This is the third seasonal hurricane forecast issued by the Independent Weather Information Center. The seasonal forecasts released prior to June of 2003 and 2004 both demonstrated a high degree of skill. However, no long-term forecast has ever been absolute, and a higher degree of accuracy is desired this season. This forecast is an unofficial one, and therefore we are not liable for one’s actions based upon the information being presented. Please read our disclaimer.
The data and research used to establish this independent forecast was obtained during the months of October to May. The primary parameters that are known to affect the frequency and steering of tropical cyclones in the Atlantic Basin have been closely analyzed throughout the offseason. A thorough explanation and forecast of these factors are outlined. In addition, 55 years of climatological evidence have been included to coincide with the anticipated 2005 summer pattern over the northern hemisphere. All of the aforementioned data has been utilized to not only forecast seasonal tropical cyclone activity, but also month-by-month activity, regional activity, and landfall probabilities. It should be noted that offseason developments prior to June 1 or ensuing November 30 will not be included within the verification of this seasonal forecast.
II. El Nino Southern Oscillation
The El Nino Southern Oscillation, commonly referred to as ENSO, is the most important parameter concerning Atlantic Basin tropical cyclogenesis. ENSO is most notably characterized by significant variation of sea surface temperatures anomalies, or SSTAs, in the equatorial Pacific Ocean, though alterations in sea level pressure, trade winds, and convection are also observed. Warm SSTAs are associated with El Nino episodes, whereas La Nina episodes correspond with cool SSTAs. If SSTAs are neither warm nor cool, ENSO is considered neutral.
Once an El Nino event takes hold, strong upper level westerly winds dominate the low latitudes of the Atlantic Basin. As a result, hurricane development from tropical waves and disturbances along the intertropical convergence zone is typically suppressed. In the mid-latitudes, a higher frequency of upper level troughs and other dynamics that support the development of cutoff lows is typical. High latitude tropical and subtropical activity is thus increased during El Nino events. The reverse is true for La Nina. There is only a small difference between the number of named storms in an El Nino versus a La Nina, but far more hurricanes and intense hurricanes are observed in a La Nina or even neutral ENSO conditions than in an El Nino event.
Current Status of ENSO
ENSO has been in a predominantly warm state since the late spring of 2002. During that year, SSTAs in ENSO Region 3.4 peaked in October and November with anomalies exceeding 1.5 degrees Celsius. By spring 2003, the SSTAs began to cool, and a return to the climatological average transpired. Nevertheless, a secondary warming phase occurred in the summer and El Nino returned. Similar fluctuations ensued in 2004, with a slight moderation period in the spring followed by a warming trend during the latter half of the year. In fact, SSTAs rose to 0.9 degrees Celsius from August through December, and El Nino critera was met for a third straight year.
By late December 2004, SSTAs in the equatorial Pacific Ocean began to show signs of moderation. The 30-day Southern Oscillation Index rose to neutral territory in January for the first time since May 2004. The 30-day Southern Oscillation Index, or SOI, is defined by the monthly sea level pressure differential between Tahiti and Darwin, Australia. The SOI is generally a good indicator of ENSO’s status. Persistent negative values correspond with El Nino, the opposite for La Nina. When ENSO is in a neutral state, the SOI tends to fluctuate between negative and positive values more often. ENSO region 3.4 SSTAs decreased from 0.9 to 0.6 degrees Celsius by February. All indices were beginning to suggest that a return to neutral ENSO conditions was imminent within the next few months. That changed in February when a strong pulse of the Madden Julian Oscillation, MJO, moved across the western Pacific Ocean and enhanced parameters needed for tropical cyclogenesis. The South Pacific Basin became extremely conducive for tropical cyclone formation. Super cyclones Nancy, Olaf, and Percy, which were extensively analyzed and discussed here at IWIC, all developed during the MJO pulse. The MJO pulse coincided with one of the most intense South Pacific Basin cyclone seasons in recorded history, which in turn helped to enhance a Westerly Wind Burst, or WWB, that traveled across the Pacific Ocean. As a result, the SOI plummeted to -52 on February 26, the lowest daily measurement recorded since the extreme El Nino in 1997. Such WWBs can enhance warm depth temperature anomalies, or DTAs across the equatorial Pacific Ocean, which then surface off the coast of Peru two to three months later. Fears of a moderate or strong El Nino returning by summer began to run rampant. However, the sudden drop in the SOI was masked by the South Pacific Basin super cyclones, and other signs of El Nino development were not present.
The WWB, which was triggered by the MJO and intense cyclones, did in fact create a Kelvin wave, and warming beneath the surface became apparent by the month of March. Some warming of the ocean’s surface in the Nino regions 1 and 2 within one to two months was expected. By late April and early May, weekly ENSO region 1.2 SSTAs rose from -1.1 to 0.5 degrees Celsius. A continued rise in subsurface temperatures would suggest that El Nino formation is underway, but that is not the case this time around. The warm pool below the ocean surface has all but reached the surface and moderated, and a new WWB just as intense as the one in observed in February is unlikely.
Climatology is one of the best available tools in ENSO prediction. 1950 to 2004 seasonal ENSO region 3.4 SSTAs were utilized to obtain these following statistics. The 2002 through 2005 warm ENSO episode has been one of the longest-lasting since 1950, second only to the 1991 to 1995 episode. This warm biased ENSO period lasted longer than ten previously recorded episodes. The probability of warm equatorial Pacific Ocean SSTAs lasting another three to six months is nine percent based on those statistics. In addition, there have been 15 El Nino events that persisted through the winter months, and exhibited signs of weakening between December and April the following year. Such has been the case so far in 2005. Only three out of those 15 years, 20 percent, experienced El Nino conditions during the climatological peak of hurricane season. Furthermore, the analogous ENSO years for the December to April period, 1970, 1978, 1988, and 1995, all experienced significant ENSO cooling throughout the spring and summer. La Nina conditions were present during the 1970 and 1988 hurricane seasons while 1978 and 1995 were cool biased. Climatology highly argues against a prolonged El Nino episode lasting well into the 2005 hurricane season. There is even a slight chance of cool biased ENSO conditions by the peak, but a return to neutral conditions is much more likely than either an El Nino or La Nina.
