2003 Extended Range Atlantic Hurricane Season Forecast
Written and published by Kevin Budd, Rob Mann, and Jason Moreland
As of May 25, 2003
I. Introduction
This forecast is based on months of independent tropical cyclone research studies. Over the offseason, we have discovered several climatological patterns that relate to the upcoming season, which we will discuss in this forecast. Additionally, with the data that is available to use over the Internet, we have evaluated the many known factors that are crucial to North Atlantic tropical cyclone variability. Finally, we have examined previous seasons with the atmospheric conditions most similar to what we expect to occur this year, to enhance our evaluation confidence.
It should be emphasized that all three of us are amateurs, not professionals. By no means is this forecast an official one. Thus, we advise you to not use this forecast for life and death situations. We will not be responsible for any tragic consequences that may occur from the information given here.
II. El Nino Southern Oscillation
An El Nino event is characterized by anomalous changes across the equatorial Pacific Ocean. These alterations include warmer sea surface temperature anomalies, increased cloud cover, negative values of the Southern Oscillation Index, and weaker trade winds. A La Nina event causes just the opposite. In an El Nino, the anomolous heating of sea surface temperatures near Peru causes convection to develop over ther eastern Pacific which leads to increasing 200 millibar vertical shear over the Atlantic basin. Subsidence also increases. This was the case last season. El Nino was one of the many factors that inhibited tropical cyclone activity. The 2002 El Nino episode was weak to moderate in intensity.
The 2002 El Nino event peaked in mid November. Sea surface temperature anomolies in the equatorial Pacific Ocean rose between 1.5 and 2.0 degrees celsius. The Southern Oscillation Index dipped down to -14 in January.
Since then, the El Nino Southern Oscillation has become neutral, and the 2002 El Nino episode has come to an end. All indications are showing that the lingering pockets of above average anomalies in the Pacific will continue to diminish in overall coverage.
Southern Oscillation Index
Negative values of the Southern Oscillation Index often indicate El Nino conditions while positive values of the oscillation often indicate La Nina conditions. Fluctuations in the Southern Oscillation Index occur because of the large changes in the Walker circulation, which is closely linked to the pattern of tropical Pacific sea temperatures. The 90 day Southern Oscillation Index average has risen from -14 back in January to -5. The 30 day index average is now slightly above average. This is typical once an El Nino has decayed.
Subsurface Temperatures
Subsurface temperatures are water temperatures below the surface of the ocean. Upcoming changes in sea surface temperature anomalies can often be first spotted below the surface. This is why watching subsurface temperatures in the Pacific is so important while making an El Nino Southern Oscillation forecast. Since December, the layer of warm anomolies below the surface has continued to diminish. In fact, a cool tongue of subsurface temperature anomolies is now building across the Pacific under the surface. The recent cooling under the surface may very well be the first indications of the return of La Nina.
Sea Surface Temperature Anomalies
The cooling of subsurface temperatures is now beginning to show up at the surface. Sea surface temperature anomalies in Nino regions 3 and 4 have steadily lowered from 1.4 degrees celsius to 0.2 degrees celsius. Temperature anomalies in Nino regions 1 and 2 have already cooled below normal. Negative anomolies range between -0.5 ad -1.5 degrees.
Long range El Nino Southern Oscillation forecast models usually have a difficult time forecasting sea surface temperature anomolies between the months of March-June. However, we still use these model forecasts as guidance. Over the past few months, the majority of long range El Nino Southern Oscillation models have continued to forecast near-normal conditions during the peak of the 2003 hurricane season. 7 out of 11 models are forecasting neutral conditions, and the other 4 are forecasting cool conditions. It should be noted that none of the models currently forecast warm El Nino conditions during the second half of the year.
2003 El Nino Southern Oscillation Forecast
We have seen a return towards neutral El Nino Southern Oscillation conditions recently. Both sea surface and subsurface temperatures have cooled to normal, and even slightly below normal levels in sections of the eastern Pacific. We expect this long term trend to continue throughout the remainder of the year. The 2003 season will start off under neutral conditions. As the season progresses, the sea surface temperature anomalies will trend to more negative values, and by the peak of the 2003 season, a weak La Nina should be present. As the 2003 season comes to an end, we expect weak to moderate La Nina conditions.
