July 30, 2009 Meeting Summary
At today's meeting Young Kwon presented some of his recent work on gravity wave drag (GWD) parameter experiments using four major Atlantic hurricanes from the 2008 season--Fay, Gustav, Hanna, and Ike. This presentation focused on results from the Fay and Ike runs, as the other runs have yet to be completed. These four storms were selected for this study because of their interaction with land and strength. These storms could also help determine definitively whether GWD will be included in future HWRF operational code. The H209R version of the HWRF is being used for these experiments, which is the 2008 operational HWRF with upgraded scripts, minus bug fixes. Three experiments are being conducted for each storm: H9RN represents H209R runs without GWD, H9RG is H209R runs with GWD, and HGJA is H209R runs with a tuned GWD suggested by Jordan Alpert. Young noted that since most of these storms were long-lived, both he and Janna O'Connor were completing HWRF runs to speed up the process.
Next, Young described the specific tuned parameters used in the HGJA code. First, the sigfac parameter was tuned from sigfac=0 to sigfac=1. This parameter basically scans for the maximum terrain height in a given grid box: elevmx = elevmx + sigfac*hprime, where hprime is the deviation of the terrain height difference in the grid box. The second tuned parameter is hncrit, which is the limitation of the minimum terrain height allowing mountain blocking. Specifically: elevmx = min(elevmx, hncrit). This parameter was originally equal to 8000, and it was modified to be equal to 4000 for HGJA. The last tuned parameter was cdmb, which can be described as the amplitude of subgrid mountain drag. It was modified from 1.0 to 4.0.
Young then showed plots of intensity and track errors for Fay and Ike, the two storms completed thus far. All plots feature H9RN, indicated by the magenta line, H9RG, indicated by the solid blue line, and HGJA, indicated by the dotted blue line. For Fay, the intensity error to 48h shows similar error values for all three experiments. At 72h, HGJA has a larger error value than the other experiments, however, past 72h, error values for H9RG and HGJA are lower than those for H9RN. The track error for Fay shows all three experiments with similar errors, all lower than for CLP5, through about 72h. Then, the error for H9RG and HGJA became higher than those for H9RN. Thus, while intensity error after 72h improved using GWD, track error got worse for Fay. For Ike intensity, the H9RN errors are lower than both experiments using GWD through 72 hours. After that time, the H9RG experiment has lower errors, but the HGJA run has the highest intensity errors. Through 48 hours, there are fairly large differences in the errors between all experiments, with H9RG having the highest errors at 48h. The track errors for Ike for the three experiments are all very similar and lower than errors for CLP5.
Young's future work plan includes a continuation of H209R GWD experiment runs for Gustav and Hanna. Future work also includes performing sensitivity experiments with wave breaking parameters, such as CLEFF. Finally, some East Pacific storms should be run using this configuration to see if the tuned GWD still has a positive impact in the East Pacific basin.