ENSO Prediction Models
Coupled and statistical ENSO models have been in use for several years to try to accurately predict the evolution of ENSO up to eight months in advance. Each individual model has a bias to one extent or another, and at certain times of the year. Therefore, it is better to look at the model consensus as a whole, and more importantly, the model consensus trend. This spring, the average of all the models combined is a slight warm bias. Some, such as the POAMA, were recently forecasting a weak to moderate El Nino to develop over the summer. These select models have been trending cooler however, and again, it is the trend that is really important. The rest of the models have been consistent on an overall neutral ENSO, though it is worth noting that the CLIPER, a relatively good statistical model, has trended to show nearly a weak El Nino during the hurricane season.
|ENSO Models||September 2005||December 2005|
|NOAA LINEAR INVERSE||Neutral||Neutral|
2005 Summer ENSO Forecast
There are clear signals that point towards a neutral ENSO state during the 2005 hurricane season. After the WWB, the DTAs have resumed to a slightly cool state and with little signs of warming further west. This trend does not match the spring state of DTAs in years where an El Nino developed later on such as last year. Furthermore, the model consensus and trend similarly points to a late summer ENSO with a ONI value between -0.5 and 0.5 degrees Celsius. Climatological ENSO statistics would also argue against any El Nino, and if anything it would favor a La Nina. Finally, it is worth noting that three the best five global predictors for predicting ENSO a half year to year in advance, courtesy of Seseske 2004, suggest neutral to cool ENSO conditions, whereas the other two support a warm ENSO. Taking all of this information into account, yet another El Nino regeneration appears very unlikely this year. A La Nina is improbable as well, mostly due to the lack of stronger cool DTAs and model support. The currently warm-biased ENSO, residual from the weak El Nino event that recently dissipated, should continue to slowly relax over the summer and fall. Although there is some uncertainty regarding the bias towards the end of the year, spot-on neutral ENSO conditions should prevail during the heart of the 2005 Atlantic Basin hurricane season. This will neither help nor inhibit tropical cyclone formation.
III. Quasi-Biennial Oscillation
The Quasi-Biennial Oscillation, QBO, is a periodic variation in the direction of stratospheric winds across the deep tropics. The two phases, easterly and westerly, generally last from 12 to 16 months, with the easterly phase often having a slightly longer duration. The exact mechanism in which the QBO influences activity is not yet completely understood, but it is suggested that the QBO is responsible for at least some variation in upper level vertical wind shear in the tropical Atlantic Basin.
Aside from the fact that its exact role is a mystery, there has been significant dispute among the amateur and professional meteorological community as to whether the QBO is truly a crucial factor involving Atlantic Basin hurricane activity. Dr. William Gray at Colorado State University cites that the influence of the QBO has not been very noticeable since the multi-decadal upturn of activity, which began in 1995. After further investigation, it has been decided that the QBO will still be considered a significant parameter in this forecast. Climatological research shows that the QBO particularly influences major hurricane frequency in the tropical Atlantic Basin, with a relatively insignificant impact on the total amount of named systems and hurricanes. The westerly phase of the QBO tends to increase the number of major hurricanes while easterly years usually have an average number of intense hurricanes. In the past ten years, the average number of major hurricanes is 3.8 for both the westerly and easterly phases of QBO, respectively. However, two of the five westerly QBO years had a moderate to strong El Nino present. None of the easterly QBO seasons had a noticeable El Nino. If one were to eliminate 1997 and 2002, the new major hurricane average for westerly QBO years is increased to 5.3. Based on this recent ten year research sample, combined with the well exhibited correlation in earlier years, it is deduced that the QBO does affect major hurricane activity in the Atlantic Basin.
Unlike ENSO, the QBO’s alternation between phases is timely, making it quite simple to forecast. Last hurricane season, the westerly phase of the QBO peaked in May, with a value of 13. Since then, it slowly declined and just recently this year has transitioned to the easterly phase. The QBO will continue to fall and peak later this year, with the next transition not occurring until sometime in 2006. Therefore, the QBO is judged to be a suppressing factor for major hurricane activity in the Atlantic Basin this season.
IV. Atlantic Thermohaline Circulation
The Atlantic Thermohaline Circulation, or ATC, is a density-driven circulation in the Atlantic Basin that undergoes cycles on decadal timescales. It has been proven that the ATC influences Atlantic Basin hurricane activity. When the ATC is in its warm or strong cycle, hurricane activity is increased. During the strong cycle of the ATC, the most notable consequence is warmer SSTAs across the tropical Atlantic Basin, which provides more energy for a tropical storm or hurricane to sustain deep convection. Additionally, lower vertical wind shear and sea level pressure anomalies, or SLPAs, are typically observed across the Atlantic Basin with a strong ATC. The exact opposite is true for the weak or cool phase of the ATC, which is inhibiting to tropical cyclone formation. From the mid 1920s through the late 1960s, the ATC was running strong for the most part, and as a consequence, hurricane activity increased during that period. The ATC shifted to its weaker state in the early 1970s, and that cycle lasted through the early 1990s. A sharp decrease in tropical cyclogenesis was noted. Since 1995, the ATC has shifted back to the strong cycle, which explains the significantly above average hurricane seasons that have occurred over the past ten years. However, even during the ATC’s warm phase, it does occasionally fluctuate for several months. These small fluctuations are difficult to forecast in advance, but observing SSTAs in the Mean Development Region over a monthly period is an easy way to monitor such trends.
Since mid 2003, the SSTAs in the eastern Atlantic Basin have for the most part remained much warmer than average. Notable warm SSTAs have been especially noted over the Mean Development Region this past winter and spring, with monthly values reaching as a high as 0.85 degrees Celsius above average. SSTAs have been just as consistently warm around the Azores and Canary Islands, a tell-tale location of the ATC’s mode. In 2002, there were already signs of the weakening ATC by this time of the year, not only in the SSTAs but also the very positive North Atlantic Oscillation that was setting up. Since neither is the case, a cooling scenario similar to what happened that year is highly unlikely. Therefore, a continued strong ATC and associated warm SSTAs is anticipated throughout the remainder of the year, acting as a robust enhancing factor for tropical cyclone formation, particularly that in the Mean Development Region.
