Data obtained throughout the past several months indicates that 2007 Atlantic Basin hurricane activity will be somewhat above the long-term average, with an estimated 13 named storms, 8 hurricanes, and 3 major hurricanes. An abnormally high number of tropical cyclones will form east of the Lesser Antilles during the bimonthly period of August and September, with at least one significant hurricane striking the northeast Caribbean. The Yucatán Peninsula also faces a high risk of a major hurricane strike during the final third of the season. No tropical cyclone landfalls are expected along the United States mainland.
This is the fifth Atlantic Basin hurricane season forecast issued by the Independent Weather Information Center. The primary parameters that are known to affect the frequency and steering of tropical cyclones in the Atlantic Basin have been closely analyzed throughout this off-season. A thorough explanation and forecast of these factors are outlined. Analogous pattern and parameter combinations derived from 57 years of historical climate data have also been examined to further justify the anticipated synoptic setup over the northern hemisphere. The aforementioned methodology has been utilized to forecast not only seasonal, but also monthly, regional, and landfalling tropical cyclone activity.
While an overwhelming consensus of members within the meteorological community dismisses the idea of providing accurate landfall forecasts months in advance, it is the long-term goal of IWIC to prove that it is in fact possible. Countless hours of research and resulting statistical correlations combined with theory have led the authors to believe, in convincing fashion, that several recent findings are of significance. Bare in mind this forecast is experimental and unofficial, hence IWIC is not liable for one’s actions taken based on the information being presented. Please read the disclaimer.
Several verification guidelines are strictly followed. First, any subtropical or tropical cyclone formation before June 1 or beyond November 30 will not be included within the verification of this seasonal forecast. Subtropical Storm Andrea has not been included in the forecast totals. Second, classified depressions are not featured in any aspect of the seasonal verification. Third, a depression that develops in a particular region, but moves into a secondary region before becoming a named storm will be verified as a named storm formation in the region in which it officially becomes a named storm. Fourth, a depression that develops in a particular month, but does not strengthen into a named storm until the following month will be verified as a named storm formation in the month in which it officially attains a name. Finally, any named storm that strikes the Gulf (Atlantic) coast of the Florida peninsula and re-emerges into Atlantic (Gulf) waters will not be considered a landfall along the east (west) side of Florida. This guideline also applies for western and eastern coastal areas of the Yucatán Peninsula.
2. 2006 Methodology Modification and Forecast Verification
Although there was moderate success in 2003 and 2004, neither the landfall nor total tropical cyclone activity forecasts verified well in 2005. It would have been completely imprudent to apply the same methodology to forthcoming forecasts after it displayed no skill in 2005. Upon an intense post-review, it remains accepted that several, pre-2006, hypothesized correlations were flawed due to misrepresentative statistics. Thus, many of the original forecasting methods were abandoned and a new methodology was introduced in the 2006 Atlantic Basin hurricane season forecast.
Confidence in the new methodology increased leading up to the 2006 season despite the recognized failures of 2005 and forecast discontinuity between IWIC and other respected groups and agencies. The majority of members within the media and meteorological community began to warn that hurricane season 2006 would be hyperactive, and feature significant landfalling tropical cyclones. On the other hand, IWIC did not foresee a season of such hyperactivity and landfalls. Fortunately, the much advertised season of hyperactivity and destructive landfalls never materialized. Rather, the outcome was quite similar to IWIC’s expectations.
The 2006 total tropical cyclone activity forecast called for 14 named storms, 8 hurricanes, and 4 major hurricanes. In reality, only 10 named storms, 5 hurricanes, and 2 major hurricanes formed. It is somewhat disappointing that the actual totals did not match the projected estimates. However, it should be noted that the three longstanding and highly respected seasonal forecasting agencies had indicated there would be at least as much activity as forecast by IWIC. The following table compares the seasonal forecast totals released in May 2006.
Comparison of May 2006 Atlantic Hurricane Season Forecasts
Regardless, the fact that most agencies were guilty of forecasting more activity than observed is no excuse for IWIC’s forecasting shortfalls. The unexpected development of El Niño conditions in the equatorial Pacific by October was the primary cause of IWIC’s overestimation. If one were to break down the monthly tropical cyclone activity forecasts, it would become evident that 7-10 named storms, 4-5 hurricanes, and 2 major hurricanes were projected to form by October 1. In reality, 10 named storms, 5 hurricanes, and 2 major hurricanes had developed by the end of September. In other words, the monthly observed activity was easily meeting IWIC expectations until equatorial Pacific conditions reached El Niño thresholds in October. Furthermore, it is no surprise the forecast of 5 named storms and 2 hurricanes during the bimonthly period of October-November did not verify. Had the late onset of El Niño conditions been accurately diagnosed in May, no more than 12 named storms would have been forecast for 2006. Additional adjustments to the ENSO forecasting approach have been incorporated into the 2007 seasonal forecast, thus reducing the probability of such flawed ENSO forecasts in the future. In the meantime, it is the opinion of the authors that the 2006 total activity forecast was, in large part, a success. Such skill has been consistently displayedsince the first IWIC forecast in 2003.
With all that said, the success of IWIC landfall forecasts is of greater importance. Thankfully, the 2006 IWIC landfall forecast verified quite well. The following landfall forecasts verified with no question: no landfalling storms along the storm-ravaged north Gulf coast, 1-2 named storms exiting the Caribbean Sea and striking the Florida peninsula, no significant landfalls along the east side of the Florida peninsula, no direct hits along the northeast United States, 1-2 threats to the Canadian Maritimes, minimal storm threats to the Bahamas originating from the Caribbean Sea, 1-2 minimal storm hits on Cuba from the Caribbean Sea, and 1-2 minimal threats to Hispaniola, Puerto Rico, and the Lesser Antilles.
Some of the landfall forecasts did not verify, however. Approximately 2 minimal tropical cyclones were forecast to originate in the southwest Gulf of Mexico and impact Texas and northeast Mexico, but these storms never materialized. This forecast could have been corrected with the additional data acquired since last May. Nonetheless, it is interesting to note that multiple tropical disturbances in the southwest Gulf of Mexico were nearly classified as tropical cyclones by the National Hurricane Center before moving inland over Mexico and Texas. The second incorrect landfall forecast included 1-2 tropical cyclones, one possibly being intense, impacting Central America. It is presumed that the onset of El Niño during the latter half of the 2006 season greatly diminished Central America’s chances of being directly impacted by a tropical cyclone. Finally, no more than a brush was expected in North Carolina, but one tropical storm landfall was observed. This was more of an error in phraseology than methodology, as several weak tropical cyclones directly struck North Carolina in selected analog years. Nevertheless, the primary idea was there would not be any significant hurricane landfalls along the East coast.
