Adapted from Preprints, 18th AMS Conference on Severe Local Storms, San Francisco CA, 19-23 Feb 1996


Roger Edwards and Steven J. Weiss

Storm Prediction Center

Norman, OK


A number of studies have examined cool season return flow from the Gulf of Mexico, some directly addressing operational forecasting problems. Severe weather forecasting in much of the U.S. east of the Rockies, especially during fall, winter, and early spring, depends on the ability of forecasters to assess the thermodynamic character of low level return flow from the Gulf. Historically, this process has been hindered by a lack of direct observations over the Gulf -- which adversely affects both numerical model guidance and real-time human diagnostics -- and deficiencies in model physics (Janish and Lyons 1992). Weiss (1992) examined short-range numerical model performance in return flow episodes when severe weather was forecast and/or occurred, and found that prediction errors in the model thermodynamic fields were directly related to the accuracy of Day-2 severe thunderstorm outlooks issued by the SEvere Local Storms Unit (SELS) at the National Severe Storms Forecast Center (NSSFC).

Some recent studies have related sea-surface temperature (SST) to return-flow moisture. In this context, the idealized equilibrium state between ocean and convective boundary layer (Betts and Ridgway 1989) is especially relevant. Merrill (1987) determined equilibrium dew points for near-surface air based on SST. Merrill (1992) also described the usefulness of equilibrium assumptions in many synoptic-scale events that involve rapidly modified return flows. Thompson (1992) showed that calculations of equilibrium theta-e, based on surface trajectories over waters with weak SST gradients like the "open" Gulf and Caribbean Sea, are reliable predictors of low level return flow theta-e. Thompson et al. (1994) found that warm SST anomalies maintained high theta-e in a return flow air mass leading to a major severe weather outbreak, in November 1988. In that event, increased boundary layer theta-e contributed to instability and an increase in potential for severe thunderstorms.

To our knowledge, no published paper has attempted to directly compare Gulf SST anomalies with severe thunderstorm incidence, or to statistically document correlations between them implicit in much of the return-flow research. In this study, hypothetical correlations are examined between monthly anomalies in Gulf of Mexico SST and quantitative occurrence of severe thunderstorms over the southern United States, during the cool season months of November through April.


Gulf SST anomalies were obtained for 67 cool season months during the period from November 1981, the first year of available anomalies, through December 1992 (U.S. Dept. of Commerce 1981-1992). SST anomalies were gridded by latitude and longitude, with each grid square measuring one degree on a side, over the Gulf. For additional comparisons, a smaller grid represented the "open" Gulf -- away from northern Gulf continental shelf waters that are most thermally modified by post-frontal boundary layer air masses. Hsu (1992) found that an SST gradient of roughly 7 deg C exists in the middle of the cool season from the northwest Gulf shelf to the abyss, due to strong evaporation and latent heat fluxes associated with frontal passages. We similarly used the 200-m isobath as an approximate shelf break across the central and northern Gulf (north of 24 degrees latitude), assigning SST anomaly grids that straddled this isobath to either the "shelf" or "open" category based on which side had greater spatial coverage in each grid square.

Figure 1. Sample map of gridded monthly Gulf of Mexico SST anomalies from November 1988, in degrees C. Grid points north of the heavy black line represent "shelf" waters. The west (east) Gulf grid points are west (east) of the 90 degree longitude line.

Whole-Gulf and open-Gulf SST anomaly grids were each further subdivided into eastern and western sections, arbitrarily separated by the 90-degree west longitude line. [A representation of the data grids and subdivisions is found in Fig. 1.] Differences in return-flow air mass modification between the eastern and western Gulf are possible due to the presence and orientation of a warm southerly current, called the Loop Current, over the eastern open Gulf. Molinari (1987) examined the relationship of sensible and latent heat fluxes in and near the Loop Current to air mass modification, which varies according to the direction of air flow and the position of the Current. In addition, Lewis and Crisp (1992) showed that changes in model trajectories of return flow air masses over various parts of the Gulf alter the thermodynamic character of return flow along the Gulf coast.

