April 23, 2009 Meeting Summary
Yihua Wu presented some of his recent work comparing the 2009 HWRF and the HWRF using the NOAH LSM. Results from four storms were shown, Hanna, Fay, Dean, and Katrina. In Yihua's graphs N883 (represented by red lines in track and intensity plots) is the 2009 HWRF or ctl and N893 (represented by blue lines) is the 2009 HWRF with the NOAH LSM or test. Yihua noted that HWRF runs were performed on Cirrus and the HWRF code used contained the land temperature bug fix.
For Hurricane Hanna, tracks and intensities for N883 and N893 are almost identical through 72h, with higher intensity values than was observed. At 72h, they diverge slightly with N893's wind max peaking slightly later than N883's. Rainfall swaths for Fay (with N883 values on the left and N893 values on the right) are very similar for both runs, however, the rainfall for N893 has a higher maximum value that is also larger in spatial extent.
For Hurricane Fay, track values are almost identical for N883 and N893 through almost 72h before they diverge with N893's track more westward than N883's. Intensity plots show the same lag in peak as seen for Hanna with values for the model runs much larger than what was observed. Once again, the rainfall swath for N893 (on the right) has a higher maximum value that is spatially larger than that for N883 (on the left).
Hurricane Dean's track for N883 and N893 were once again almost identical through about 72 hours before a very slight divergence. The intensity plots show model values slightly less than what was observed with slight differences between N883 and N893 seen throughout the forecast period. The prominent lag seen with Hanna and Fay is not really present here. Rainfall swaths for Dean show larger max values for N883 than N893 with more rain produced north of Cuba in the N893 run than in the N883 run.
Track errors for Hurricane Katrina showed very similar values for N883 and N893. These errors were slightly higher for HWRF experiments than for GFDL. Hurricane Dean track errors for N883 and N893 were very similar through about 72h before higher values are observed for N893. For this storm, errors for the HWRF experiments were lower than those seen for GFDL.
Yihua mentioned that the HWRF runs for today's results were cold starts. Future work includes using cycled HWRF runs to get results as well as more investigation into the lag seen using the NOAH LSM. Overall, use of the NOAH LSM produced small differences in track and intensity compared to the 2009 HWRF. However these differences would be more valuable when used to predict inland flooding.