Additional skill was displayed in forecasting regional activity. For instance, 4 named storms, 2-3 hurricanes, and a major hurricane were forecast to develop in the Mean Development Region, which is located between the Lesser Antilles and west coast of Africa. Over the course of the season, 4 named storms, 2 hurricanes, and a major hurricane originated from the area. Up to two storms were forecast to, “probably enter or form close to the eastern Caribbean Sea,” hence tropical cyclones Chris and Ernesto. Above normal activity was anticipated in the west Atlantic due to a combination of in-situ developments and recurving storms from other regions, and such was in fact the case. Tropical cyclone activity estimates for the west Caribbean Sea and west Gulf of Mexico did not fare as well, hence the inaccurate landfall forecasts among those regions. If nothing more, one should at least be able to recognize that the forecast unquestionably nailed the overall pattern observed during the 2006 hurricane season.
The verification of the IWIC 2007 seasonal hurricane forecast will be even more promising based on what is hereby assumed to be positive additions to the forecast methodology. The preexistence of an applicable forecast technique to build upon, as opposed to the lack of such in early 2006, was a significant advantage in the construction of the 2007 forecast. Moreover, the latest modifications being implemented in the 2007 forecast are intended to reduce the types of errors observed in 2006, which fortunately were few and far between.
3. Summer-Fall El Niño Southern Oscillation
The El Niño Southern Oscillation (ENSO) is an important parameter concerning Atlantic Basin tropical cyclogenesis. ENSO is most notably characterized by significant variations in sea surface temperature anomalies (SSTAs) in the equatorial Pacific, though alterations in sea level pressure anomalies (SLPAs), trade winds, and convection are also observed. Warm (cool) SSTAs are associated with El Niño (La Niña) episodes, and if SSTAs are neither warm nor cool, ENSO is considered neutral.
Once an El Niño (La Niña) event takes hold, strong (weak) upper-level westerly winds dominate the low latitudes of the Atlantic Basin, thus hindering (enhancing) hurricane development from tropical waves and disturbances along the intertropical convergence zone. Being that El Niño can allow for more baroclinic-induced development in the higher latitudes, there is only a small difference in named storm frequency between El Niño and La Niña episodes. Nonetheless, more hurricanes and intense hurricanes are generally observed in La Niña or even neutral ENSO conditions.
El Niño conditions developed in October 2006. Warm SSTAs peaked in ENSO Region 3.4 on December 6 with a value of +1.4ºC. SSTAs in Region 3.4 have steadily declined since early December and are currently entering negative territory. Furthermore, a large pool of negative SSTAs below the ocean surface of the equatorial Pacific has been present since December. Subsurface cooling is often one of the first precursors to the formation of a La Niña episode. Some mild warming has recently occurred in tandem with a traversing Kelvin wave. This will act to temporarily delay, but not prevent, La Niña development. Another important aspect of the ENSO is the Southern Oscillation Index (SOI), defined by SLP differential between Tahiti and Darwin, Australia. Negative (positive) values correspond to El Niño (La Niña), and as one would expect, daily negative values dominated the SOI during the late formation of El Niño in 2006. The monthly SOI tanked in January 2007 with a value of -7.3. In recent months, both daily and monthly SOI values have become increasingly positive. The April SOI value was -3.0. That reading is slightly higher than the -1.4 value recorded in March, but some fluctuations in the SOI are to be expected. The overall slow trend toward a more positive SOI remains evident.
While the sole use of climatology is ill-advised, it is a vital component of long-range ENSO forecasting methodology. One of the climatological research methods used is the evaluation of past and current values of the Multivariate ENSO Index (MEI). The MEI helps determine the overall condition of ENSO and its level of influence on atmospheric circulations. The index accounts for the SOI, equatorial convection, and other global conditions known to be affected by ENSO. Positive (negative) values of the MEI correspond with El Niño (La Niña) conditions. The MEI recently peaked at +1.3 in November 2006. There was a secondary peak in January 2007, but MEI values have dramatically fallen to near 0 within the last three months. The peak MEI value in late 2006 followed by a steady decrease into neutral territory during the early spring is an important trend to note. The following five MEI analogs were then selected and analyzed: 1963-1964, 1965-1966, 1972-1973, 1987-1988, and 1994-1995. Three of those analogs featured La Niña conditions throughout the following hurricane season. The two exceptions, 1966 and 1995, featured ENSO cool bias and late season La Niña conditions, respectively. None of the hurricane seasons that followed had to contend with El Niño or even warm bias ENSO conditions. Thus, trends within the MEI suggest that there is little to no chance of El Niño this summer, but there is at least an 80 percent probability of La Niña conditions by fall.
A second climatological method used to forecast ENSO is more basic yet equally as effective. The observation of trends in trimonthly values of ENSO region 3.4 SSTAs can benefit ENSO forecasting much like those observed within the MEI. The latest El Niño peaked during the trimonthly period of November-December-January with a 3.4 region SSTA value of +1.1ºC. Soon after, trimonthly values decreased substantially. SSTAs moderated by 0.8ºC between the trimonthly periods of NDJ and JFM. Similar trends were observed in 1963-1964, 1987-1988, and 1994-1995. La Niña conditions were present throughout the 1964 and 1988 hurricane seasons. Similarly, La Niña made a late appearance during the 1995 season.
A third climatological method of ENSO forecasting based on global SLPA patterns is being studied. Additional elaboration of the specifics is not desired at this point in time. Nonetheless, this scheme is much more sophisticated and potentially more valuable to long range ENSO predictions than simple observation of MEI and trimonthly SSTA trends. The method, albeit in the experimental stage, is signaling that La Niña conditions will be established by the trimonthly period of June-July-August at the latest.
Finally, a new technique involving a connection between ENSO and solar activity is in the works. In fact, this scheme may permit accurate ENSO forecasts up to several years lead time. While the details will have to be explained in a future paper, it can be said that the method, combined with the aforementioned data, is indicating that at least cool bias ENSO conditions will form this year and possibly last through the end of 2008.
Little weight was given to the available ENSO model forecasts this year. The model consensus more often than not seems to miscalculate the timing and development of crucial ENSO episodes in the long range. Regardless, all of the available model data suggests that ENSO will either remain neutral or strengthen into La Niña by the summer. It should be noted that CLIPER and GMAOGCM, two of the more reliable ENSO models that correctly predicted the onset of the 2006 El Niño, are predicting cool ENSO conditions by the peak of the hurricane season.
2007 ENSO Forecast
The El Niño episode that developed in late 2006 and persisted through early 2007 has dissipated and will not return this year. Moreover, ENSO is forecast to reach La Niña criteria by September and last through the second half of the Atlantic hurricane season. This solution is supported by the following: recent trends in the SOI, moderating SSTAs over the equatorial Pacific, persistent subsurface cool anomalies, historical MEI analogs, trimonthly ENSO SSTA analogs, two experimental climatology methods, and a consensus among ENSO forecast models. The anticipated La Niña will act to enhance hurricane and major hurricane activity in the tropical Atlantic Basin during the trimonthly period of August-September-October.
