Historically, the Gulf of Mexico has been considered a primary source of water vapor that influences the weather for much of the United States east of the Rocky Mountains. Although severe thunderstorms and tornadoes occur most frequently during the spring and summer months, the periodic transport of Gulf moisture inland ahead of traveling baroclinic waves can result in significant severe-weather episodes during the cool season.
To gain insight into the short-range skill in forecasting surface synoptic patterns associated with moisture return from the Gulf, operational numerical weather prediction models from the National Meteorological Center were examined. Sea level pressure fields from the Limited-Area Fine-Mesh Model (LFM), Nested Grid Model (NGM), and the aviation (AVN) run of the Global Spectral Model, valid 48 h after initial data time, were evaluated for three cool-season cases that preceded severe local storm outbreaks. The NGM and AVN provided useful guidance in forecasting the onset of return flow along the Gulf coast. There was a slight tendency for these models to be slightly slow in the development of return flow. In contrast the LFM typically overforecasts the occurrence of return flow and tends to “open the Gulf” from west to east too quickly.
Although the low-level synoptic pattern may be forecast correctly, the overall prediction process is hampered by a data void over the Gulf. It is hypothesized that when the return-flow moisture is located over the Gulf, model forecasts of stability and the resultant operational severe local storm forecasts are less skillful compared to situations when the moisture has spread inland already. This hypothesis is tested by examining the performance of the initial second-day (day 2) severe thunderstorm outlook issued by the National Severe Storms Forecast Center during the Gulf of Mexico Experiment (GUFMEX) in early 1988.
It has been found that characteristically different air masses were present along the Gulf coast prior to the issuance of outlooks that accurately predicted the occurrence of severe thunderstorms versus outlooks that did not verify well. Unstable air masses with ample low-level moisture were in place along the coast prior to the issuance of the “good” day 2 outlooks, whereas relatively dry, stable air masses were present before the issuance of “false-alarm” outlooks. In the latter cases, large errors in the NGM 48-h lifted-index predictions were located north of the Gulf coast.