V. North Atlantic Oscillation
The North Atlantic Oscillation, or NAO, is the fluctuation in 500 millibar heights in the northern Atlantic, primarily between western Europe, Iceland, and Greenland. Despite the fact that the difference in phases has little to no impact on seasonal tropical cyclone frequency, the NAO has been shown to have an important role in determining the general steering pattern across the Atlantic Basin. One problem that has prevented reliable techniques in forecasting the NAO months in advance is that it is extremely variable on a monthly and even weekly basis. It would be foolish for us to try to forecast its evolution through the rest of the year without the proper understanding that most have yet to acquire. However, our 55-year climatology does indicate that winter and spring NAO values may influence the steering that sets up during the hurricane season via lag influences.
The NAO was predominantly positive this past winter, although it dropped very low in February and March, with the monthly mean indices standing at -1.33 and -2.76 respectively. It was less so negative in April and has been lingering closer to normal so far this month. Excluding other steering factors, a positive NAO in the winter will modestly correlate with a stronger ridge over the central Atlantic Basin and a stronger trough offshore the United States east coast during the hurricane season. One of our interesting finds was that out of all the tropical cyclones since 1950 that moved westward through the entire Caribbean Sea without dissipating, only ten percent of them occurred in a season that had a positive NAO the prior winter. If this research is indeed a true correlation and not just a coincidence, it suggests that storms forming east of the Lesser Antilles this season will either dissipate or more likely recurve east of 80 degrees west rather than tracking all the way into Central America or the Gulf of Mexico.
VI. Pacific Decadal Oscillation
The Pacific Decadal Oscillation, hereby referred to as the PDO, is a multi-decadal pattern of high and low pressure systems in the northern Pacific Ocean, sometimes seen as a longer-term version of ENSO. However, unlike ENSO, the PDO index is calculated by spatially averaged monthly SSTAs over the northern portions of the Pacific Ocean, not the equatorial. In terms of its variability, the PDO is most similar to the ATC, given that it too fluxes on a multi-decadal basis between a cool and warm cycle. Also like the ATC, the PDO sometimes temporarily swings for a month or even a few years until returning to its dominant cycle phase. The cycle of the PDO switched to warm around 1977, and began showing signs of reverting back to the negative cool phase in the late 1990s. However, since late 2002, the PDO has been predominantly positive; very much so over the past few months. It is uncertain whether the cool PDO that lasted a few years after 1998 was a deviation from the ongoing warm phase, or the warm PDO in the recent few years is a temporary fluctuation away from the new cool phase. One option is that the PDO is in a transitional stage, meaning that it is in a somewhat lengthy process of switching.
Regardless of the dominant cycle the PDO is in, it has been positive this past winter and spring. Some of our climatological research suggests that the February to March PDO values combined with winter tropical Atlantic Basin SSTAs has a lag correlation on the Atlantic Basin steering and tropical cyclone track patterns. In the 55 years that were sampled, it was noticed that in years when the PDO was positive and the winter Atlantic SSTAs were generally cool, three in every four storms that developed in the Mean Development Region recurved out at sea. In the seasons with the opposite conditions, a cool PDO and warm Atlantic Basin SSTAs, about every other storm recurved. Although the relationship is not tight enough to explain a large amount of the track variance, it is worth taking into consideration. This year, the winter SSTAs in the tropical Atlantic Basin were considerably above average, but the PDO was also positive in February and March. 2005 therefore falls in the middle range where the PDO and the Atlantic Basin SSTAs will have opposing lag connections. At face value, this argues against the extremes of having several low latitude tropical cyclones track into the Gulf of Mexico or all of the storms staying harmlessly at sea. However, to obtain a more clear-cut understanding of the track pattern, other predictors have to be examined.
VII. Geopotential Heights and Precipitation
One way to gain an accurate representation of the steering pattern during the upcoming hurricane season is by observing geopotential heights during the April to May timeframe. Geopotential height refers to the potential energy per unit mass of a body as a result of the earth’s gravitational field, but to explain it easier, low geopotential heights usually correspond to low pressure and vice-versa. The 500-millibar level of the atmosphere is used as it is a fair compromise between pressure anomalies in the upper and lower portions of the troposphere. The geopotential height pattern over the past 60 days has been characterized by low height anomalies over the eastern part of the United States and in the central Atlantic Basin just west of the Azores. High height anomalies, on the other hand, have been noted south of the Canadian Maritimes and just off the Iberian Peninsula. This particular regime is concerning for some, as we will explain a little further down.
Another fair, though more indirect indicator of the mean ridge and trough positions comes from analyzing precipitation totals by state across the United States. Precipitation average departures for spring similarly show correlation with the tropical cyclone track pattern in the succeeding hurricane season. Since March, the eastern seaboard of the United States in general has been predominantly wetter than average. This relates nicely to the overall low geopotential height anomalies observed over the same area. However, areas west of the Mississippi River, particularly Texas and the Great Lakes region, have had below average precipitation in the same time period.
From observing the geopotential heights and precipitation departure data, it can be deduced that frequent trough activity has occurred around the United States eastern seaboard over the past three months at least. The Bermuda High, the dominant steering feature for western Atlantic Basin tropical cyclones, has been centered offshore for the most part. Also, there has been more ridging in the central and southern portions of the United States. Climatologically, it is this type of pattern that favors one or more hurricane strikes along the United States east coast. The stronger than normal Bermuda High that is setting up will push any storm that moves under it westward. However, the frequent troughs along its western periphery will make it difficult for a storm to remain on a westward course into the Gulf of Mexico, therefore leaving the areas from the Keys up through New England more vulnerable for a major strike than anywhere further west. This hypothesis was tested by looking at the spring rainfall and height anomalies for our seven analog years. All but one, 2001, featured a hurricane hit along the United States east coast. Scarily enough, all except 1960 and 2001 also had a tendency for above average precipitation and low geopotential heights in that region. 1960 is even a borderline case since it was more average than either dry or wet across the Mid-Atlantic states. Although this steering pattern will not favor long-tracking storms moving westward into the Gulf of Mexico, tropical cyclones that do form in the Gulf of Mexico may tend to move west due to the ongoing tendency of a stronger ridge over Texas and the central Plains.