4. Summer-Fall Atlantic Multidecadal Oscillation
The Atlantic Multidecadal Oscillation (AMO) is a long-term pattern of SSTA variability in the North Atlantic Ocean. It is theorized that the oscillation is dependent on decadal density-driven changes in the thermohaline circulation. The warm (cool) phase is known to enhance (suppress) hurricane activity. The most notable aspect of the warm (cool) phase is above (below) normal SSTAs across the tropical Atlantic Basin, which provides more (less) energy for tropical cyclones to sustain deep convection. Since 1995, the AMO has been in a predominantly warm cycle, which explains the several significantly above average hurricane seasons that have occurred over the past 12 years. However, the AMO does occasionally fluctuate to the opposing phase of the long-term cycle.
From mid 2003 to this day, SSTAs in the tropical Atlantic Basin have, for the most part, remained well above average. However, a new experimental model that may explain as much as 16 percent of tropical Atlantic Basin SSTA variance suggests that modest cooling will gradually occur throughout the remainder of 2007. In fact, current SSTAs could moderate to near neutral levels during the peak of the hurricane season. While there is confidence in this scheme, additional elaboration of model specifics is not desired at this point in time. Considering summer and fall tropical Atlantic Basin SSTAs will not reach the extreme levels observed over the past 4-5 years, the warm AMO will only moderately enhance tropical cyclone activity in 2007.
5. Other Factors
The combination of La Niña conditions in the equatorial Pacific and a slightly warm AMO certainly suggests that an active Atlantic Basin hurricane season is likely. However, there are several other parameters that must be accounted for to accurately forecast both monthly and total activity as well as formation and track distribution.
Spring North Atlantic Oscillation
The North Atlantic Oscillation (NAO) is an atmospheric pattern over the North Atlantic that is defined by differences in sea level pressure between the Icelandic Low and Azores High. Positive (negative) NAO values denote higher (lower) pressures near the Azores compared to Iceland. Although the NAO is a central figure in the concurrent steering pattern across the Atlantic Basin, it is extremely variable on a monthly and even weekly basis. However, an in-depth reanalysis of the last 57 years indicates that spring NAO values have a lag influence on large-scale atmospheric conditions that evolve over the Atlantic Basin during hurricane season. The importance of the positive NAO observed this spring is discussed in detail in later sections.
Spring North American Pattern
Analysis of the 57 year climatology database indicates that the springtime pattern over the continental United States, northeast Pacific Ocean, and surrounding areas has a lag affect on large-scale atmospheric conditions that evolve over the Atlantic Basin during the hurricane season. As such, the authors have proposed a unique parameter hereby called the Spring North American Pattern (SNA). The proposed four phases of the pattern have been found to influence homegrown tropical cyclone frequency and steering patterns surrounding the southeast United States and Central America. The significance of the Phase 2 SNA pattern observed during much of spring 2007 will be discussed in later sections of the forecast.
Winter El Niño Southern Oscillation
Winter values of the ENSO are vital to several aspects of the seasonal hurricane forecast. First and foremost, winter ENSO episodes are known to have a lag influence on large-scale atmospheric conditions well after they dissipate. Therefore, the winter ENSO signal is given weight when analog springtime patterns are selected. Second, winter ENSO has been found to have a direct influence on tropical cyclone landfall probabilities in certain regions. Finally, it was recently discovered that winter ENSO episodes play a key role in determining early season tropical cyclone frequency when other summertime parameters are less pronounced than usual.
Summer-Fall 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 westerly phase of the QBO peaked last June with a value of +11.47. Afterwards, the QBO steadily declined and has just recently transitioned to the easterly phase. The QBO will continue to fall throughout this year, with the next transition not occurring until sometime in 2008. While the QBO has a moderate affect on tropical cyclone activity in the Mean Development Region, it is otherwise one of the lesser factors considered in this forecast.
Summer-Fall Pacific Decadal Oscillation
The Pacific Decadal Oscillation (PDO) is a multi-decadal pattern of high and low pressure systems in the north Pacific Ocean, sometimes seen as a longer-term version of ENSO. Unlike ENSO, the PDO index is calculated by spatially averaged monthly SSTAs over the northern, as opposed to equatorial, portions of the Pacific Ocean. The PDO switched to a long-term warm cycle in 1977 and aside from a few fluctuations has remained in that phase up to the present time. However, the PDO has been closer to neutral since mid 2006. In fact, a recently discovered link between the PDO and solar activity suggests that it will indeed switch to a long-term cool state later this year as La Niña conditions unfold in the equatorial Pacific. Although the PDO is not known to exhibit a significant influence on year-to-year Atlantic Basin tropical cyclone activity, one study hints that its long-term state influences United States landfall odds.
6. Activity By Region
The Atlantic Basin has been divided into five regions to help portray the anticipated tropical cyclone formation and steering tendencies of 2007.
Mean Development Region (south of 20ºN, east of 60ºW)
The majority of tropical cyclogenesis observed in the Atlantic Basin is initiated by tropical waves that migrate from Africa to the Mean Development Region, hence the name. Several of the waves that trigger cyclogenesis do so within the borders of the region. The remaining percentage of embryonic tropical cyclones form upon entering the Caribbean Sea, west Atlantic, or Gulf of Mexico. Common logic dictates that warm SSTAs and a lack of El Niño conditions would result in abnormally high activity in this area. However, the necessary formula is not that simple. Several seasons have featured warm SSTAs, a lack of El Niño, and average to below average tropical cyclone activity east of the Lesser Antilles. For example, record warm SSTAs and cool-bias ENSO conditions were observed in 2005, but as active as the season was overall, the Mean Development Region escaped with only two weak named storms. This conundrum has been a subject of study post 2005, and sure enough a critical parameter of Mean Development Region activity was discovered: spring values of the NAO.
After analyzing 57 years of data, it has been concluded that positive (negative) spring values of the NAO correspond with enhanced (suppressed) tropical cyclone activity in the Mean Development Region. Hurricane seasons following positive NAO spring months yield 1.0 named storms, 0.4 hurricanes, and 0.3 major hurricanes more than those following negative spring months. The correlation is magnified when AMO values are positive. Warm AMO hurricane seasons that follow positive NAO spring months yield 2.5 named storms, 1.5 hurricanes, and 0.8 major hurricanes more than warm AMO seasons following negative spring months. The lengthy process by which spring values of the NAO affect Mean Development Region tropical cyclone activity will have to be explained in a future paper. In short, however, opposing modes of the spring NAO have conflicting effects on the large-scale atmospheric circulation. Moreover, atmospheric fingerprints of the spring pattern persist over the Mean Development Region during the summer months. Summer variability of the NAO does not override the summer pattern set forth by the spring NAO dipole pattern because the dipole is considerably less intense during summer.
