July 23, 2009 Meeting Summary
Mingjing Tong gave a presentation on her recent work with assimilating pseudo-MSLP data. Mingjing focused on Hurricane Gustav, and mentioned that for all data assimilation (DA) experiments, the first guess is the HWRF 6-h forecast plus the storm relocation only. Intensity and storm size corrections were not performed for the first guess. For the assimilations, only MSLP data or MSLP data plus all other data except radar data were assimilated. The plots that Mingjing showed were the result of sensitivity experiments which were performed by tuning the observation error of the MSLP data. Mingjing also noted that the goal of this work was to see how assimilating MSLP data could improve the hurricane assimilations.
For the plots showing surface pressure and 10-m wind and the E-W cross section of Gustav, the observed values for the 2008082600 UTC run are seen in red at the top left of the slide. The top row of plots represents no DA values, the second row shows values with an observation error of 1.35 mb, the third row uses an observation error of 0.6 mb and the last row of plots uses a 0.1 mb observation error. The min. pressure and max wind values for each observation err are listed on the left side of the slide. From this, Mingjing noted that as observation error decreased, the min. pressure and max. wind got closer to observed values. However, the storm structure was further from observations. Thus, for the other experiments, an observation error for 0.6 mb. was used.
Track and intensity plots were then shown for the 2008082600 UTC run of Gustav where only MSLP data were assimilated. The green line (red line for intensity) represents the H209 track while the blue line (NTCP) is for no DA. The rest of the tracks and intensities have varying observation error values as the only difference between them. Most of the tracks and intensities are quite similar. The next slide showed track and intensity plots for the 2008082912 UTC run of Gustav where MSLP and all other data except radar data were assimilated. Here, the cyan-colored line (for TCP6) for track and intensity was closer to observations for track and max winds.
Mingjing mentioned that she used the TCP6 observation error for cycled HWRF runs from 2008082912 UTC through 2008083006 UTC. For the track plots using TCP6 (blue line) compared to H209 (green), the TCP6 track is a bit closer to obs. than H209. Max wind plots show H209 (red line) much stronger than TCP6 (blue line) at all times. The same is true for minimum pressure. Mingjing noted that the largest difference between H209 and TCP6 could be seen at the 2008083006 UTC time. Mingjing concluded by saying that this could be due to a background error covariance issue. More investigation into this is ongoing.