HR: 16:00h
AN: H44A-01 INVITED     [Abstracts]
TI: Continuous Assimilation of 3-hourly TRMM Rainfall and NCEPR2 with the FSUGSM
AU: * Cocke, S
EM: scocke@fsu.edu
AF: FSU, Dept. Meteorology, Tallahassee, FL 32306 United States
AB: Currently most data assimilation systems do not assimilate precipitation directly, including NCEP Reanalysis II (NCEPR2). Thus, precipitation derived from a model (even the analysis model) using analyses may differ considerably from observed precipitation. In order to improve the derived precipitation, as well as the associated heating and moisture profiles, a physical initialization procedure is used. The PI procedure has long been used to assimilate satellite derived daily precipitation in numerical models to improve the initial state. In this study, we will assimilate 3 hourly merged TRMM precipitation and NCEPR2 in a continuous assimilation period of a few months. Such "re-reanalysis" could be useful for seasonal climate studies, in much the same manner as the NCEP Regional Reanalysis Project which assimilated observed precipitation at high resolution. However, the focus of this study is on the fidelity of the precipitation assimilation, especially at shorter than daily time intervals. In addition, we seek to determine how dependent the assimilated precipitation is on the cumulus scheme. For these experiments we use the FSU Global Spectral Model in Assimilation Mode (AM) and in Forced Mode (FM). In AM the mass and momentum fields are nudged toward 6 hour NCEPR2, whereas the moisture field is directly modified so that the model rain rate approaches the observed rain rate. The temperature fields are indirectly modified via the cumulus scheme. For AM the Kuo cumulus scheme is currently used, though adaptation for other schemes is in progress. In FM all prognostic variables are nudged toward NCEPR2. Using FM with each of 6 different cumulus schemes, we compare the derived rain rates with observations. Preliminary results indicate that the FSUGSM-AM is able to assimilate the 3 hourly rain rates rather well, without degradation to the prognostic fields on the large scale. Results from using FM show that most cumulus schemes reproduce the assimilated rain, with varying degrees of success. With some cumulus schemes, precipitation spinup is a major issue. Using a simple tuning of the convective time scale, we are able to mitigate the spinup effect to some extent. The range of response to using the array of convective schemes can provide further insights to improving the assimilation method and the cumulus schemes as well.
DE: 1854 Precipitation (3354)
DE: 3314 Convective processes
DE: 3315 Data assimilation
DE: 3354 Precipitation (1854)
SC: Hydrology [H]
MN: Fall Meeting 2005