Warm AMO and La Niña conditions are forecast this hurricane season. While warm AMO combined with neutral and especially La Niña conditions typically feature above normal activity in the Mean Development Region, such is never a guarantee unless spring NAO conditions are positive. Spring 2007 did feature a positive NAO. Therefore, it has been concluded that abnormally high tropical cyclone activity will be observed in the region. All three parameters aligned in the same fashion during the 1995 and 1998 hurricane seasons. 1995 and 1998 featured 9 and 7 named storms, 6 and 3 hurricanes, and 1 major hurricane each in the Mean Development Region, respectively. All three measurements of tropical cyclone activity were exceptionally high compared to the climatological average in each of those seasons. 2007 may rival what was observed in the deep tropical Atlantic during 1995 and 1998, but slightly less activity is anticipated overall. The westerly QBO of 1995 and remarkably warm AMO of 1998 are thought to have maximized the favorable conditions set forth by ENSO and the NAO. The current easterly QBO and steadily moderating AMO should place somewhat of a ceiling on total numbers in 2007. Thus, 7-8 named storms, 4-5 hurricanes, and 1-2 major hurricanes are forecast to originate from the Mean Development Region. It should be noted that such numbers are substantially higher than the long-term climatological averages of 3.0 named storms, 1.4 hurricanes, and 0.5 major hurricanes.
Northeast Atlantic (north of 20ºN, east of 65ºW)
Unlike the Mean Development Region, tropical cyclogenesis that occurs in the northeast Atlantic is typically a result of baroclinic forcing as opposed to initiation by tropical waves. However, tropical cyclones that develop in the Mean Development Region often recurve poleward and move into the area. While the northeast Atlantic is by far the largest region covered in this forecast, it also tends to be the most ignored due to the fact that most tropical cyclones that traverse these waters do not impact land.
To assess the amount of tropical cyclones that will enter the northeast Atlantic, the spring mid-level pattern in the general area was analyzed. One noticeable characteristic of the past spring pattern was a large weakness centered near 60ºW. The weakness acted to separate two high pressure systems, one over the continental United States and the other over the northeast Atlantic, respectively. With 7-8 named storms forecast to develop in the Mean Development Region, it is reasonable to suspect that many storms will recurve into the northeast Atlantic along this weakness. In fact, a concentrated recurving corridor may set up along the mean trough position between 55-65ºW. An estimated 5-6 named storms, 4-5 hurricanes, and 1-2 major hurricanes are expected to move into the northeast Atlantic from the Mean Development Region and west Atlantic.
Although there will certainly be numerous tropical cyclones entering the northeast Atlantic, it does not appear that there will be much tropical cyclone development within the region. Several studies were conducted to better understand how various parameters influence development in the northeast Atlantic. The most promising study is primarily based on the winter ENSO. Seasons following both winter La Niña and neutral ENSO conditions yield an average of 1.4 named storms, 0.85 hurricanes, and 0.1 major hurricanes developing in the northeast Atlantic, whereas the average for seasons following an El Niño is 2.3 named storms, 1.5 hurricanes, and 0.2 major hurricanes. Interestingly, years with winter El Niño conditions followed by neutral ENSO conditions average 3.5 named storms, 2.7 hurricanes, and 0.5 major hurricanes forming the northeast Atlantic. On the other hand, years with winter El Niño conditions followed by La Niña conditions average only 1.4 named storms, 0.9 hurricanes, and 0 major hurricanes. Of the latter years, only 1964 and 1998 had more than one named storm formation. Furthermore, two of the tropical cyclones in 1964 were simply delayed Mean Development Region developments. Such developments are less likely in warm AMO and positive NAO years. The physical reasoning behind such correlations is currently being investigated. With that said, 1-2 named storms and 0-1 hurricanes are forecast to develop in the northeast Atlantic.
West Atlantic (north of 20ºN, west of 65ºW)
Although the west Atlantic does not sustain the strongest hurricanes, tropical cyclones in the region often rapidly intensify uncomfortably close to the United States East coast. Furthermore, tropical cyclone formation via initiation of tropical waves or baroclinic forcing are both common in the west Atlantic, as are tropical cyclones moving into the area from the Mean Development Region and northeast Atlantic.
However, the said trigger mechanisms of tropical cyclogenesis are forecast to yield few formations during August-September 2007. Baroclinic-initiated tropical cyclogenesis occurs most frequently during El Niño episodes and cool AMO periods, but neither will be present this season. A high frequency of baroclinic-induced storms has also been noted during several La Niña episodes, but nearly all of them also featured cool AMO conditions. In fact, La Niña-dominated seasons that feature a cool AMO yield 1.1 more named storms than those with a warm AMO due to the increase in baroclinic-initiated development. In addition, the lack of any significant upper air troughing over the region, which was the case last spring, is often a precursor of less cutoff lows in the west Atlantic during the summer. Thus, the probability of development via baroclinic forcing is exceptionally low. Notwithstanding, it is not out of the realm of possibility that a few tropical waves will induce tropical cyclone formation in the portion of the west Atlantic bordered by the Bahamas, Bermuda, and the 65ºW parallel. This may be most likely to occur in July when conditions in the Mean Development Region are still seasonably unfavorable. Any tropical cyclone that does form in this quadrant of the west Atlantic would recurve northeastward due to close proximity of an upper air weakness near 60ºW. With that said, 2-3 named storms and 0-1 hurricanes are forecast to originate from the west Atlantic this season.
The lack of in-situ storms does not mean the west Atlantic will be void of multiple tropical cyclone tracks, however. An abnormally active Mean Development Region combined with an upper air pattern conducive for several long-tracking storms over the central Atlantic signifies an above normal number of tropical cyclones entering the west Atlantic. A large number of recurvatures are forecast along a weakness in geopotential heights near the 65ºW parallel, and half of the recurving cyclones are bound to briefly traverse west Atlantic territory. An estimated 3-4 named storms, 2-3 hurricanes, and 1-2 major hurricanes are expected to enter the region from the Mean Development Region and northeast Atlantic.
Despite the fact that numerous tropical cyclones have passed through the east Caribbean with little complications, in-situ formation is rare due to the summer presence of an upper-level trough. This feature induces strong southwesterly shear, which is unfavorable for tropical cyclogenesis. In-situ formation in the east Caribbean is even less likely in warm AMO, La Niña, and positive spring NAO years. 1995 and 1998 featured all three conditions, yet neither featured a single in-situ formation. The aforementioned regimes greatly enhance tropical activity in the Mean Development Region, thus decreasing the number of delayed, tropical wave-initiated formations that commonly occur west of the 60ºW parallel. In-situ east Caribbean formation has been observed in only 25 percent of past La Niña-dominated seasons. Likewise, only 12.5 percent of seasons with a combination of warm AMO and positive spring NAO conditions featured in-situ formation. Consequently, 2004 was the only year with such a parameter combination that featured an east Caribbean formation, but El Niño conditions were also present. The AMO, NAO, and ENSO will all be in modes that are least conducive for in-situ formation this year. Thus, no tropical cyclones are forecast to originate in this area. However, 1-2 significant tropical cyclones are expected to pass through the far northeast Caribbean Sea given the recently strong mid-level ridging north of the Mean Development Region.
