P4.2
1DVAR analysis of temperature and humidity using GPS radio occultation data
Paul Poli, Meteo-France, Toulouse, France and NASA/GSFC, Greenbelt, MD; and J. Joiner, R. Kursinski, and M. Kolodner
The Global Positioning System uses a constellation of 24 satellites. Its signals are sensitive to the atmosphere. Using the radio occultation technique, it is possible to realize soundings by measuring the bending angles of the rays. The information can be converted into refractivity, and then related to temperature and humidity. The problem linked to the amibiguity in the interpretation of the refractivity is solved by a 1DVAR analysis.
A forward model (or observation operator) has been developed. It converts an atmosphere depicted as a function of pressure, into a refractivity profile, described as a function of the altitude. A theoretical study, confirmed by a Monte-Carlo simulation, shows the potential impact of the GPS within a 1DVAR analysis where both temperature and water vapor content are retrieved. The information content of GPS soundings is significant for the temperature at all latitudes, whereas the best improvement for humidity is expected in the tropics in the lower troposphere.
Real data were collected during the GPS/MET experiment in 1995. They demonstrate the strong capability of the GPS measurements to resolve the tropopause. The first guess used (GEOS Strat) has errors which can exceed 5K in this region. The use of GPS data is very sensitive to the gravitational constant used in the analysis; depending on the approximation on g, the retrieved temperature may vary by 1K. Due to the high vertical resolution of the GPS measurements (about 1 km), the analysis is performed on the 46 sigma levels of the guess in order to extract as much information as possible from the observations. A validation is made by comparing the retrieved profiles with nearby radiosondes. The chosen collocation criteria, +/-3 hours and no more than 280 km, enables comparisons with close to 150 occultations. In the Northern Hemisphere and in the Tropics, both bias and standard deviation of temperature are being reduced, as compared with the guess. The same statistics are also computed by comparing with the same radiosondes the direct retrieval of temperature assuming a certain humidity amount (and vice-versa with a direct retrieval of humidity).
This technique (1DVAR) does not require enormous computing facilities. The results show that an analysis using GPS data is a significant improvement over the background and a direct retrieval.
Poster Session 4, At Tribute to Lewis Kaplan: Part II
Wednesday, 12 January 2000, 9:30 AM-11:15 AM
Previous paper Next paper