VIII. Selected Analog Years
Analog years are a necessity when determining long range forecasts. Today’s available weather tools, such as long range forecast models, can only make predictions with a fair amount of skill so many weeks and months in advance. A decent-sized dataset of climate history, or climatology, provides either reassurance or a reason to reassess the conclusion of what has already been analyzed. After making an assessment of the main factors, a limited number of years, analog years, that best match the forecast summer pattern are selected. More so than anything else, years with a predominantly neutral ENSO, easterly QBO, and warm ATC were selected. The trends with ENSO and ATC, along with the NAO and PDO, were also taken under consideration. The analog years chosen for the 2005 hurricane season are below.
|Analog Year||Named Storms||Hurricanes||Major Hurricanes|
*Recorded named storm amounts prior to the satellite era are suspect. Some may have been missed.
However, all of those years had at least one difference in the pattern than what is setting up this year. The ATC was not nearly as strong in 1960 compared to this year, and it was temporarily weak in the beginning of 1952, 1989, and 2001. A weaker ATC even before the season begins has noticeable implications on the steering pattern and deep Atlantic Basin conditions during August and September. Also, 1989, 1996, and 2001 had a La Nina episode the preceding winter, which is the opposite of the weak El Nino that was present in the beginning of 2005. The El Nino that formed in 1957 did not dissipate until the summer of 1958, which posed a slight lag influence on that season, and warm bias ENSO conditions rebounded during the fall of 2003, helping to weaken tropical cyclone activity in the latter portion of that season. None of the analog years are perfect, but taking everything into account, the top three of the list are 1958, 1996, and 2003.
IX. Activity By Region
The Atlantic Basin has been divided into separate regions so that the amount of tropical cyclone activity in each sector can be analyzed.
Mean Development Region
The Mean Development Region is the area between the Lesser Antilles and the western coast of Africa south of 20 degrees north. This is where tropical waves traverse and often develop into hurricanes either in the region itself, or later on in the Caribbean and western Atlantic Basin, hence its name. The extremely warm SSTAs in the Mean Development Region this spring will set the stage for lower SLPAs and more energy to sustain deep convection. Furthermore, the predominantly negative NAO since February also translates to a weaker Azores High, which in turn reduces shear across the area. Also worth noting is the strength and mean position of the intertropical convergence zone, or ITCZ. The ITCZ is essentially a zonal area of low atmospheric pressure and ascending air located around 10 degrees north during the season, and often spins off tropical cyclones when it is not too close to the equator. Although it has been suggested that an easterly QBO suppresses the ITCZ to some degree, the lack of an El Nino combined with the weaker Azores High and warmer SSTAs should still allow for a more northward-shifted and stronger ITCZ this season.
Taking the above into consideration, an above average level of tropical cyclone activy in the Mean Development Region is likely this year. All of our analog years except 1960 saw at least five tropical systems develop in this area. The problem with 1960 as well as 2001, which did not see any hurricane form in the Mean Development Region, can be best explained by the temporarily relaxed ATC during those summers. This was a strong reason why SLPAs in the deep Atlantic Basin were increased in both years. In 1960 the frequency of Mean Development Region storms was lessened, whereas in 2001 the less favorable conditions put a hamper on the intensity. However, such an ATC relaxation is not the case this year, and therefore 2005 should not be faced with that kind of problem. Five to six named storms are forecasted to originate in the Mean Development Region, with three to four being hurricanes, and one to two of those achieving major hurricane status in the area.
The subtropical area of the Atlantic Basin is generally more favorable than the Mean Development Region in years with El Nino or cool ATC conditions. Neither is the case this year, so activity in the high latitudes will be restricted to at least some extent. Nonetheless, a warm ATC and neutral ENSO combination is climatologically not too unfavorable. The amplified trough and ridge pattern that is expected to prevail during the season also implies better dynamics for cut-off lows to develop, especially early and late in the season. Furthermore, there is always the possibility of a highly amplified tropical wave not developing until it is in close to the Bahamas, an uncommon but not unseen example of tropical-originating development in this region. A simple look at our analog years reveals a spread of diversity in subtropical tropical cyclone activity, but given the trough and ridge setup combined with the neutral ENSO and warm Atlantic Basin SSTAs, the area should be slightly more favorable than usual this year. Three to four named storms, with one to two being hurricanes, are forecasted to form above 20 degrees north in the basin in 2005.
We first begin with the eastern Caribbean Sea, often referred to as the dead zone of the tropical Atlantic Basin. This nickname was earned by strong southwesterly shear often in the area due to the summer and autumn presence of the Tropical Upper Tropospheric Trough, or TUTT. This feature, while almost always there, is stronger in some seasons than others, thus having implications on approaching tropical cyclones. Because of the TUTT, very rarely do tropical cyclones actually form in this region. It has been hypothesized that the easterly phase of the QBO could strengthen the TUTT over the summer. However, SSTAs in the eastern Caribbean Sea have been extremely warmer than normal over the past few months. This will aid the barotropic nature of the atmosphere, which in turn will cancel and rule over any influence that is exhibited by the QBO. Therefore, a slightly weaker than normal TUTT is likely this season. Conditions in the eastern Caribbean Sea will not be completely hostile, but activity will likely be on the light side because of the lack of storms expected to move into the area.
In the western Caribbean Sea, conditions are usually more favorable. The problem is, in 2005 most of the storm activity in this region will probably come from in-situ development, rather than a system moving into it from the eastern Caribbean Sea. This will limit the net total of activity, but the low SLPAs and exceptionally warm SSTAs since winter could assist in early season development. Later in the season, there is some concern for a major hurricane to develop in the western Caribbean Sea, as we will explain further down in the monthly section. All in all, the Caribbean Sea as a whole should be somewhat more active than usual with three to four named storms, two to three hurricanes, and possibly one major hurricane.
