March 19, 2009 Meeting Summary
Bob Tuleya presented his work on additional intensity analysis tools. These tools included the 10m maximum wind histogram and the chart of bias distribution as a function of intensity, both courtesy of Tim Marchok (GFDL). For his presentation, Bob looked at the 2008 Atlantic storms from Bertha through Paloma. The 10m wind max. histogram shows the number of occurrences of max wind speed (m/s) for the HWRF (red line), GFDL (green line), best track corresponding to GFDL (solid black line) and best trackh, which corresponds to the HWRF (dashed black line). Both the HWRF and GFDL produce too many intense storms (40-50 m/s range) compared to best track values and too few weaker storms (10-20 m/s range). The histogram for H209 vs. HWRF shows an underprediction of weaker storms (25-35 m/s range) by both the HWRF (solid red line) and H209 (dashed red line) and an overprediction by both the HWRF and H209 of storms in the 35-45 m/s range. Examination of the plot of bias as a function of intensity shows more positive bias (day-1 bias: red line) throughout from H209 compared to HWRF which had a low day-1 bias earlier on. As Bob noted, the biases are not weighted for these plots. Next, Bob discussed the sensitivity to HWRF surface wetness for Hurricane Fay. Seven cases for H209 were run as a "dry" condition where wet=0.3. The plot of 10m max winds versus forecast day shows many cases where H209 (CNTL) is more intense than the "dry" run (wet.3) and many cases where there "dry" run is more intense than the H209 value. When both the H209 (CNTL) and "dry" run are compared to the HWRF, the H209 and "dry" run are more similar and cyclogenic compared to the HWRF for Fay. As Bob noted, it does not seem like wetness is a major contributor to the model's overintensification.
Next, Qingfu Liu presented his work with the impacts of storm size correction on track and intensity forecasts, which he briefly discussed at last week's meeting. Qingfu ran tests for Hurricane Bertha using alpha=1.0, which meant no size correction in the HWRF initialization. This led to a very large storm size and a degraded track forecast. Track error plots for Bertha, where HWRF (in red) is the operational HWRF, H209 (in blue) is 2009 T&E version, H047 (in orange) is new GFS/GSI, and QL02 (in pink) is Qingfu's test, show QL02 with the highest track errors from 48 hours onward. For intensity error, the QL02 runs had peak error values (of all models) at 12h before decreasing to the lowest error values from 48-108h. The track plot for 2008071200 shows the QL02 (in yellow) with a much faster track compared to the other models. Intensity plots for the same day show QL02's peak wind max occurring at 12h before decreasing closer to observed values.