November 13, 2008 Meeting Summary |
John Derber presented Daryl Kleist's work on assimilation of pseudo-MSLP
observations for TCs. Pseudo-MSLP observations involve generating
observations for every TC from tc_vitals files producing a single MSLP
observation for each TC. For weaker storms, where observation minus
forecast values are small, the storm position is reinforced. For
stronger storms, which have larger observation minus forecast values,
the storm position is reinforced and the storm is strengthened. It was
shown that this method improved TC initial intensity and position in the
GFS. Track was not vastly improved. Future work would include running
more cases to assess impact on forecast tracks and storm environment.
Mingjing Tong presented her work on assimilation of airborne Doppler
radar data with GSI for hurricane initialization and prediction. The
impact of assimilating airborne radar radial velocity data is examined
for two hurricane cases, Karen (2007) and Felix (2007). The new aspect
of the analysis is the use of the flow-dependent anisotropic background
error covariance, which produced better analysis than the isotropic one.
The use of radar data showed some positive impact on intensity forecast,
and no negative impact was found for track forecast if anisotropic error
covariance is used. Future work includes more tuning and further
analysis of results as well as running more cases for stronger and
weaker storms.
Mingjing Tong presented her work on assimilation of airborne Doppler
radar data with GSI for hurricane initialization and prediction. The
impact of assimilating airborne radar radial velocity data is examined
for two hurricane cases, Karen (2007) and Felix (2007). The new aspect
of the analysis is the use of the flow-dependent anisotropic background
error covariance, which produced better analysis than the isotropic one.
The use of radar data showed some positive impact on intensity forecast,
and no negative impact was found for track forecast if anisotropic error
covariance is used. Future work includes more tuning and further
analysis of results as well as running more cases for stronger and
weaker storms.