Severe thunderstorm report totals were tabulated and mapped (as in Fig. 2) using the SVRPLOT software (Hart 1993) from the NSSFC database for the U.S. roughly south of 40 degrees north latitude, between 80 and 103 degrees west longitude, for each of the 67 months. Mean report totals were computed for each month across years (i.e., November, 1981-1992).

Figure 2. SVRPLOT map of severe thunderstorm reports for November 1988, representing the geographic domain for monthly report totals. Triangles and paths are tornadoes; dots are hail reports; and crosses are severe gust or wind reports.


We hypothesized that warmer (colder) than normal SST might indirectly lead to higher (lower) incidence of severe weather reports. This is based on the concept of equilibrium thermodynamics and its contribution to return flow instability. To assess this hypothesis, contingency tables of SST and severe event anomalies were built; and the relationships between the variables were examined. Examples of these tables are shown in Figs. 3a-b.

Figure 3a. Contingency table for the western open Gulf, showing number of months (out of 67 total) in which positive and negative SST anomalies corresponded to above- and below-average numbers of severe weather reports.

Figure 3b. Contingency table for the whole Gulf, similar to Fig. 3a.

For the whole Gulf and the various sub-grids used, most months with negative (cold) SST anomalies had below average severe thunderstorm reports. For months when severe reports were above average, a substantial majority exhibited warm SST anomalies. Both findings agreed with our initial hypotheses. However, most months with positive SST anomaly actually yielded below average severe reports, contrary to the hypothesized correlation. In fact, severe reports were below average for roughly 80% of the months in the study period, regardless of the SST anomalies. This apparent imbalance is attributed primarily to the presence of a small number of months containing significant severe storm outbreaks.

Also, notable graphical similarities between trends of severe reports and SST anomalies were identified for specific months across years. The most pronounced example was April (Fig. 4), where upward and downward trends in SST anomaly from year to year corresponded to like trends in severe thunderstorm occurrence, except for 1987-88. Little difference was noted between east- and west- Gulf comparative trends for any of the months. However, for the whole Gulf, and to somewhat lesser extent all subdivisions, upward (downward) trends in the other months' SST anomalies also generally corresponded with upward (downward) trends in severe reports.

Figure 4. Graphical comparison of trends in whole-Gulf SST anomalies (solid) and severe thunderstorm reports (dashed) for April, 1982-1992.

Relationships between SST anomalies and severe reports were examined for months showing the most well-defined SST anomalies. For the ten months of the data set showing the largest whole-Gulf SST positive anomalies, severe thunderstorm reports were 118% of monthly average for the 12-year period. For the ten months with the largest whole-Gulf SST negative anomalies, severe thunderstorm reports were 69% of monthly average. Therefore, it appears that extremes in warm SST anomalies correspond to above average numbers of severe thunderstorm reports; and vice versa.

Finally, SST anomalies were also examined for the ten months showing the greatest positive and negative deviations from average of severe thunderstorm reports. For the ten most active severe weather months versus mean, SST anomalies averaged +0.426 deg C over the whole Gulf, and +0.552 degrees C over the "open" Gulf. For the ten months with the least severe weather versus mean, SST anomalies averaged +0.162 deg C over the whole Gulf, and +0.376 deg C over the "open" Gulf. Although both samples exhibited warm SST anomalies, the larger anomalies were associated with the most active severe weather months.


Over the sampled period, there were few strong correlations found between positive monthly SST anomalies and increased occurrence of severe thunderstorms except in the extreme months. However, strong negative correlations were observed. Several possible explanations for these patterns are offered for further study, especially regarding links to observed synoptic-scale weather features:

1) Cold SST anomalies are representative of more frequent deep frontal penetrations over the Gulf, which lower the SST and adversely interrupt the return flow cycle. Thus, low level instability and resultant severe thunderstorm potential are greatly diminished.

2) However, warm SST anomalies in any given month suggest a lack of strong baroclinic waves over the southern U.S. responsible for the deep frontal penetrations over the Gulf.

Without the passage of such waves, anomalously high SST could be "wasted" with relatively little or no severe thunderstorm production. So, high SST by itself does not necessarily foretell an unusually active cool season for severe storms; but instead may contribute to enhanced severe potential in a few individual events. By contrast, however, low SST would more reliably suggest relatively inactive severe storm periods.