On the other hand, the springtime pattern over North America and the far north Atlantic combined with spring values of the NAO also argues against any Mean Development Region-originating tropical cyclones passing west of 75ºW in the Caribbean. Rather, more northerly tracks into the subtropical Atlantic region are favored due to the presence of a mid-level weakness located near 60ºW. In fact, any tropical cyclone that enters the Caribbean Sea is likely to recurve back into the Atlantic east of Hispaniola. Moreover, the lack of any east Caribbean-originating storms suggests it is unlikely that any tropical cyclones will enter the west Caribbean Sea.
Finally, no storms are expected to originate in the west Caribbean Sea until late September or October. 1995, 1998, and 2000 are the three sampled seasons that featured positive spring NAO, warm AMO, and La Niña conditions. 1995 is the only year that featured an early and late season in-situ formation. Hurricane Allison formed on June 3, 1995, but such an early formation is not likely in 2007 due to the early onset of La Niña (see section 7). Also, 1995 is the only positive spring NAO year to ever produce more than one named storm in the west Caribbean. Excluding 1995, none of the 18 of positive spring NAO years had more than one west Caribbean in-situ development. However, positive spring NAO conditions and the early onset of La Niña do not exclude the possibility of late season development. All three aforementioned years featured a late-season major hurricane, and this area is notorious for producing major hurricanes, especially toward the latter half of the season. The development of La Niña conditions combined with the presence of warm AMO alone greatly increases the probability of a late season major hurricane. Late season major hurricane formation has been observed in the west Caribbean during 7/8 (88 percent) of recorded La Niña and warm AMO years. Due to overwhelming statistical support, one late season major hurricane is forecast to develop in the west Caribbean Sea.
Gulf of Mexico
The Gulf of Mexico has sustained some of the most notorious hurricanes in the Atlantic Basin. Occasionally, low vertical wind shear, abundant moisture, and seasonably warm SSTs of the Gulf of Mexico align in a manner that can transform incipient systems into monstrous hurricanes. Since the Gulf of Mexico is almost completely surrounded by land, tropical cyclones under supporting conditions have no option but to impact coastal and inland communities. Such was the norm in the destructive 2005 hurricane season, with five major hurricane landfalls along the Gulf coast, including Katrina. In sharp contrast, only one tropical storm traversed the Gulf of Mexico in 2006, so high variability amongst even warm AMO years is apparent.
With that said, the Gulf of Mexico faces another relatively calm season. One of the most intriguing discoveries over the last few months was the connection between the spring pattern and Gulf of Mexico tropical cyclone variability. The SNA, as the authors call it (see section 5), has four phases based on geopotential heights over the continental United States and adjacent Pacific and Atlantic regions. The definitions of each phase will be explained in a future paper. The following statistics demonstrate how the varying configurations of the four phases influence Gulf of Mexico tropical cyclone activity: Phase 1 averages 4.9 named storms, 2.4 hurricanes, and 1.3 major hurricanes; Phase 2 averages 2.1 named storms, 0.9 hurricanes, and 0.1 major hurricanes; Phase 3 averages 3.0 named storms, 1.7 hurricanes, and 0.6 major hurricanes; Phase 4 averages 1.3 named storms, 0.7 hurricanes, and 0.3 major hurricanes. The variance in tropical cyclone activity amongst each phase is surprisingly low as well. Spring 2007 falls under Phase 2 of the SNA. Of the 14 years in the Phase 2 dataset, only one had over three named storms in the Gulf of Mexico. However, the geopotential height pattern in 2007 is very unique compared to past Phase 2 years. The anomalous mid-level ridging that has been present over the entire of the Gulf of Mexico this spring is actually more reminiscent of Phase 4. Regardless, the statistics from both datasets suggest below-average Gulf of Mexico activity overall.
Additionally, other statistical datasets suggest that the number of tropical cyclones entering the Gulf will be minimal. First, it has been concluded that there is little to no chance of a storm entering the Gulf from the west Atlantic and Florida. As aforementioned, an analysis of the spring pattern depicts a prominent trough centered near 60ºW, which is expected to persist in August-September. Thus, all tropical cyclones moving into the west Atlantic will lift poleward far away from the Gulf of Mexico as they follow the path of least resistance. Any tropical cyclone that forms in the west Atlantic will follow a similar track well to the east of the Gulf of Mexico. On the other hand, the late season major hurricane forecast to develop in the west Caribbean Sea will likely move into the Bay of Campeche in a weaker, sub-major state. After analyzing past years with tropical cyclones in the west Caribbean Sea, it is clear that the central Atlantic trough that became established this past spring is not positively-tilted or deep enough to pull a west Caribbean tropical cyclone northeastward into the west Atlantic. Therefore, 1 named storm, possibly a hurricane, is forecast to enter the Gulf of Mexico from the west Caribbean Sea.
Finally, it does not appear that many tropical cyclones will develop in the Gulf of Mexico either. The aforementioned high geopotential heights in the Gulf of Mexico suggest that tropical waves that would otherwise spawn a tropical cyclone will be kept south of the Gulf of Mexico this season. This is even more likely with persistently neutral to high geopotential heights across the Caribbean Sea, which will keep tropical waves from deepening enough to commence a poleward motion component. Moreover, only 1/12 (8 percent) of the Phase 4 SNA years had as many as two named storms form in the Gulf of Mexico: 1977. One of the storms was baroclinically-initiated, made possibly by the El Niño and cool AMO in place. Neither will be present this year. Thus, 0-1 named storms and 0-1 hurricanes are forecast to develop in the Gulf of Mexico during 2007.
7. Activity By Bimonthly Periods
The Atlantic Basin hurricane season officially spans six months, beginning on June 1 and ending on November 30. The season has been divided into three bimonthly periods to help portray the anticipated tropical cyclone formation and timing tendencies of 2007.
Typically, the first two months of the Atlantic Basin hurricane season are relatively quiet. On average, 1-2 named storms form before August. In 2005, a record-setting 7 named storms and 2 major hurricanes developed during this period. Although near record level activity continued throughout all three bimonthly periods of 2005, the amount of activity that occurs in June and July does not necessarily have any bearing on how much activity occurs in the remaining four months of the season.
The primary parameters known to influence June-July activity are the AMO, summer ENSO, and winter ENSO. It is no surprise that warm AMO conditions support above normal activity, albeit slightly. But one statistic could not be explained until halfway through this off-season. Warm AMO and neutral ENSO summers yield more activity than warm AMO and La Niña summers. It was discovered that winter values of the ENSO are to be implemented as part of the seasonal forecast, but only when the summer ENSO is expected to be neutral. When the summer ENSO is neutral, it leaves the door open for residual winter ENSO conditions to influence early season activity. Furthermore, residual winter El Niño conditions are most enhancing. The majority of the most active June-July periods fall within that dataset. It is theorized that residual El Niño effects and neutral summer ENSO conditions result in slightly above normal probability of baroclinic-induced formations and less than normal hindrance of deep tropical activity due to the lack of abnormally strong subtropical ridging typical seen in La Niña summers.