It is important to accurately forecast the El Nino Southern Oscillation in order to accurately forecast the amount of Atlantic tropical cyclone activity this hurricane season. The El Nino Southern Oscillation is the most crucial seasonal factors for the Atlantic basin.
III. Sea Surface Temperature Anomalies
Sea surface temperature anomalies are important when it comes to hurricane development. Warm temperatures provide fuel for hurricanes to grow.
Gulf of Mexico
A weather pattern similar to the pattern we saw in 2000 is expected to take shape in the deep south. A dome of high pressure will likely build over the region for the majority of the summer. This ridge should keep the main storm track well to the north. Precipitation should be limited due to high pressure dominating the area. Therefore, sea surface temperature anomalies in the Gulf of Mexico should be above average.
Caribbean Sea
The Caribbean Sea is the most difficult region to forecast in terms of sea surface temperature anomalies. In the northwest Caribbean, we anticipate slightly above average values. See the Gulf of Mexico section above for more details. Temperature anomalies in the central Caribbean Sea should be near average. In anticipation of a moderate Tropical Upper Tropospheric Trough and a weak to moderate La Nina, sea surface temperature anomalies in the eastern Caribbean are expected to be slightly below average.
Western Atlantic
The two most dominate factors controlling sea surface temperature anomolies in the western Atlantic are the Bermuda High and North Atlantic Oscillation. When there is a positive North Atlantic Oscillation and strong Bermuda High, sea surface temperatures in the area tend to be above average, and vice-versa. Given the current North Atlantic Oscillation trend, we expect it to be in a positive mode this summer. The Bermuda High has been anomalously intense so far this Spring, and with the North Atlantic Oscillation climbing, we expect the high to remain well developed through the hurricane season. Thus, warm sea surface temperature anomalies in the western Atlantic are forecasted.
Mean Development Region (8-20N, 20-60W)
Based on analyses of sea surface temperature anomaly archives, we have found that there is an indirect correlation with temperature anomalies in the Mean Development Region and Nino regions 1 and 2 values. During a negative phase of the El Nino Southern Oscillation, sea surface temperature anomalies in the Mean Development Region tend to be below average. Conversely, when there is an El Nino event, Mean Development Region temperature anomalies are usually above average. The correlation is most evident near the coast of western Africa. As noted in the El Nino Southern Oscillation section of our extended forecast, weak to moderate La Nina conditions are in the forecast. The positive phase of the North Atlantic Oscillation this season will also help cool the region, primarily by strengthening the Azores High. Due to these expectations, we are forecasting below average sea surface temperature anomalies in the Mean Development Region.
IV. Azores High
The Azores High is a crucial factor for tropical cyclone development in the Mean Development Region. When a strong high is observed, Mean Development Region activity tends to be surpressed, as it causes cooler temperatures, more subsidence, and a surpressed equatorial trough. The Azores High was strong last season, making it one of 2002's many inhibitors for low latitude development. A weak Azores High, on the other hand, enhances formation in the Mean Development Region. So far this spring, the high has been running very close to average in strength. We judge the Azores High as a neutral factor for this upcoming Cape Verde season.
V: Bermuda High
The Bermuda High is one of the most crucial factors when forecasting east coast landfalls. A strong Bermuda High can cause tropical storms to move too far west to avoid land before curving out to sea. A weak Bermuda High allows a large trough to move into the east, blocking the coast from dangerous hurricanes. Over the past three years, we have seen a strong trough protecting the east coast. Keep in mind that there is usually less trough activity during La Nina events, which we expect this year. Furthermore, as already mentioned in our sea surface temperature anomaly section, the Bermuda High has been quite strong this past winter and so far this Spring. Thus, a stronger Bermuda High and less trough activity off the United States east coast is in the forecast.
VI: Sea Level Pressure Anomolies
Tropical storms and hurricanes are associated with areas of low pressure. Therefore, a broad area of abnormally below or above normal sea level pressure anomolies may have significant implications on tropical cyclone activity. In fact, Shapiro (1982) suggests that one sixth of tropical cyclone activity is caused by such variations in sea level pressure. Low values tend to correlate with more tropical cyclone activity, and vice-versa.
The western Caribbean Sea has experienced below average sea level pressure over the past couple years. This trend has continued throughout the winter months. Sea level pressure anomolies remained slightly below average, and little change is expected this season.