Gulf of Mexico
We do not anticipate much activity in the Gulf of Mexico during 2005. As we outlined earlier, the winter and spring statuses of the NAO, PDO, and Atlantic Basin SSTAs do not support the idea of storms moving through the Caribbean Sea and into the Gulf of Mexico due to a stronger trough along the United States east coast. However, data suggests that there will not be many in-situ Gulf of Mexico systems either. One interesting correlation we found this offseason is that when the March through April SLPAs off the Mid-Atlantic states coastline are positive, more activity is generally observed in the Gulf of Mexico the succeeding hurricane season. In fact, after separating the past 55 years into two groups based upon this, and then omitting a few outliers that all had the same unrelated conditions present, the average number of named storms, hurricanes, and major hurricanes were all doubled in the positive Mid-Atlantic SLPA years. A reason for this relation could be that higher SLPAs equate to a stronger ridge in the eastern United States, which helps guide tropical cyclones into the Gulf of Mexico and ventilates ones that develop there. Reversely, negative SLPAs are more indicative of a trough, which may induce a an unfavorable southwesterly flow across the Gulf of Mexico.
In March and April this year, the SLPAs off the Mid-Atlantic states were strongly negative, again emphasizing the idea of fewer storms. The analog years that had negative SLPAs in this area were 1952, 1958, 1996, and 2001, all having no greater than three named storms or one hurricane. Another prominent correlation is with the QBO. Based on 55-year statistics, any given season with a westerly QBO has an 88 percent of having a Gulf of Mexico major hurricane, whereas the chance is only 36.7 percent during an easterly QBO. Sure enough, only one of our analog years, 1960, had a Gulf of Mexico major hurricane, and it also had positive SLPAs off the Mid-Atlantic coastline in March and April. Based on all of this information, only two to three named storms, one hurricane, and no major hurricanes are forecasted in the Gulf of Mexico this season.
X. Local Landfall Data
The theory of being able to forecast which areas of coastline will be directly affected by tropical cyclones months in advance is and should be the long-term goal of seasonal forecasting. In order for such predictions to progress, its boundaries must be tested. Long range forecasting also becomes more accurate with an increase in experience. Therefore, experimental landfall forecasts will be included in IWIC’s seasonal forecast for a third year. Many researchers argue that landfall probabilities cannot be given with accuracy since the synoptic pattern is responsible for the steering of tropical cyclones. However, we argue that synoptic trends during spring along with climatological statistics regarding the trends can result in accurate landfall forecasting. Some of these correlations have worked in the past two years, whereas other have not. This year, we are investigating some new methods, and incorperating the ones that have shown skill in the past two seasons.
Texas and northern Mexico
The western Gulf coast was hit by at least one tropical system in all of our analog years except 1952. Upon further examination of 1952 in comparison to the other analog years, it was noticed that it is our only year that had above average spring rainfall in Texas. The rest of the analog years more or less saw dryness in Texas and other South-central portions of the United States. Since less precipitation is usually a product of stronger high pressure, and stronger high pressure prevents storms from curving poleward, this correlation is fairly logical. A stronger ridge over Texas will tend to steer tropical cyclones in the Gulf of Mexico towards the west.
This spring, precipitation over Texas and the southern Plains has been below normal. Therefore, 2005 is not in the same boat as 1952. Some of our analog years with similar dryness, such as 1989 and 2003, had three named storms, two being hurricanes, hit the western Gulf coast. However, these years also saw more overall activity in the Gulf of Mexico, which should not be the case this year. Taking this and the extent of the dryness into account, the best analog years for this landfall region are 1958, 1960, and 1996. One to two named storms, one possibly being a minimal hurricane, are forecasted to hit the Texas or northern Mexico coastline this season.
Louisiana was hit by only one named storm out of the seven selected analog years. The best three Gulf of Mexico analog years, 1958, 1996, and 2001, witnessed no landfalls, with the exception of Subtropical Storm Allison. Allison was not included within the sample due to the fact that it already made landfall in southeast Texas, and remarkably managed to re-enter the Gulf of Mexico after losing some tropical characteristics. The likelihood of such an event taking shape this season is considered low. The forecasted 2005 synoptic pattern does not favor a tropical cyclone moving northward towards the central Gulf coast without recurving prior to landfall. The Bermuda High positioned off the United States east coast interacting with a weakness over the Mississippi Valley is more likely to steer any central or eastern Gulf of Mexico storm to towards Florida. Any storm in the western Gulf of Mexico, on the other hand, will likely make landfall further west under the influence of a prominent ridge in the United States central Plains. Therefore, Louisiana should avoid being hit by any named storms this year.
Mississippi through Florida Panhandle
This region was hit by a total of three named storms and one hurricane out of the selected analog years. The only hurricane to make landfall was Ethel in 1960, and 1960 does not seem to fit the overall pattern that has been surmised. If the top three Gulf of Mexico analog years are used, the panhandle was hit by one named storm two out of three years. As explained in the previous section, a longwave trough over the southeast and northern Gulf of Mexico would likely recurve any central or eastern Gulf of Mexico tropical cyclone towards Florida. It is hence interesting to note that every analog storm originated in either the central or eastern Gulf of Mexico. 1960 is the only analog year that supports the possibility of a big hurricane threatening the western side of Florida from the Caribbean Sea or Bahaman area, but that was also a favored season for high Gulf of Mexico activity, unlike this year. If this area does witness tropical cyclone activity, then no more than one tropical storm landfall is supported by our statistics.
Western Florida Peninsula
After witnessing borderline Category Five Hurricane Charley in 2004, it will be interesting to see how the 2005 hurricane season pans out for the Gulf Coast side of the Florida peninsula. There has been a lot of hype about the possibility of this year setting up in a similar fashion as did the 2004 season. The expected positioning of the main steering parameters is enough to express at least some concern. As you will read under the section referring to Cuba, a track similar to that of Hurricane Michelle in 2001 cannot be ruled out this season. Remember, only a couple hundred mile variation to the north could have meant disaster for southern Florida. Luckily, the trough over the United States east coast was enough to steer Michelle east of the peninsula.
There are several examples in our analog years of tropical cyclones originating in the northwest Caribbean Sea. Nearly all of these cyclones turned northeast, but they passed just to the south of the Florida Keys and hit central and western Cuba instead. On the other hand, while Hurricane Donna in 1960 was not exactly a Caribbean Sea hurricane, it did not curve until 80 degrees west, which is right over the Florida peninsula. Such a path can never be ruled out, but none of the remaining data suggests that this type of track is likely. Finally, we expect the mean positioning of the United States east coast trough to be far enough south to recurve storms like Charley before entering the southeast Gulf of Mexico.
Any named storm that develops in the central or eastern Gulf of Mexico very well could curve to the east or northeast, however. Fortunately, most Gulf originating storms struggle to begin rapid deepening cycles, and any Florida hit from the west would likely be from a tropical storm. The less activity in the Gulf of Mexico in the first place would argue against not only the region being a target to multiple hits, but also anything intense. One tropical storm is forecast to hit the western side of the Florida peninsula this season.
Eastern Florida Peninsula and Keys
The eastern Floridian coastline was hit hard by two significant hurricanes last season, Frances and Jeanne. From a climatological point of view, this region does not look to be a target this year. The only tropical cyclone in our analog years that hit Florida from the east was major Hurricane Donna. However, Donna appears to be an outlier, and the expected synoptic pattern does not support such a track. Further reasoning for less than average probabilities for this region is simply the fact that the setup may not allow hurricanes to enter the Gulf of Mexico. Nine out of the 11 hurricanes that have hit the eastern Florida coast since 1950 were in years with positive SLPAs off the Mid-Atlantic in March to April. Ten of the same 11 hurricanes, 90.9 percent, were also in years when the preceding winter NAO was negative. Both of these favor a more westward displaced Bermuda High, and both are not the case in 2005. Thus, once a tropical cyclone is in the Bahamas, a more northerly track out to sea or further up coast is more likely this season. No landfalls are anticipated along the eastern Florida peninsula, though there will probably be at least one close call. This expectation might as well be viewed as a hit to err on the side of caution, because only a small break in the overall progged pattern could result in another troublesome season for Florida.
It should also be mentioned that the small section of Georgia’s coastline just above Florida has not been directly hit by a hurricane in over 100 years. This leaves us with a limited climatological sample of lanfalls for the region, and thus it is too difficult to provide landfall forecasts for Georgia. However, just because it has not been hit in a long time does not mean it will never be hit again.
Early indicators suggest that the Carolinas will be extremely susceptible to hurricane landfalls this season. As noted in the pattern discussion, six out of seven analog years featured at least one hurricane landfall, the outlier being 2001. Five out of the same seven analog years had abnormally wet conditions in and around the Carolinas; the two exceptions were 1960 and 2001. The flaws found in 1960 have already been outlined in different sections of this seasonal forecast, so our attention is drawn to 2001. That year, the Carolinas were drier than average. In addition, upper level ridging along the Eastern Seaboard was present more than any other year. The remaining five analog years had already shown indications of a longwave trough by April to May along the eastern states. Greater trough frequency became even more prevalent during their respective hurricane seasons.
Unfortunately, similarities between 2005 and the analog years with Carolina hurricane hits have become apparent. A detailed description of this year’s pattern has already been explained under prior sections, but the following points are highly crucial for this region in particular. This year, two abnormally positive areas of high pressure in the mid-levels have developed: one over the Iberian Peninsula and the other just south of the Canadian Maritimes. Both anomalies could have been predicted based upon the five Carolina landfall analog years. Second, there has been a minor weakness between these positive anomalies near 30 degrees west and 50 degrees north, a feature also noted in the exact same area within the landfall analog years’ geopotential height average. Finally, a weakness in the pattern has been in place over the eastern United states the majority of the spring. The positioning of the trough is the most important key. All of the landfall analog years had an eastern United States trough in one form or another.
The evolution towards the synoptic pattern required for a busy Carolinas season is already underway. Further evidence is seen with the strong ATC combined and the lack of an El Nino. One or two named storms, with one being a significant hurricane, are expected to hit the Carolinas this season. There is a chance that these numbers may be slightly conservative. 1996, which had three named storm and two hurricane landfalls, including Fran, was one of the five landfall analog years.
Mid-Atlantic States and New England
The progressive trough over the eastern United States combined with a stronger than normal subtropical ridge to the east may allow a few recurving tropical cyclones to come rather close to this portion of the coast. However, the ridge does not appear to be strong enough to guide any tropical systems due north into the northeast or New England states. Concerns for an in-situ hurricane similar to Belle or Bob are low this year, as these types of systems are more likely to occur in years with an El Nino or weak ATC, both which enhance dynamics needed to develop tropical cyclones from lingering fronts. Therefore, it appears that if any tropical cyclone is going to impact the United States east coast from Virginia to Maine, it would likely come from a recurving system that already made landfall in the Carolinas.
There is a good chance that the Canadian Maritimes will feel the effects from at least one tropical cyclone this season. 1952 and 1960 were the only two analog seasons in which no tropical cyclone made it as far north as the Canadian Maritimes, and both had close calls. This inference fits with the forecasted track pattern of more recurvatures but some getting close and hitting the United States east coast. Many storms that recurve just east of the United States are guided into Canada before becoming extratropical. Some tropical cyclones that form in the subtropical latitudes can move into Canada as well, such as Karen and Juan, interestingly both in two different analog years. Although a Canadian hurricane to the magnitude of Juan is extremely rare, the Canadian Maritimes are likely to experience one named storm with tropical characteristics this year.
