Atmospheric Precipitable Water Amount Retrieval And Its Preliminary Study UsingGround-based GPS Data

Water vapor and its variation is an important subject of meteorology and climatology. However, the scarcity in water observation greatly restricts the development of these disciplines and understanding of water vapor. GPS soundings retrieve water vapor content in atmosphere using delay of GPS signal caused by refractivity when passing through the atmosphere. With its high precision, all weather, high temporal resolution, high reliability and no calibration, it is a complementary to the regular atmospheric soundings, The data source it provides will undoubtedly bring great advancement to meteorology, environment and hydrology and so on.This paper focused on study ground-based GPS data processing and atmospheric parameter retrieval methods using software GAMIT . Analysis of data obtained from Beijing GPS/MET experiment in 2000 and Hongkong region in 2004 indicates that the retrieved ground-based GPS Precipitable Water Vapor have higher pricision. The retrieved GPS integrated water vapor has 3mm RMSE and 0.4mm BIAS compared with that from radiosounding in 2000 Beijing experiment. The analysis of distribution of integrated water vapor over a specific area can distinguish the evolution and source of water vapor. We also found that the time series of PW is close linked with the synoptic process, especially precipitation, intensive precipitation is associated with PW mushroom and appearance of the peak value. Therefore, the research on GPS PW will lead to the understandings about the role of water vapor in weather forecasts.Being regarded as the previous work of GPS data assimilation, this paper compared the PW evaluated by MM5 24h-simulaton with the retrieved PW of GPS observation. The result shows that the humidity error of MM5 reanalysis fields can be minimized by increasing the horizontal resolution of the model, can also slightly promote the ability to PW prediction, but not very obvious. The 24h prediction (step by step) can reflect the daily change trend of PW on the whole. Although have 5%-10% relative bias in the beginning of simulation, MM5 model still have a ability in PW prediction over 1h to 10h integration period, the PW prediction error increase obviously after 20h integration, consequently, high temporal resolution water vapor is necessary for meso-scale model in order to best simulate the model water distribution feature. In addition, GPS observation itself maybe have system error, so it is needed to take into account GPS self-reliability when assimilating the GPS data into the numerical prediction model.