With the Kelvin wave delaying the onset of La Niña, neutral ENSO conditions are expected through the end of July (see section 3). Previous warm AMO seasons that followed winter El Niño conditions and had neutral ENSO conditions during June-July include 1966, 1995, 2003, and 2005. All four of these years advertised above average tropical cyclone activity before August, with no fewer than 3 named storms and 2 hurricanes. A similar evolution of ENSO has been witnessed in the first half of 2007, thus above average June-July activity is likely. However, other parameters suggest that observed activity will remain below that of several ENSO analogs. First, the strong ridging that has extended through the Gulf of Mexico and into the west Atlantic this past spring argues against formation in the Gulf of Mexico or west Caribbean until later in the season. Second, the AMO was very strong in summer 2005 and also slightly stronger in 1966 and 1995 than it is at present. The extremely warm tropical Atlantic SSTAs in 2005 helped fuel the record-breaking June-July activity. On the other hand, the AMO in summer 2003 was closer to neutral and more comparable to current trends. It is no surprise that 2003 had the least amount of June-July activity in the four-year sample. Therefore, 2-3 named storms and 0-1 hurricanes are forecast to develop before August. These developments will likely be confined to the Mean Development Region or west Atlantic.
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 activity being early to mid September. In nearly all cases, an above average hurricane season equates to above average activity during August-September.
The establishment of a worthwhile August-September tropical cyclone forecast method was not as arduous to develop as the June-July formula. First, the warm (cool) AMO signals above (below) normal activity. Second, summer La Niña (El Niño) is the most enhancing (suppressing) phase of ENSO, thus matching the traditional ENSO-tropical cyclone variability relationship. The primary difference in bimonthly forecast methodology is the elimination of winter ENSO as a parameter. Additional statistics developed in recent months indicate that any residual winter ENSO effects become too minimal to have any significant influence on Atlantic tropical cyclone activity by August and September.
The climatological peak of hurricane season 2007 is expected to underscore the effects of a moderately warm AMO combined with La Niña conditions that will have already been present for at least a month. Accordingly, more emphasis has been placed on previous warm AMO and summer La Niña seasons: 1950, 1955, 1961, 1988, 1995, 1998, and 1999. All (7/7) seasons with such conditions produced between 7 and 10 named storms during August-September, which are above the climatological average of 6.5 named storms. Furthermore, all of the aforesaid years advertised average to well above average activity in the Mean Development Region, and most Mean Development Region formations occur within August-September. It has already been concluded that above normal Mean Development Region activity will be observed this year. However, the bimonthly August-September period of 2007 is not forecast to approach the upper confines of the 7-10 named storm range provided by those years for two primary reasons. First, although activity east of the Lesser Antilles will be above normal, it should not be as active as 1995. Additionally, the mid to upper-level geopotential height pattern is not conducive for tropical cyclogenesis in each of the remaining four regions (see section 6). An estimated 7-8 named storms, 5-6 hurricanes, and 2 major hurricanes are forecast to develop during the bimonthly period of August-September, with most formations occurring in the Mean Development Region.
As October approaches, a noticeable decline in Atlantic Basin tropical cyclone activity is typically observed. On average, 2-3 named storms form during October-November, which is far less than August-September, but slightly more than what is typically monitored in June-July. However, in rare cases more activity can occur during October-November than the climatological peak. This scenario last materialized in 2005. Otherwise, several of the most intense hurricanes ever observed in the Atlantic Basin have formed during the October-November period. In short, the Atlantic hurricane season is often far from cessation by late September.
The October-November tropical cyclone forecast methodology is nearly the same as that used for the bimonthly period of August-September. The forecast is highly dependent on the AMO, summer and fall ENSO, and springtime mid to upper-level geopotential heights over the Atlantic Basin and North America. Warm AMO and fall La Niña conditions have yielded anywhere from 2-6 named storms in past bimonthly October-November periods. However, the use of the already given geopotential height pattern should help lower and narrow that range quite a bit. For example, 1950 is one of eight warm AMO and fall La Niña analogs. October 1950 featured 6 tropical cyclones, but 3 of which originated in the Gulf of Mexico. This year, it is unlikely that even one named storm will develop in the Gulf during October for several reasons (see section 6), one of which being the geopotential height pattern. Tropical cyclones simply are not likely to form in the areas where they occurred during analog seasons with more than 3 October-November tropical cyclones. On the other hand, the west Caribbean and Mean Development Region will be primed to support at least one tropical cyclone development, the former likely being a major hurricane. Therefore, 2-3 named storms, 1-2 hurricanes, and 1 major hurricane are forecast to develop during the bimonthly period of October-November.
8. Landfall Activity By Coastal Areas
The concept of correctly forecasting tropical cyclone landfall distribution months in advance is and should be the long-term goal of seasonal forecasting. Seasonal forecast boundaries must be tested in order for such forecasts to become more accurate and useful. Otherwise, seasonal hurricane forecasts will continue to be of little to no value to the general public. In short, many researchers insist that the synoptic steering pattern and consequential tropical cyclone landfalls cannot be forecast months in advance. To the contrary, evaluation of the synoptic springtime pattern in combination with analysis of parameters known to influence Atlantic Basin hurricane activity can yield significant skill in seasonal landfall forecasting. Since the landfall forecasts issued in 2006 were a great success, it has been decided that similar landfall forecasts will be issued for a fifth year.
The method used to generate regional landfall forecasts is essentially unchanged since 2006. Springtime mid-level geopotential heights across North America, the Atlantic Basin, and adjacent regions were analyzed. Additionally, the newly developed four-phase SNA configurations, springtime NAO, AMO, and ENSO are being given more weight this year.
Venezuela and Leeward Antilles
One typically does not think of this area being directly affected by tropical cyclones. But in recent years, a number of notable hurricanes have passed close enough to the Leeward Antilles and Venezuela to cause minor damage and indirect fatalities. Hurricanes Ivan and Emily in 2004 and 2005, respectively, are two noteworthy examples. No such low-latitude tracking tropical cyclones are expected this season for two primary reasons.
First, there are quite a few notable differences between the overall setup during the aforementioned seasons and 2007. In short, the onset of El Niño conditions in 2004 delayed tropical cyclone formation in the Mean Development Region, thus allowing for more storms to develop in close proximity to the Caribbean Sea. The ridging pattern over the central Atlantic was also more supportive of westerly tracking tropical cyclones. In 2005, it was the negative spring NAO that delayed formation in the Mean Development Region. The ridging pattern also favored westerly tracking tropical cyclones for a second consecutive year. This year, nearly all parameters known to affect activity in the deep tropical Atlantic are favorable, thus corresponding to formations farther east and away from the Caribbean Sea. East Atlantic-originating tropical cyclones are naturally less likely to effect the Caribbean since there is more time and opportunity for them to be steered north by atmospheric troughs in conjunction with the Coriolis effect.