Sea level pressures in the Gulf of Mexico have varied significantly over the past few months, which is normal for this time of the year. As mentioned in the sea surface temperature forecast, the weather across the Gulf of Mexico and surrounding areas is expected to be dominated by a ridge of high pressure this summer. As a result, we are forecasting average to slightly above average sea level pressure anomolies in the Gulf of Mexico.
Sea level pressure anomolies in the Mean Development Region have been average to slighly above average this year. The anomolies are in response to a normal strength Azores High and El Nino. Although El Nino has now faded, our expectation of a positive North Atlantic Oscillation should result in little change in these anomolies during the season.
VII: Tropical Upper Tropospheric Trough
The Tropical Upper Tropospheric Trough is located in the eastern Caribbean Sea and central Atlantic Ocean. Basically, this is a large upper level trough that can be enhanced by upper level lows in the central Atlantic. A strong tropical trough is not conducive for tropical development. It tends to cause strong levels of vertical wind shear, which act to rip developing tropical storms apart. This factor was a major inhibitor in 2000. During the 2001 it weakened a bit, and in 2002 it became practically nonexistant. The Tropical Upper Tropospheric Trough is likely to return this year. Sea level pressure anomalies have been above average in the eastern Caribbean Sea and central Atlantic. Above average pressure is assocaiated with strong sinking air which enhances baroclinicity. In turn, the trough itself is enhanced. Plus, an easterly phase of the Quasi Biennial Oscillation (see below) also strengthens the Tropical Upper Tropospheric Trough. This year, will it be as strong as it was in 2000? Probably not. But it will be stronger than it has been over the past two seasons. A moderate Tropical Upper Tropospheric Trough is being forecasted.
IX: Stratospheric Quasi Biennial Oscillation Winds
Quasi Biennial Oscillation winds are one of the more important factors in determining the amount of tropical cyclone activity in the Atlantic. These winds come in two phases; easterly and westerly. When they are abnormally easterly, higher shear values across the Mean Development Region of the Atlantic is most evident, with westerly being vice-versa. Hence, Quasi Biennial Oscillation winds tend to have more of an effect on tropical cyclone intensity, rather than frequency. These winds mostly influence the number of major hurricanes, rather than the number of hurricanes and tropical storms.
On average, both Quasi Biennial Oscillation phases last about year. During the 2002 season, Quasi Biennial Oscillation winds were westerly, an enhancing factor for major hurricane development. However, we had an average number of major hurricanes. A westerly phase does not always mean that there will be an above average number of major hurricanes, it just increases the likelyhood. A number of other factors besides these winds inhibited major hurricane development last season.
During the 2003 season, Quasi Biennial Oscillation winds will be in an easterly phase, thus a small inhibiting factor on major low latitude hurricane activity.
X: Other Factors
There are several other factors that affect tropical cyclone activity in the Atlantic basin. But there are just too many to mention. There are two factors that are well-known, but will not be used in our 2003 seasonal forecast numbers.
African Sahel Rainfall
Dr. William Gray of Colarado State University did find a correlation between African Sahel rainfall and hurricane activity in the mean development region. Periods of above average rainfall in the African Sahel correlated with above abover average hurricane activity and vice versa. However, the factor has failed in recent years. More research still has to be conducted on this possible factor.
Saharan Air Layer
A Saharan Air Layer is made up of an easterly low level wind surge that starts off over the continent of Africa. This sudden surge of wind carried dry Saharan dust along with it. Once over the Atlantic, it causes rapid increases in vertical wind shear and subsidence. An unusually strong Saharan air layer can prevent a Cape Verde season from becoming active. So this factor is crucial. However, the main reason why we do not use this in long range seasonal forecasting is simply because there is no way of knowing when one of these major outbreaks will occur. More research studies still needs to be conducted.
XI: Climatology
We have spent most of our research this offseason on past hurricane seasons, especially the ones that we believe will be similar to the 2003 season. First, we started by eliminating the years that were either El Nino or westerly Quasi Biennial Oscillation wind years. We focused mainly on the years that had an easterly Quasi winds and a weak La Nina. We have quite a few numbers for you to ponder.
