den Beobachtungen zu vergleichen. In beiden Fällen ergibt sich eine guteÜbereinstimmung für größere Elevationen, bei niedrigen Elevationen treten jedoch größere Abweichungen auf. Insgesamt ergibt sich eine mittlere Abweichung von 1,18 mm und ein RMS von 6,0 mm zwischen den Zenitdelays aus Radiometerbeobachtungen und der GPSAnalyse und eine Abweichung von 3,3 mm mit einem RMS von 2,9 mm zu den Modelldaten.