March 15, 2012 Meeting Summary
Young Kwon from the HWRF group at EMC presented recent results from 2010 HWRF upgrades.
Young began by outlining features of the 2012 HWRF. As a result of HFIP funding and help from multiple agencies, a high-resolution hurricane model (27-9-3km) is being implemented for the first time. The 2012 HWRF is the result of an extensive list of pre-implementation tests that were run on the Jet machine, courtesy of HFIP. Overall, the 2012 HWRF (or H212) showed significant improvement to storm structure. Upgrades to the 2012 HWRF include a vortex initialization at the 3km resolution, model code speed-up, adding I/O server, changing the time step of model integration to make the model more numerically stable, and a bug fix in the mask which caused a discontinuity in the lateral boundary. One dimensional ocean coupling was also added to the Eastern Pacific basin. Physics upgrades include adding a GFS shallow convection scheme, modification of several microphysical parameters, changing the critical Richardson number from 0.5 to 0.25, and using a constant Ch profile. GSI was also upgraded to version 3.5.
Next, Young showed statistics for track error, intensity error, and intensity bias. H212 runs for the 2011 hurricane season used the upgraded GFS and were run on the CCS machines. H212 runs for 2010 hurricane season used the older GFS and were run on Jet. For the 2011 Atlantic, H212 track error showed a maximum improvement of 20% over operational HWRF (2011 operational); intensity error for H212 was comparable to operational except at day 5; and intensity bias, while still slightly positive later in the forecast, was reduced, compared to operational. For the 2011 East Pacific, H212 track error showed a maximum 25% improvement over operational HWRF; intensity error showed a max improvement of 40% at day 5 with a slight degradation at day 1 and 2; and intensity bias for H212 was more negative than operational early in the forecast but was less positive than operational later in the forecast. For the 2010 Atlantic, H212 showed a maximum 25% improvement over the 2010 operational HWRF in track error; there was a maximum 15% improvement over operational HWRF early in the forecast for intensity error; and intensity bias for H212, while still negative, was less than operational HWRF. For the 2010 East Pacific, H212 track error was comparable to operational HWRF with a slight degradation due to Frank 09E; intensity error showed a max 20% improvement over operational; and H212 intensity bias showed a negative bias but less so than for operational.
Young then showed verification of storm structure for H212 through graphs of storm size, plots of PBL height, plots of the wind-pressure relationship, and bar graphs showing initial spin up/spin down values. In plots of 50kt wind radius error for 2011 Atlantic storms, H212 error values were much lower than operational HWRF in all quadrants, especially at later forecast times. The same was shown in 65kt wind radius error plots. This indicates that H212 is not expanding the storm with time as much as was seen in the 2011 operational HWRF. In a figure showing PBL height for the 2011082518 cycle of Irene 09L, the PBL appears more shallow for H212 than for operational HWRF. In a graph comparing the wind-pressure relationship for operational HWRF versus best track and H212 versus best track, H212 is much closer to best track values than operational HWRF. This indicates that H212 can sustain strong winds in the model without deepening the storm too much, which is consistent with the findings from the wind radius error plots. In bar graphs showing the initial 6-hour spin up/spin down values for max wind and min pressure, H212 showed reduction in spin up/down values compared to operational HWRF.