Thursday, 15 January 2009: 2:15 PM
The Impact of GPS Radio Occultation Data on Typhoon Prediction: An OSSE Study of Typhoon Shanshan (2006)
Room 130 (Phoenix Convention Center)
Impact of GPS radio occultation (RO) measurements on Typhoon Shanshan (2006) was investigated using an Observing System Simulation Experiments (OSSE) approach. Both the local operator and nonlocal operator were employed to assimilate the GPS RO refractivity soundings with WRF 3DVAR. To generate the GPS RO refractivity soundings, a 2-D raytracing model was used to obtain the bending angles of hypothetical rays as they penetrate in the model atmospheric state produced by a MM5 simulation of 20-km resolution for Shanshan. A fixed azimuthal angle of 45o was used for all the rays with a tangent point ranging from about 300 m to 18 km in the vertical. The refractivity soundings at 300-km resolution then were produced from the Abel transform of the modeled bending angles as the RO retrieved from the data center. The model state then was degraded to simulate the large-scale analysis. In this study, the impact of the RO refractivity soundings located at the upstream environment of the typhoon vortex and the Pacific high was investigated, in particular. After assimilation of these soundings (as target observations), there is a cold dome produced in model initial increments near the oceanic surface, which encloses the sounding zone associated with an anticyclonic circulation with flow intensity up to 10 m s-1. It was found that this intense flow is induced due to a small difference of 2-3 units in refractivity between the (true) model state and the retrieved soundings, regardless of use of the local operator or nonlocal operator of 3DVAR. In reality, these small differences over the ocean (as cold biases) account only for 1% errors which are comparable with the accuracy of the routinely data retrievals. Although, the model resolution is only 20 km and has assumed a uniform angle of the rays, the retrieved refractivity soundings should reflect an upper bound of the accuracy of RO refractivity. As expected, such errors tend to dominate the model performance of track prediction, in spite that the Shanshan's track in the degraded model state can be improved when the true refractivity soundings are assimilated.
Supplementary URL: