A numerical study of the effect of GOES-8 sounder data on the prediction of Hurricane Felix

In this paper, we discuss a procedure to assimilate satellite radiance data into the initial condition for a NWP model. We next examine the impact of GOES-8 sounder data on the predicted intensity and track produced by an 84-h numerical weather prediction of Hurricane Felix. A mesoscale adjoint modeling system provides a framework within which to assimilate the radiance data, conventional data (surface and raobs), and ship reports of sea level pressure observations, the latter which serve to define a realistic hurricane vortex in the initial condition. We discuss the quality-control process for the satellite data and the bogus data assimilation (BDA) procedure used to generate the initial condition for the forecast. The radiance data is shown to affect several features in the model initial condition and produced identifiable improvements in the subsequent forecast when compared with 1) a control forecast started from initial conditions based upon a conventional analysis and observations only; 2) a forecast started from initial conditions in which the BDA procedure was used with conventional and sea level pressure observations (no satellite radiance data) to initialize the hurricane vortex. We present a discussion of the synoptic and mesoscale features that were enhanced or changed as a result of including the satellite data. We also discuss evidence of the impact of the satellite data assimilation on the general structure of the hurricane and associated precipitation pattern. The most notable effect of the satellite data was to improve the forecast track and intensity of Hurricane Felix. This improvement was evident even after 84-h of simulation.