The Bahamas have the potential to be hit from two different regions this season. First, a few tropical cyclones are expected to be directed towards the east coast of the United States, particularly the Carolinas. Such tracks often place the Bahamas under the first bull’s-eye depending on how much longitude is gained prior to the beginning of recurvature. Second, the potential threat of recurving hurricanes from the southwest Caribbean Sea in the latter half of the season must also be taken into account. With this in mind, the Bahamas have a good chance of experiencing two hurricanes this year, one possibly being major. The Bahamas are often overlooked at as a landfall region, but in 2005, the islands certainly stand out as one of the main target areas.
Out of the seven selected analog years, Cuba was hit by a total of six named storms, four hurricanes, and three intense hurricanes. The island was not directly hit by any tropical cyclones in 1960, 1989, and 2003. In order to obtain more consistent statistics, each analog year had to be broken down. While no significant problems could be found with 1960, it is interesting to note that Havana barely escaped the inner core of major Hurricane Donna, which passed just to the north and hit southern Florida. Second, the two remaining years that did not have any landfalls, 1989 and 2003, experienced more mid to upper level ridging over the northwest Caribbean Sea and Florida Straits and less troughing over the southeast United States than other analog years. An even share of tropical cyclone activity was located in the Gulf of Mexico, and a more persistent trough allowed most of the Mean Development Region originating systems to pass to the north.
If 2005’s spring synoptic pattern is any indication, and our research has led us to that conclusion, then 1952, 1958, 1996, and 2001 are the favored analog seasons for this region in particular. These years suggest that a mid to upper level trough will generally hang around the Eastern Seaboard for the peak of the hurricane season. Meanwhile, the favored positioning of the subtropical ridge is expected to be somewhere near 30 degrees north and 55 degrees west. The positioning of these two parameters are highly crucial. If the overall pattern does setup as forecast, then a decent southwest flow stretching from the northwest Caribbean Sea northeastward into the Bahamas will develop. Such a pattern would be a dangeroues one for Cuba, especially towards the second half of the season. All four landfalls occurred from hurricanes that originated in the southwest Caribbean Sea in October, a region that is expected to be highly favorable this season due to the lack of an El Nino and a strong ATC. Two out of those four systems were major hurricanes. One or two named storms, and one major hurricane is forecast to directly hit the island of Cuba.
Hispaniola, Puerto Rico, and Lesser Antilles
Hispaniola, Puerto Rico, and the Lesser Antilles are among the more vulnerable areas in any given hurricane season, and climatological evidence tends to point that they are especially so this year. However, there is great disparity in how hard the region was hit in our analog years. 1960 and 1989 had major Hurricanes Donna and Hugo respectively. 1958 had Hurricane Ella hit Hispaniola, and 1996 also saw Bertha and Hortense. The rest of the analog years, on the other hand, only saw tropical storm hits. The high amount of named storms we expect developing in the Mean Development Region, more so than many of our analog years due to the stronger ATC, rationally increases the chance of a significant hit in this area. More intense ridging on the east side of the strong trough along the United States east coast suggests a greater risk to the eastern Caribbean islands as well. At least two named storms should pass through these islands, with at least one being a hurricane. A major hurricane hit is a decent possibility taking the above into account.
Eastern Yucatan Peninsula and Central America
This large region was completely spared a hit during the monstrous 2004 season. In 2005, the threat of a storm tracking through the entire Caribbean Sea and into Central America is fairly low, though one in July or August cannot be ruled out. Any western Caribbean Sea storms that develop in October or November will be prone to move poleward due to a stronger trough, putting Cuba at a greater risk during that time of the season. Therefore, virtually the only way Central America or the Eastern Yucatan Peninsula will be hit this year is if a storm develops in the western Caribbean Sea during the early or middle season. Tropical cyclones in the area in July through September will be more likely to be pushed westward into this region, with the dominion of the United States east coast trough not setting in so far south until October. Considering that the western Caribbean Sea should be favorable for such storm development during the season, this type of storm is possible. One named storm, perhaps a hurricane, is forecasted to hit the Eastern Yucatan Peninsula and Central America.
XII: Monthly Breakdown
The Atlantic Basin hurricane season officially spans six months, beginning on June 1 and ending on November 30. Unlike landfall and regional forecasting, it is very hard to venture out and predict tropical cyclone activity for each month or pair of months by observing the recent patterns alone. Therefore, we looked primarily at climatology and data from analog years for this section of the forecast.
June and July
Generally speaking, the first two months of the Atlantic Basin hurricane season are relatively quiet. On average, one to two named storms form in these months combined. The amount of activity that occurs in June and July does not have any bearing on how active the remaining four months or the season altogether will be. In 1997, the tally stood at five named storms by August, and yet only three others developed after that. On the contrary, the first named storm in 2004 did not form until August, and it turned out to be one of the most intense seasons on record.
Regardless, climatological evidence points towards above average activity during June and July this season. Every single one of our analog years had its first named storm in June or, in the case of 1952, earlier. This is consistent with our 55-year data showing that years with an easterly QBO have a 73 percent chance of a June or pre-June storm, versus a 48 percent chance in westerly QBO years. While the statistics raise eyebrows, it is difficult to come up with a logical reason for this relationship, as is often the case when the QBO is involved. The QBO exhibits a similar, though less robust, correlation with July activity. Years with a neutral ENSO similarly tend to have more activity during June and July than when either La Nina or El Nino conditions are present. A possible explanation for this is that in an El Nino year, other factors aside, the late spring and early summer pattern tend to feature an abnormally strong and low jet stream. The strong zonal flow associated with this is not favorable for early tropical cyclone development. In a La Nina summer, there are usually much less frontal features dipping down into the tropics in the first place. This is a problem as well, as many early season storms are spawned in part by remnant frontal boundaries. The most conducive ENSO scenario is neutral. Fronts are not completely hindered, but at the same time the longwave flow is still amplified enough to allow for cyclogenisis in the tropics off remnant frontal tail-ends. What may be an even more favorable setup is not only a neutral ENSO during the summer, but one succeeding a previous warm ENSO winter. This was the case in three strong ATC seasons: 1966, 1995, and 2003, which featured five, five, and four named storms before August respectively.