Active Gulf of Mexico Years Followed by Slower Gulf Years 1950-2002
1909-1910: 6 and 3
1947-1948: 6 and 2
1959-1960: 6 and 3
1970-1971: 6 and 2
1995-1996: 6 and 2
1998-1999: 6 and 2 2002-2003: 6 and ???
The average number of storms is 2.3
On average, there is 38% the level of Gulf activity in a season followed by an active Gulf year.
Comparison Between El Nino Southern Oscillation & Regional Development
Year
La Nina Phase
# storms W of 60W
# storms B/T 40W and 60W
# storms E of 40W
1984
Mod.
3
2
1
1985
Mod.
2
0
1
1988
Strong
0
4
1
1999
Mod.
1
3
2
1.5 named storms develop east of 40W during moderate to strong La Nina years.
2.2 named storms develop between 40 and 60W during moderate to strong La Nina years.
1.5 named storms develop west of 60W during moderate to strong La Nina years.
3.2 named storms develop or move into the Gulf of Mexico during moderate to strong La Nina years.
2.0 named storms develop in the Gulf of Mexico during moderate to strong La Nina years.
Year
La Nina Phase
# storms W of 60W
# storms B/T 40W and 60W
# storms E of 40W
1989
Weak
0
2
5
1995
Weak
2
5
6
1996
Weak
1
1
5
1998
Weak
0
2
5
2000
Weak
1
4
3
4.8 named storms develop east of 40W in neutral to weak La Nina years.
2.8 named storms develop between 40 and 60W during neutral to weak La Nina years.
0.8 named storms develop west of 60W during neutral to weak La Nina years.
2.7 named storms develop or move in the Gulf of Mexico during neutral to weak La Nina years.
2.2 named storms develop in the Gulf of Mexico during neutral to weak La Nina years.
Analog Years for the 2003 Season
The purpose of selecting analog years is to find an average amount of activity and United States landfall activity in years with with weak La Nina conditions and an easterly Quasi Biennial Oscillation.
The analog years we chose were 1952, 1954, 1984, 1988, 1989, 1996, 1998, 2000, and 2001.
Average number of named storms: 12
Average number of hurricanes: 8
Average number of major hurricanes: 3
Strange, there are a few analogs here which are different than Gray's. Yet, we average the same number of named storms as in Gray's December 2002 forecast.
Average total United States landfalls: 3
Average East coast landfalls: 2
Average Gulf coast landfalls: 1
Next, years that were following an El Nino with westerly Quasi winds were selected. United States landfall totals and overall activity was averaged.
The years we chose were 1952, 1954, and 1988.
Average number of named storms: 10
Average number of hurricanes: 6
Average number of major hurricanes: 3
Average total United States landfalls: 3
Average East coast landfalls: 1
Average Gulf coast landfalls: 2
Note, 1952 and 1954 had one East coast brush each.
After some controversy, we have decided to consider 1964 a special analog year.
Named Storms: 12
Hurricanes: 6
Major Hurricanes: 6
1964 must be considered an analog because 1963 had a weak El Nino and a westerly Quasi Biennial Oscillation. 1964 did have westerly winds too, but unusually weak; very close to a 0 value. 1964 also had a weak La Nina phase developing late in the summer.
Climatology Conclusion
This climatological research presents some rather suspicous results. For all of the years that were averaged, there was no more than two tropical cyclone landfalls along the Gulf of Mexico and east coast. When we equate the several lesser factors, the best analogs become 1952, 1954, 1964, 1988, and 1998.
XII: Regional Highlights
Gulf of Mexico
We are forecasting 2 to 3 named storms to either originate or move into the Gulf of Mexico. Based on climatology, the chances of having 3 or less Gulf storms in a season following an active Gulf season is good. Other factors point towards a rather quiet Gulf of Mexico this year. One of those factors is the North Atlantic Oscillation. Typically, when the North Atlantic Oscillation is in its negative phase, the Gulf of Mexico experiences more tropical cyclone activity. This was the case in 2002. Conversely, when the oscillation is positive, tropical cyclone activity shifts away from the Gulf of Mexico. This should be the case this upcoming season. Climatologically speaking, the easterly phase of the Quasi-Biennial Oscillation does not support a major landfall along the Gulf coast either. Does this mean that there is absolutely no chance of a major landfalling hurricane in the Gulf this year? Of course not.
