**4. WRF model simulation**

**Figure 9** shows wind conditions at 1500 m (about 850 hPa) at the time of model integration. Compared to **Figure 2**, the global WRF reproduced large‐scale atmospheric circulations very closely and effectively. The color of the wind arrows represents magnitudes of wind speed; wind speed increases as the color changes from red to yellow to green to blue (0–46 m s–1). Arrows in green represent winds exceeding 10 m s–1. The yellow‐to‐purple scale in the background represents the 1500 m high water vapor in the range of 12–20 g kg–1. The defor‐ mation of the low‐level circulation by Hurricane Vince (October 8–11; **Figure 9a**) and the subsequent development of the anomalous low‐level cyclone over and northeast of the Caribbean Sea by the westerly winds from eastern North Pacific were simulated very realisti‐ cally (**Figure 9b**).

**Figure 9.** Samples of global WRF model simulation outputs: the anomalous circulation development of the subtropical high in the North Atlantic due to Hurricane Vince (October 8–11) (a) and the anomalous low‐level cyclone over, and northeast of, the Caribbean by the westerly winds from eastern North Pacific (b). Wind conditions at 1500 m (about 850 hPa) at the time of model integration are presented with arrows. The color of the wind arrows represents wind speed: wind speed increases as the color changes from red to blue (0–46 m s–1). The yellow to purple in the background repre‐ sents the water vapor in the range of 12–20 g kg–1.

However, the global WRF model simulation deteriorated by the seventh day after initializa‐ tion (**Figure 10**), preventing the model from replicating the TCG of Wilma and its subsequent development into a hurricane over the western North Atlantic (not shown). From **Figure 10**, it seems that the major errors occurred in the high latitudes and mid‐latitude systems that are directly affected by the high‐latitude circulations. In fact, during the prolonged global WRF simulation, the major failure occurred in reproducing the interactions between the mid‐ latitude trough over the northeastern US and the subtropical low to the east of the US. As a result, the erroneous forecast in intensity and location of the mid‐latitude trough over the northeastern US that was positioned to the northeast of the disturbance in the Caribbean predicted that the subtropical high would be restored in the North Atlantic, and vigorous easterly winds would resume over Central America (**Figure 11**) by October 16–20. This zonally enhanced low‐level wind condition is opposite to the meridionally enhanced, low‐level condition in the actual case of Wilma. In fact, the restored easterly winds over the Caribbean Sea impeded the development of a storm in the region.

**4. WRF model simulation**

102 Recent Developments in Tropical Cyclone Dynamics, Prediction, and Detection

sents the water vapor in the range of 12–20 g kg–1.

cally (**Figure 9b**).

**Figure 9** shows wind conditions at 1500 m (about 850 hPa) at the time of model integration. Compared to **Figure 2**, the global WRF reproduced large‐scale atmospheric circulations very closely and effectively. The color of the wind arrows represents magnitudes of wind speed; wind speed increases as the color changes from red to yellow to green to blue (0–46 m s–1). Arrows in green represent winds exceeding 10 m s–1. The yellow‐to‐purple scale in the background represents the 1500 m high water vapor in the range of 12–20 g kg–1. The defor‐ mation of the low‐level circulation by Hurricane Vince (October 8–11; **Figure 9a**) and the subsequent development of the anomalous low‐level cyclone over and northeast of the Caribbean Sea by the westerly winds from eastern North Pacific were simulated very realisti‐

**Figure 9.** Samples of global WRF model simulation outputs: the anomalous circulation development of the subtropical high in the North Atlantic due to Hurricane Vince (October 8–11) (a) and the anomalous low‐level cyclone over, and northeast of, the Caribbean by the westerly winds from eastern North Pacific (b). Wind conditions at 1500 m (about 850 hPa) at the time of model integration are presented with arrows. The color of the wind arrows represents wind speed: wind speed increases as the color changes from red to blue (0–46 m s–1). The yellow to purple in the background repre‐

**Figure 10.** Comparison of the large‐scale low‐level circulations between the global WRF model results after a 7‐day continuous simulation (a) and at the model initialization (b) both at 0000 UTC October 14 using FNL and RTG SST datasets.

**Figure 11.** A failed forecast of Wilma's TCG.

By contrast, the second simulation that was initialized at 0000 UTC October 14 successfully simulated the TCG of Wilma and its subsequent development (**Figure 12**). The WRF global model reproduced every major vortex and circulation at 850 hPa level not only over the North Atlantic but also over the neighboring basins, including the eastern Pacific, South Atlantic, and subpolar regions. But the forecasted track of Wilma shifted to the east from the actual best track when Wilma was in its hurricane stage. This error seems to be attributed to the use of a relatively large grid size for the model simulation (about 56 km) to represent the meso‐scale features accurately, while the inner‐core dynamics of the storm actually might have played a more important role in steering its path and determining its intensity.

Planetary‐Scale Low‐Level Circulation and the Unique Development of Hurricane Wilma in 2005 http://dx.doi.org/10.5772/64061 105

**Figure 12.** A successful forecast of Wilma's TCG.

**Figure 11.** A failed forecast of Wilma's TCG.

104 Recent Developments in Tropical Cyclone Dynamics, Prediction, and Detection

By contrast, the second simulation that was initialized at 0000 UTC October 14 successfully simulated the TCG of Wilma and its subsequent development (**Figure 12**). The WRF global model reproduced every major vortex and circulation at 850 hPa level not only over the North Atlantic but also over the neighboring basins, including the eastern Pacific, South Atlantic, and subpolar regions. But the forecasted track of Wilma shifted to the east from the actual best track when Wilma was in its hurricane stage. This error seems to be attributed to the use of a relatively large grid size for the model simulation (about 56 km) to represent the meso‐scale features accurately, while the inner‐core dynamics of the storm actually might have played a

more important role in steering its path and determining its intensity.

The successful forecast of the merger of the subtropical cyclone with the mid‐latitude trough off the east coast of the US around October 16 seems to be the key point that led the subsequent successful forecasts in the unique low‐level, large‐scale circulation development in the case of Wilma (see **Figures 3**–**6** and **12**). This result suggests that the role of mid‐latitude systems in TC activity is not negligible; a result that is incongruent with the current TCG forecasting emphasis solely on tropical atmospheric conditions. The comparison of the unsuccessful and successful forecasts of the TCG and development of Wilma suggests that every major vortex and circulation component at least in the immediately neighboring storm development area is important for TCG progression.

Although the calculation errors in the WRF global model simulation grew significantly after seven days of model integration, it seems that the global WRF can be used for the purpose of operational short‐range TCG forecasting. This result is encouraging, considering the fact that the global WRF model was initialized 42 hours before the simulated TCG and subsequent development of a TC, and yet the forecast will be fairly accurate in one continuous simulation for the 7‐day period. It should be also noted that the 7‐day global WRF model simulation required less than six hours in a Linux cluster computer with 96 cores.
