July 16, 2009 Meeting Summary
Bob Tuleya presented his work on Noah LSM runs for Fay and Hanna. Bob noted that he used the H209 configuration for the HLSM runs. Looking at a plot of forecasted stream flow, it was shown that stream flow is higher in the HWRF than in NLDAS and NAM, especially in the southeast area of the domain. This could be attributed to subsurface GFS moisture being too high. The goal of this work was to run more cases to test the HWRF and the stream flow routing scheme. Bob then presented plots of track error for Fay and Hanna. In these and future plots, HWRF is represented by the blue line, H209 by the red line, and HLSM (the H209 using the LSM) by the green line. For Fay, H209 has a higher error for track than HWRF (production). The HLSM track error is lower than H209 and closer to the error for HWRF. For Hanna, HWRF actually shows a higher error value from forecast hours 72 onward. The HLSM error value is lowest at this point, thus, it was shown that HLSM has improved track errors over both H209 and HWRF. Looking at model track forecasts for the entire forecast period for each model separately shows that the GFS track forecasts were much closer to the best track than forecasts from HWRF, which had many outliers.When looking at the H209 and HLSM tracks, we see that H209 had more outliers than those for HLSM. As Bob mentioned, outliers really affect statistics. To conclude, Bob showed two track forecasts for Hurricane Fay, one that showed a good forecast by HLSM with a much improved track over H209 and HWRF and a second forecast where all three models made a bad forecast. Overall, it was shown that HLSM reduces the track errors seen with H209, but intensity issues still remain. For his future work, Bob mentioned running more cases, such as for Ike and Gustav, and refining the HLSM system.
Next Vijay Tallapragada presented a few slides about his work with HWRF diagnostics for landfalling storms. The H209 configuration, which included gravity wave drag, initialization modifications, and bug fixes for radiation and land surface temperature, produced a negative impact on track and intensity forecasts and a high positive intensity bias for Fay, Hanna, Gustav, and Ike from the 2008 hurricane season. To improve the H209 configuration, H209 and operational HWRF output will be examined to see what kind of effects the following items have on track and intensity forecasts: 1) Reducing moisture supply for storms initialized over land thus reducing cyclogenesis over land; 2) Assessing the impact of the bug fixes on radiative, heat, and momentum fluxes, convective and large-scale heating rates, and stability analysis in the PBL; 3) Comparing MPI, IKE, oceanic heat content, net SST cooling, surface temperature, and evaporation over land; and 4) Tuning gravity wave drag (GWD) and mountain blocking parameters. Vijay then showed intensity error plots for Fay and Hanna together and Gustav and Ike together. For Fay and Hanna, which were relatively weaker storms, HWRF intensity error was less than that for H209. H209 had a higher positive bias as well. For Gustav and Ike, which were stronger storms, H209 error is larger than that for HWRF, but the differences are smaller. Preliminary experiments (represented by the yellow HXCS line) have been conducted for Fay using reduced cyclogenesis over land--or no warm core in the composite storm. This decreases the intensity component compared to H209.
Hyun-Sook Kim also showed a few slides of her recent work with Gustav simulations. She noted that for her work, aGus08 represents experiments with GWD and bGus08 represents experiments with OUT GWD. Hyun-Sook noted that all the experiments shown here used H209 with HYCOM, and the only difference in a and b is the use of GWD. Looking at the wind stress fields, the fields are more symmetric and the intensity is much stronger in the experiments without GWD. Hyun-Sook also noted that over the whole domain, the storm was very deep in the experiments without GWD and very shallow with GWD turned on.