Western Caribbean Sea
The western Caribbean Sea should be generally favorable. We expect warm sea surface temperature anomalies and slightly low sea level pressure anomalies to be in place in 2003. Add in La Nina and boom, ideal conditions for a major hurricane. It should be noted that a lot of our best analog years featured a late season Caribbean major hurricane (Fox 1952, Joan 1988, Mitch 1998, Michelle 2001, just to name a few). While we cannot say with 100 percent confidence that one will occur this year, the odds are in favor.
Western Atlantic
With a positive North Atlantic Oscillation, thus a strong Bermuda High and very warm sea surface temperature anomalies, we expect quite a lot of storms to develop or deepen in this region. Interestingly, a lot of our strongest analog years had modest Mean Development Region activity (as we forecast this year), but a lot of rapid intensification and development once far to the west. Can we expect a similar pattern this year? Yes. One factor that might be a small problem is the Tropical Upper Tropospheric Trough. While we do not expect it to be as strong as it was in 2000, it should temporarily help to prevent strengthening of several storms that move towards the Bahamas and east coast. However, we do not expect scenarios liken to 2000, where several systems we ripped to shreds before approaching the Bahamas. The reason is simply because the tropical trough should not be nearly as strong as it was that year.
Mean Development Region (8-20N 20-60W)
Conditions will favor a normal to slightly above normal Cape Verde season this year. The reason why we do not expect the Mean Development Region to be dangerously active is due to the expectations of cooler sea surface temperature anomalies and a neutral Azores High. However, one major factor that will favor tropical cyclone development in this area is La Nina, which is the reason we are still calling for far more tropical cyclone activity in the region than in 2002. All in all, we are forecasting 4 to 5 named storms to form in the Mean Development Region; most of these forming further west at that.
Subtropical North Atlantic
Over the past two hurricane seasons, we have seen a lot of subtropical storm activity. Subtropical and frontal storm activity was enhanced last year by the El Nino, whereas the very warm waters in the northern portion of the Atlantic basin fueled subtropical development in 2001. This season, we will have La Nina, which is not conducive for subtropical storms. However, the sea surface temperature anomalies in the high Atlantic have been persistently warm so far this year, which we judge to be an enhancing subtropical factor. Furthermore, a moderate Tropical Upper Troposheric Trough helps clean the baraclonic environment, which should also aid in subtropical development. We are forecasting 3 to 4 high-latitude forming systems this season, as opposed to 6 as observed in 2001 and 2002.
XIII: United States Landfall Highlights
One of the things we decided to do differently this season, is instead of forecasting landfall probabilities, we should provide actual numbers. Our landfall forecasts are a lot more detailed than others, so do expect errors. These forecasts are based mostly upon our 2003 El Nino Southern Oscillation forecast, long range forecasts of the North Atlantic Oscillation and precipitation totals between January and May.
Texas
Forecasting Texas landfalls this season will be difficult. Based on La Nina alone, the chances of a Texas landfall are decent. Rainfall data also suggests that there is a decent chance of a landfall along the Texast coastline. However, the North Atlantic Oscillation says otherwise. When this is positive, the chances of a Texas landfall decrease dramatically. So we have decided to go with the North Atlantic Oscillation this season. Only one tropical storm is forecasted to hit Texas this year, near Brownsville. There is also a small chance for a tropical cyclone landfall between Lake Charles and Houston.
Louisiana
After experiencing a rough 2002 season, Louisiana may be in for a break this season. Conditions were prime for a direct landfall. Based on the North Atlantic Oscillation and La Nina, a direct landfall is a lot less likely this year, especially for southeast Louisiana. If Louisiana were to get hit this season, it would likely occur near Lake Charles.
Mississippi, Alabama, and Florida Panhandle
This section of the Gulf Coast is usually quiet during positive North Atlantic Oscillation years. No landfalling tropical storms are expected from Biloxi, Mississippi through Pensacola, Florida. The landfalling tropical storm or hurricane probability increases slightly between Panama City and Tallahassee. Thus, one landfall along the Florida Panhandle is being forecasted.
Florida Peninsula
This section of our lanfalling forecast covers all of Florida except for the Florida panhandle. There has been a lot of talk about Tampa being due for a major landfalling hurricane. Being due for a hurricane does not mean much when a La Nina event and positive phase of the North Atlantic Oscillation does not favor a Tampa landfall.