Although the sign of the anomaly was often unrelated to the numbers of severe reports, there were notable similarities between trends of SST anomalies and trends of reports. In other words, there was a correlation between increases (decreases) in SST anomalies and increases (decreases) in severe weather, when comparing the same month from year to year. This finding further supports a link between SST anomaly and severe thunderstorm incidence; however, its long-range predictive implications await reliable numerical prediction of SST on a seasonal time scale.

There could be an accessory relationship between monthly severe thunderstorm incidence and SST anomalies one month prior, which is a planned offshoot of this study.

The set of ten months with the strongest positive deviations of severe reports from mean had over two and a half times the mean whole-Gulf SST anomaly than the set of ten months with the lowest reports totals versus mean. Also, severe thunderstorm reports were above (below) average for the set of ten months with the warmest (coldest) SST anomalies, for both whole Gulf and "open" Gulf. These findings suggest a positive relationship between SST anomaly and severe thunderstorms for extreme months of either variable.

A goal of this work is to further familiarize forecasters with the importance of ocean-air interaction to low level return flow from the Gulf, and its relationship to severe thunderstorm potential. Our study supports the growing operational philosophy that forecasters concerned with the central and eastern U.S. should closely examine Gulf of Mexico SST and the process of air mass modification during return flow cycles. This will help to more thoroughly evaluate the nature of the returning air and better anticipate the potential for severe thunderstorms in the one to three day forecast frame.

To aid in this forecasting approach, SST data from the Gulf is available at several places on the World Wide Web, as are SST anomalies and climatologies. Climatological SST data is also available from National Climatic Data Center.


The authors thank: Jim Henderson, Bob Johns, John Lewis, Fred Ostby, Rich Thompson, and Mike Vescio for their helpful insight and suggestions; also Jenifer Clark and Tom Ross for helping to procure SST anomaly data.


Betts, A.K., and W. Ridgway, 1989: Climatic equilibrium of the atmospheric convective boundary layer over a tropical ocean. J. Atmos. Sci., 46, 2621-2641.

Hart, J.A., 1993: SVRPLOT: A new method of accessing and manipulating the NSSFC severe weather database. Preprints, 17th Conf. Severe Local Storms, St. Louis, 40-41.

Hsu, S.A., 1992: Effects of surface baroclinicity on frontal overrunning along the central Gulf coast. J. Appl. Meteor., 31, 900-907.

Janish, P.R., and S.W. Lyons, 1992: NGM performance during cold-air oubreaks and periods of return flow over the Gulf of Mexico with emphasis on moisture-field evolution. J. Appl. Meteor., 31, 995-1017.

Lewis, J.M., and C.A. Crisp, 1992: Return flow in the Gulf of Mexico. Part II: Variability in return flow thermodynamics inferred from trajectories over the Gulf. J. Appl. Meteor., 31, 882-898.

Merrill, R.T., 1987: Equilibrium dewpoint temperature over the Gulf of Mexico in March. Technical paper presented at GUFMEX planning meeting, National Severe Storms Laboratory, Norman, OK, 4 pp.

_____, 1992: Synoptic analysis of the GUFMEX return flow event of 10-12 March 1988. J. Appl. Meteor., 31, 849-867.

Molinari, R.L., 1987: Air mass modification over the eastern Gulf of Mexico as a function of surface wind fields and Loop Current position. Mon. Wea. Rev., 115, 646-652.

Thompson, R.L., 1992: Equilibrium theta-e over the Gulf of Mexico and its relationship to severe thunderstorms in the cool season. M.S. thesis, School of Meteorology, University of Oklahoma, 119 pp.

_____, J.M. Lewis, and R.A. Maddox, 1994: Autumnal return of tropical air to the Gulf of Mexico's coastal plain. Wea. Forecasting, 9, 348-360.

U.S. Dept. of Commerce/National Oceanic and Atmospheric Administration, 1981-1992: Oceanographic Monthly Summary, Nov.-Apr., 22.

Weiss, S.J., 1992: Some aspects of forecasting severe thunderstorms during cool-season return-flow episodes. J. Appl. Meteor., 31, 964-981.