This year, the QBO is easterly and the ENSO should stay primarily neutral. Furthermore, there was a weak El Nino this previous winter. All of this is statistically promising for above average levels of activity in June and July, and with a strong ATC, 2005 fits in the list of three years above. A named storm is forecasted to develop sometime in June. Two to three more are expected to form in July, with one or two of those likely being hurricanes.
August and September
Atlantic Basin tropical cyclone activity tends to experience a fairly sharp increase during the month of August, particularly after August 15. August and September are by far the most active months of the season, with the climatological peak of being in early to mid September. In almost all cases, an above average hurricane season equates to above average activity in these two months.
A hyperactive hurricane season is expected. Thus, an active combined August and September is logically probable. Upon investigating previous hurricane seasons, we noticed two separate groups in the years with high activity in these months. One group contains years that featured a general range of six to eight named storms in August and September, which is above normal, but not overly so. On the other hand, there are other years that had over ten named storms in the same timeframe, a rather explosive amount of systems for such a duration. This poses the question of exactly how active the peak two months will be this season. We further examined the eight years since 1950 with ten or more named storms and interestingly enough discovered that seven of them, 87.5 percent, had no more than one tropical storm prior to August. In fact, only one year, 1995, saw its first named storm before July 20. But the overwhelming majority of the seasons with extreme levels of systems in August and September got off to very slow starts. Although we cannot perfectly explain the outlier, the overall strong trend observed here does make sense. One of the main roles of a tropical cyclone is to transport heat from the tropics to the polar latitudes. More oceanic heat is still preserved by the time August arrives in years with little June and July activity. In those kinds of seasons, a higher number of storms are able to develop during August and September, provided the main atmospheric factors are favorable.
As we went into detail about in the above sub-section, an above average amount of tropical cyclone activity is expected before August. Assuming that is true, then it is unlikely that more than eight named storms will develop during the middle two months of the season. Considering that a lot of this activity will probably come from the Mean Development Region, a fairly high percentage of the tropical storms that form should strengthen into hurricanes. Major hurricane activity will no doubt occur, though with QBO being in its easterly phase, we can expect it to be suppressed at least to some degree. Six to eight named storms, four to five hurricanes, and three major hurricanes are forecasted for August and September, with September being the most active of the two.
October and November
When October approaches, a noticeable decline in Atlantic Basin tropical cyclone activity is observed. On average, there are two to three named storms in October and November, which is far less than what is typically the case in August and September but slightly more than the first two months. However, in uncommon cases, there can be more activity during these months than around the peak, such as what was observed in 2001. On the other hand, even the most active seasons like 2004 sometimes almost completely shut down after September.
Our research indicates an above average level of storms during the latter portion of the season. A simple look at our analog years was not enough to come up with a logical prediction due to the surprisingly wide range of tropical cyclone frequency in October and November. The first trend that was noticed upon examining all of the most recent 55 seasons was that the stronger the ATC, the more activity observed in the last two months. This is not surprising considering a strong ATC is favorable for Atlantic Basin activity as a whole. The relation with ENSO is a bit more interesting. When a moderate or strong El Nino is in place, activity during the last two months is almost always reduced below average due to the stronger subtropical jet stream and associated westerly shear over the basin. Conversely, and to no amazement, more intense tropical cyclones are observed late in the season when a La Nina is present. However, it is found that seasons with neutral or even warm biased ENSO tend to have a higher amount of storms than what is seen in a La Nina, especially when coupled with a strong ATC. The most logical reason for this is similar to what was outlined in the June to July sub-section. While a La Nina favors the stronger low-latitude systems, there are generally fewer cut-off lows in the subtropics, which often transfer to warm-core tropical cyclones in the late season.
Since a neutral ENSO and a continuation of the warm ATC are both in the forecast, we expect four to five named storms to form after September, with two to three of those strengthening into hurricanes. Whether one of those is a major hurricane or not depends largely on how warm the ATC is and any ENSO bias. It is tough to tell for sure whether the neutral ENSO will lean more towards warmer or cooler SSTAs by autumn, but a very strong ATC is almost a certainty. Furthermore, 71 percent of all the Caribbean Sea major hurricanes after September occurred when the March through April SLPAs off the Mid-Atlantic states were negative, as was the case this year. Based on this, the 2005 season has a decent shot of having a Caribbean Sea major hurricane during October and November.
XIII: Summary and Conclusion
The recent upswing in tropical cyclone activity in the Atlantic Basin is showing no signs of ending. A strong ATC combined with neutral ENSO conditions should allow the above normal trend in tropical cyclogenesis to continue into the 2005 season. There are some notable characteristics that stand out in our preseason data. There will be a higher than average amount of hurricanes originating in the Mean Development Region, but this year the synoptic steering pattern favors the majority of them recurving at sea. One or two should impact the eastern Caribbean islands. At least one is likely to make it to the United States as a significant hurricane, most likely the Carolinas rather than Florida or anywhere along the Gulf coast. The Gulf of Mexico itself should be devoid of many storm originations as well, therefore being more tranquil than previous years. The first and final thirds of the season are forecasted to be more active than usual, but the peak will also almost surely be less intense than 2004. The western Caribbean Sea is a region to watch in particular during the early and latter portions of the season. During October or November, Cuba faces a strong chance of having a major hurricane from this region, provided ENSO does not become too warm-biased. Finally, we present our forecasted number of systems below.
IWIC 2005 Atlantic Basin Hurricane Season Forecast
|Parameter||2005 Forecast||Long Term Average|
Regardless of our exact expectations for 2005, one in a hurricane prone area should always be prepared for a landfall well in advance. Even if for some reason 2005 turns out to be inactive, it only takes one intense landfalling hurricane to make the season a devastating one. A well-known example of this is the 1992 hurricane season. This was a well below average year by almost all activity measures, yet it was the year of Category Five Hurricane Andrew, which slammed into southern Florida and resulted in over 30 billion dollars of damage.
This seasonal forecast will not be updated during the season, though smaller updates will be posted on our site if necessary. We hope to write and publish a verification of our predictions contained in this forecast when the hurricane season is finished. A preliminary outlook for the 2006 Atlantic Basin hurricane season may also be posted sometime during November.