**2. Data and methods**

**1. Introduction**

The importance of large‐scale external circulations in tropical cyclogenesis (TCG) seems irrefutable, based on the numerous persuasive previous studies [1–3]. In a composite analysis of tropical cyclones (TCs) that developed from monsoon troughs over the western North Pacific, Briegel and Frank [2] hypothesized that eastward‐propagating subtropical troughs poleward of the location of TCG support TCG by providing upper‐tropospheric vorticity advection, thereby forcing upper‐level divergence and uplift. Briegel and Frank [2] highlighted that successfully developing TCs had 850 hPa southwesterly surges in addition to the monsoonal easterly winds approximately 48–72 hours prior to TCG, potentially triggering the low‐level convergence and deep uplift necessary for TCG. Using a global numerical weather prediction model for TCG in the western North Pacific, Chan and Kwok [4] derived a similar conclusion to Briegel and Frank [2] regarding the general synoptic‐scale features present before a TCG, specifically the relatively important roles of the low‐level trade winds and the southwesterly low‐level wind surge prior to TCG. Interestingly, however, Briegel and Frank [2] also found that nongenesis cases have upper‐level troughs both to the northwest and to the northeast of the genesis region, which complicates the distinction between TCG‐triggering and non‐TCG‐triggering synoptic settings. Data limitations precluded the specification of the source of the southwesterly surge into the genesis location, other than any preexisting TCs,

Therefore, there are still some uncertainties with regard to synoptic‐scale features as predictors of TCG. The confusion partly arises from a lack of understanding of the circumstances under which the combination of synoptic‐scale features optimizes TCG. Moreover, use of these synoptic‐scale features as TCG predictors in the Atlantic basin can be problematic because of the differences in basin size and landmass‐ocean distribution. While the monsoon trough is the breeding region of most TCs in the western North Pacific basin, there is apparently no

In an analysis of Tropical Storm Arlene (2005), Yoo et al. [5, 6] found that low‐level vortex dynamics advected temporary low‐level westerly winds from the eastern North Pacific into the western Atlantic which, when combined with orographically enhanced low‐level south‐ easterly winds from Central America, promoted TCG in the western Atlantic basin. Yoo [7] also suggested a potential influence of low‐level wind enhancement in North America on several cases of western Atlantic TCG. Yoo [7] noted that when strong positive potential vorticity (PV) anomalies in the form of mid‐latitude troughs occur with strong low‐level convection over a vast region in the middle‐to‐high latitudes of North America, occasionally an alley of a low‐level wind surge develops from the mid‐latitude trough southward toward the western Atlantic, enhancing the large‐scale low‐level vortex of the developing storm. Yoo [7] also suggested that the enhancement of this low‐level wind surge alley was a harbinger of the intensification of Hurricane Cindy and Hurricane Dennis of 2005. To better understand the relationship between large‐scale geophysical features and TCG mechanisms over the western Atlantic, more case studies of TCG in the western Atlantic are warranted to analyze the

which only occur in about 34% of all the genesis cases [2].

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

monsoon trough region in the western Atlantic [4].

characteristics of interactions of such features leading to TCG.

The "best track" data from NHC are used as guidelines of the track and intensity changes of Hurricane Wilma. Large‐scale sea surface temperature (SST) patterns are described using the National Oceanic and Atmospheric Administration (NOAA) optimum interpolation (OI) ¼ degree daily SST V2 data, which include in situ SST measurements from ships and buoys, satellite observations from Advanced Microwave Scanning Radiometer‐Earth Observing System (AMSR‐E) sensor on National Aeronautics and Space Administration (NASA's) Aqua satellite and NOAA 17/NOAA 18 Advanced Very High Resolution Radiometer (AVHRR), and National Centers for Environmental Prediction (NCEP) sea ice data [9].

The NCEP/National Center for Atmospheric Research (NCAR) reanalysis (NNR) data [10] are used to show the large‐scale wind pattern during the development of Hurricane Wilma until it reached its peak maximum intensity over the Caribbean Sea by 1200 UTC October 19. The period of decreased intensity after moving from the Yucatan Peninsula to Florida will not be described in this study because the large‐scale conditions that created the early TCG stage of Wilma are the main interest. The NNR dataset includes pressure‐level variables in 17 vertical layers on a global 2.5° × 2.5° grid. Hennon and Hobgood [11] noted that NNR data are superior to the European Centre for Medium Range Weather Forecasts (ECMWF) reanalysis dataset in the tropics. Daily and long‐term mean interpolated outgoing longwave radiation (OLR) data provided by the NOAA/OAR/ESRL PSD, Boulder, Colorado, USA, from their website at http:// www.esrl.noaa.gov/psd/ are used to produce the daily OLR anomalies to represent strong surface convection [12].

The Weather Research and Forecasting (WRF) model is executed for Hurricane Wilma with a global model domain setting to reproduce hourly hemispheric‐to‐meso‐scale circulations during its TCG and intensification period with improved spatial and temporal resolution over general circulation models (GCMs). The global WRF model domain was set with (*x*, *y*) dimensions of 721 × 361 (d01) (**Figure 1**). The global domain has a grid size of 55.58874 km in the *x* and *y* directions. The model microphysical schemes are configured following the NCAR Advanced Hurricane WRF (AHW) microphysics guidelines, including (i) Lin et al. cloud microphysics scheme [13]; (ii) the Rapid Radiative Transfer Model (RRTM) scheme for longwave radiation [14]; (iii) Dudhia scheme for shortwave radiation [15]; (iv) the Yonsei University planetary boundary layer (PBL) parameterization; (v) the Monin‐Obukhov scheme for the surface layer option; (vi) the thermal diffusion scheme for the land surface physics; and (vii) Kain‐Fritsch (new Eta) scheme for the cumulus parameterization [16]. The 38 sigma (*σ*) level set of [17] was applied. The model "top" is defined at 50 hPa. The model run was set to update daily SST every 6 hours into the model integration. The daily "real‐time global" (RTG) SST data were interpolated sequentially to produce 6 hourly input data for the WRF run.

**Figure 1.** The model domain of global WRF for Hurricane Wilma in 2005. NHC best tracks were plotted with major intensity changes annotated.

In the global WRF model simulation, two different runtimes were executed. The first set of simulations runs from 0000 UTC October 7 to 0000 UTC October 21, to include the anomalous circulation that Hurricane Vince introduced in the North Atlantic about 1 week prior to Wilma's TCG—a duration of 336 hours. The second set of the simulations runs from 0000 UTC October 14 to 0000 21 October UTC, which only includes the pregenesis condition, TCG, and the development of Wilma—a duration of 168 hours. Both simulations end on 0000 UTC October 21, though Wilma continued to maintain hurricane intensity until 1800 UTC October 25. The latter half of the first model simulation result is compared to the result of the second run to evaluate the accuracy of the global WRF model.

To run the WRF model, 6 hourly NCEP Global Forecast System (GFS) final (FNL) operational global analysis data (1° × 1°) and daily RTG SST data are used for three‐dimensional input data and for SST update (available from National Weather Service at ftp://ftp.polar.ncep.noaa.gov/ pub/history/sst/), respectively. The model run is set to update SST every 6 hours into the model integration. The FNL data for this study are obtained from the Research Data Archive (RDA), which is maintained by the Computational and Information Systems Laboratory (CISL) at NCAR. The original data are available from the RDA (http://dss.ucar.edu) in dataset number ds083.2.