Southern Florida may be in for a long season. Earlier, we mentioned how tropical waves exiting the coast may not begin to develop until they move farther west this season. This would make it easier for systems to stay on a more westerly track under the Bermuda High. One of these tropical systems may continue heading westward through southern Florida and eventually into the Gulf of Mexico. The western Atlantic, including the Bahamas, should be one of, if not the, most favorable ares for tropical cyclone development in 2003. With that being said, most southern Florida landfalls from the east tend to occur during neutral or even warm El Nino Southern Oscillation events, rather than cool. So while there should be several close calls, it appears unlikely that southern Florida will get hit from the east. From the western Caribbean; that is a different story. Years with positive North Atlantic Oscillation and La Nina often feature a Cuba-crossing Caribbean storm, many which make a direct hit on Florida. In either case, this year we are forecasting one storm to strike or brush by southern Florida.
A landfall near the Jacksonville area does not appear any greater this season than it would be during any other season. Although a positive North Atlantic Oscillation can curve storms westward into northeastern Florida, this has only happened once in the past 100 years. Dora of 1964 (season with a similar pattern to this one) was the last storm to strike northeastern Florida. However, due to the overwhelming odds that climatology presents against such a landfall, no northeastern Florida landfalls are being forecasted for this season.
Georgia
A landfalling tropical storm in Georgia does not happen too often. Georgia usually experiences a direct landfall every 33 years. The coast often experiences a brush every three to four years. A direct landfall is not expected this season. The slight probability of a landfall exists more during El Nino years than La Nina years. However, there is still a good probability of another brush.
North Carolina and South Carolina
The Carolinas may be the biggest target of the 2003 season. The main landfalling factors for the east coast all favor a potentially dangerous season. The pattern during the 2002 season was just the opposite. During the 2002 season, the North Atlantic Oscillation was negative overall with building El Nino conditions in the eastern Pacific. This favored a quiet season along the east coast with most of the activity shifting farther west into the Gulf of Mexico. This will not be the case this season. This year, we will likely be facing weak to moderate La Nina conditions along with a neutral to positive North Atlantic Oscillation. These conditions have not been present since 1999, the year of Dennis, Floyd, and Irene. Furthermore, an easterly phase of the Quasi-Biennial Oscillation and a stronger Azores and Bermuda High are also conducive for storms to track into or close to the Carolinas. Based on all of the above, 2 tropical systems are being forecasted to strike or brush the Carolinas this season.
Mid Atlantic coastline and New England
A quiet season is likely from Norfolk, Virgina north to Trenton New Jersey. Landfall probabilities for this area increases when both El Nino Southern Oscillation and the North Atlantic Oscillation are neutral, and both are expected to be in weak to moderate phases.
Believe it or not, the probabilities for a landfalling tropical storm increases between New York and Cape Cod. In the past, when the two major oscillations are in the states that they should be this year, New England ends up getting struck by at least one system. Based on the current pattern, either a brush or a secondary landfall from one of the east coast storms is anticipated to occur in southern New England.
XIV: Summary and Conclusion
To sum it up, here is what we expect to be in place this upcoming season.
Warm sea surface temperature anomalies in the Gulf of Mexico, western Atlantic, and western Caribbean Sea
Cool sea surface temperature anomalies in the eastern Caribbean Sea and Mean Development Region
High sea level pressure anomalies in the Gulf of Mexico and Mean Development Region
Low sea level pressure anomalies in the western Caribbean Sea
Thus,
Slightly to moderately above average activity overall
Little amount of tropical cyclone activity in the Gulf of Mexico
Normal amount of tropical cyclone activity in the Mean Development Region
High amount of tropical cyclone intensification in the western Atlantic
Good chance of a late season major hurricane in the western Caribbean Sea
Landfall target likely being the United States east coast, especially North Carolina
2003 Atlantic Hurricane Season Forecast
Named Storms (NS)
13
Hurricanes (H)
7
Major Hurricanes (MH)
3
The above numbers are modestly above the long term 1950-2000 climatological average (approximately 10 named storms, 6 hurricanes, and 2 major hurricanes).
XV: Schedule of Upcoming Seasonal Assessments
This forecast is our first for the 2003 season, and will be the final. A verification of this forecast will be released around the end of the 2003 season, followed soon by the first preliminary outlook on the 2004 season.