Second, Phase 2 of SNA configuration combined with a positive NAO directly argues against low-latitude tropical cyclone tracks across the south Caribbean. Had the SNA met Phase 1 criteria, south Caribbean landfalls would still be a possibility despite the positive NAO. SNA Phase 1 criteria was not met because of an absence of dominant ridging along 20ºN between 60-70ºW. Phase 2 portrays the 2007 setup more accurately with ridging present over the central Atlantic and troughing located beyond 60ºW. No tropical cyclones are forecast to strike Venezuela or the Leeward Antilles.
Landfalls along the Nicaraguan coastline are not common but certainly happen. The last storm was Hurricane Beta in 2005. However, the region should be spared tropical cyclone landfalls in 2007 for a variety of reasons. First, a large mid to upper-level weakness located near 60ºW will be the focal point of recurvature for all tropical cyclones that form east of the Caribbean. Second, Phase 2 of the SNA configuration combined with a positive NAO directly argues against low-latitude tropical cyclone tracks across the lower Caribbean Sea. Third, the combination of La Niña, warm AMO, and positive springtime NAO conditions argue against in-situ east Caribbean Sea formation. With all that said, it is highly unlikely that any tropical cyclones will pass west of 75ºW while still in the Caribbean Sea.
The only potential remaining source for Nicaragua tropical cyclone threats is in-situ west Caribbean Sea development. Even then, the leading parameters argue against multiple west Caribbean Sea formations, especially preceding October. However, one significant late-season tropical cyclone is likely. Springtime pattern tendencies indicate more of a threat to Honduras and the Yucatán Peninsula than points further south. No tropical cyclones are forecast to directly impact Nicaragua.
Honduras and Yucatán Peninsula
This region was hit hard in 2005 by major Hurricanes Emily and Wilma. After a break in 2006, it appears the Yucatán Peninsula and Honduras face yet another major hurricane this year. As aforementioned, no Mean Development Region or east Caribbean Sea-originating tropical cyclones are forecast to enter the west Caribbean Sea. Moreover, only one tropical cyclone, albeit a major hurricane, is expected to develop in the west Caribbean Sea during the latter third of the season (see section 6).
Once it was determined that a late season major hurricane would likely be the highlight of west Caribbean activity, the springtime geopotential height patterns preceding seasons that featured such storms were analyzed. Analogous geopotential height composites based on the following four steering scenarios were then created: Central America landfall followed by dissipation, Central America landfall followed by recurvature into Florida, recurvature resulting in a Florida landfall, and recurvature east of Florida. Variance in the patterns among individual years within all four composites yield only minute variance, thus increasing confidence in the utilization of spring geopotential height composites for this forecast. Moreover, the geopotential height patterns defined by each composite logically match with what one would expect given the observed hurricane tracks among the composite years.
The anomalous ridge over the southeast United States and weakness along 60ºW observed in spring 2007 best match the geopotential height pattern defined by the first composite. All three seasons (1961, 1995, 2000) used in the first composite featured a late-season, major hurricane that originated in the west Caribbean and struck the Yucatán and/or Honduras. Furthermore, all three of those hurricanes were guided by anomalous ridging to the north. The analogous weaknesses located near 60ºW apparently had little influence on steering as they were too weak and far east. Therefore, recurvature over Florida or points east is highly unlikely. With that said, one major hurricane is forecast to make landfall in Honduras or the Yucatán Peninsula. This projected storm will originate in the west Caribbean Sea during the latter half of the season.
The Mexican Gulf coast was hard-hit by Hurricanes Emily and Stan in 2005, but fortunately received a break in 2006 much like the Yucatán Peninsula. The SNA configuration and unique spring geopotential height pattern have been evaluated to gauge the tropical cyclone threat posed this season. Since no tropical cyclones are anticipated to enter the Gulf of Mexico (see section 6) from the Mean Development Region, west Atlantic, or east Caribbean Sea, the threat has been reduced to in-situ west Caribbean and Gulf of Mexico formation. Furthermore, only one tropical cyclone, albeit strong, is expected to originate in the west Caribbean. While the projected tropical cyclone is most likely to strike the Yucatán Peninsula or Honduras, a secondary landfall along the Mexican Gulf coast is not a sure bet. Hurricanes Hattie and Roxanne in 1961 and 1995, respectively, are examples of late season major hurricanes that struck the Yucatán and later dissipated over the Bay of Campeche. On the other hand, Keith in 2000 hit the Mexican state of Verecruz as a category 1 hurricane. Although it is unlikely that the west Caribbean major hurricane will restrengthen into a major hurricane in the Gulf of Mexico, there is the possibility of a secondary weaker landfall along the Mexican Gulf coast.
Finally, given that the spring geopotential heights over the Gulf of Mexico and adjacent Caribbean and North American regions bear resemblance to Phase 4 SNA years, no more than one tropical cyclone is forecast to form in the Gulf of Mexico. However, the lack of any troughing immediately north or northeast of the Gulf of Mexico argues against recurvature into the United States Gulf coast. Rather, the abnormal degree of ridging that has been in place over the southeast United States supports the idea that any Gulf-originating tropical cyclones will head westward into Mexico. Therefore, the Mexican Gulf coast has a high risk of being impacted by 1-2 named storms and 0-1 hurricanes, all stemming from in-situ development in the west Caribbean Sea and Gulf of Mexico.
While Jefferson and Orange counties were significantly impacted by Hurricane Rita in 2005, no tropical cyclone has made a direct landfall in Texas since 2003. This streak should continue through 2007. Climatologically, springtime troughs or regions of low geopotential heights over Texas and north Mexico have preceded tropical cyclone landfalls along the Texas coastline. With that said, the 2007 spring pattern featured strong ridging over the central United States, northern Mexico, and Gulf of Mexico. Originally, it was theorized that the anomalously strong easterly flow associated with the ridging over the Gulf of Mexico could still allow a tropical cyclone to hit Texas. However, three years with analogous spring geopotential heights (1967, 1990, and 2000) were considered. During those years, all tropical cyclones that formed in the southwest Gulf of Mexico or west Caribbean Sea continued westward and struck Mexico, well south of Texas. Since the spring ridge extends even further south this year, and geopotential heights in the Caribbean Sea have been neutral to slightly above average, the few tropical cyclones that enter or originate in the Gulf of Mexico will almost certainly be blocked from striking Texas. Accordingly, no tropical cyclones are forecast to hit coastal Texas this season.
Louisiana, Mississippi, and Alabama
It is a significant understatement to say the coastline from Louisiana through the Big Bend of Florida was hard-hit in 2005. This region was forever changed from Hurricane Katrina, not to mention the barrage of three other hurricanes. All in all, the recovery process will take several more years, not months, and even then some areas will never be the same. All one can do is hope that mother nature spares the central Gulf coast from additional landfalls in the near future. Subsequently, one of the leading questions heading into the 2007 hurricane season is whether a tropical cyclone is likely to threaten the central Gulf coast. The 2007 outlook is encouraging for interests in this area.
The upper air pattern over the central Gulf coast has behaved similarly to that observed in 2006. Unusually strong high pressure dominated the southeast and central United States for the better half of spring 2007. Analogous patterns have kept all tropical cyclones away from the central Gulf coast in the past. The central Atlantic weakness should prevent any tropical cyclones from entering the southeast Gulf of Mexico from the Bahamas. Meanwhile, the aforementioned ridging pattern over the southeast should keep any Gulf of Mexico or west Caribbean-originating storms south and west of the central Gulf coast. Furthermore, the SNA configuration suggests that Gulf of Mexico tropical cyclone activity as a whole will be below average. Therefore, no tropical cyclones are forecast to impact the central Gulf coast.
The west coast of Florida was recently struck by major Hurricanes Charley and Wilma in 2004 and 2005, respectively. Tropical storm activity was also noted along the coastline in 2006. The risk of tropical cyclone landfalls in West Florida has been deemed to be low this year.
While the ridging pattern over the central Gulf coast does resemble that observed in 2006, there are notable differences elsewhere that pertain to Florida peninsula threats. Last season, the southeast United States ridge, in tandem with a massive central Atlantic trough and a developing El Niño, supported tropical cyclone recurvature over the southeast Gulf of Mexico. This year, the ridge has routinely expanded beyond the East coast, thus making the central Atlantic trough less of a recurvature mechanism for west Caribbean and Gulf of Mexico tropical cyclones. Close comparisons between the spring patterns of 1995, 2000, and 2007 were made since all three shared basic pattern similarities. However, the central Atlantic trough in 1995 and 2000 extended much further south and west than what has been observed this past spring. The trough, combined with low geopotential heights within the Caribbean, resulted in a pattern that enabled tropical cyclones to curve into west Florida during those years. On the other hand, 1961 and 1967 had a more similar ridge-trough orientation and no west Florida landfalls were observed. Finally, the intensity of the ridging noted over the Florida peninsula, in combination with developing La Niña conditions, argues for more westerly tracks of any tropical cyclones that form in the west Caribbean or Gulf of Mexico. Thus, no tropical cyclones are forecast to directly hit the west coast of Florida.
After experiencing three hurricane landfalls in 2004-2005, not including a brush from Hurricane Rita, the east coast of Florida was largely spared in 2006. Fortunately, the break for the east coast of Florida is expected to continue through 2007. A few correlations have been noted regarding East Florida landfalls. First, years with positive spring NAO combined with high (low) geopotential heights south of Greenland usually feature increased (decreased) east Florida tropical cyclone activity. Second, years with winter El Niño (La Niña) conditions prior to the hurricane season also average more (less) east Florida landfalls, especially during positive or negative (neutral) spring NAO. While the physical mechanisms responsible for these statistical correlations remain unexplained at the present time, they are worth taking into consideration. The geopotential heights south of Greenland this past spring have been slightly below average. On the other hand, this past winter featured an El Niño, and this past spring was characterized by a positive NAO. The contradictory signals forced a closer examination of the overall Atlantic Basin pattern.
As aforementioned, this past spring featured a weakness centered between strong ridging over the continental United States and northeast Atlantic. An analogous feature was observed in 1995, a year that also had a winter El Niño, spring positive NAO, and slightly below average geopotential heights south of Greenland. All of the tropical cyclones that developed in the Mean Development Region in 1995 recurved into the central Atlantic weakness. Likewise, all Mean Development Region or central Atlantic-originating tropical cyclones should recurve east of the Florida peninsula in 2007. The only evident differential between 1995 and 2007 that pertains to east Florida is the geopotential height pattern over the west Atlantic, which has been more positive this year. Stronger geopotential heights in tandem with developing La Niña conditions do not support such west Atlantic developments that occurred in 1995. Any in-situ west Atlantic tropical cyclogenesis in 2007 is most likely to occur well east of Florida and the Bahamas. Any formation east of the Bahamas is more likely to recurve out to sea rather than veer into Florida given the spring pattern. Therefore, no tropical cyclones are forecast to hit the east coast of Florida.
Georgia, South Carolina, and North Carolina
The Carolinas are certainly no stranger to tropical cyclones. The Outer Banks of North Carolina jut outward and are thus especially prone to hurricane strikes. However, this region will largely escape tropical cyclone activity in 2007. First, unusually strong ridging over the continental United States will prevent any Gulf of Mexico or west Caribbean originating tropical cyclones from recurving through Florida and the East coast. Second, the eastern periphery of the ridging over the East coast, in tandem with warm AMO and developing La Niña conditions, does not support coastal tropical cyclogenesis. Any in-situ west Atlantic tropical cyclone development is most likely to occur well south and east of Georgia and the Carolinas. Finally, a geopotential height weakness centered between two high pressure systems, one over the continental United States and the other over the northeast Atlantic, should permit all west Atlantic and Mean Development Region-originating tropical cyclones to recurve near 60ºW, well east of the United States East coast. Thus, no tropical cyclones are forecast to make landfall along the Georgia, South Carolina, or North Carolina coastlines.
Mid-Atlantic and Northeast United States
One may not consider states north of the Carolinas to be frequently hit by tropical cyclones, but the region has certainly seen its fair share of destructive landfalls. Fortunately, no such landfalls should occur in 2007. The springtime North American geopotential height configuration suggests that any tropical cyclones that do form in the west Atlantic or Mean Development Region will recurve well to the east near 60ºW.
Furthermore, winter El Niño (La Niña) conditions preceding hurricane season average less (more) Mean Development Region-originating tropical cyclones that impact the Northeast states, especially during neutral (extreme) spring NAO years. In fact, all seven seasons that featured such tracks followed winter neutral or cool ENSO conditions and a moderate spring NAO. It is interesting to note that the opposite is observed with east Florida landfalls, which are more frequent in seasons following warm ENSO winters and extreme spring NAO values. This might suggest that the factors utilized in this study have a lag influence on tropical cyclone steering patterns which either forces storms to curve northward into the Northeast states or continue into Florida. Another statistic worth pointing out is that west Atlantic-developing tropical cyclones that hit the Northeast states were all in seasons with an El Niño, cool AMO, or both. Since neither will be present in 2007, it is unlikely that there will be many formations in the west Atlantic, much less one making landfall in the Northeast states. Accordingly, no tropical cyclones are expected to hit the Mid-Atlantic or Northeast states this season.
Contrary to popular belief, the Atlantic coast of Canada is no stranger to tropical cyclones, with the most notable being Hurricane Juan of 2003 that devastated Halifax, Nova Scotia. The mean trough position during August-September should be located near 60ºW, which runs perpendicular to the easternmost half of the Atlantic Canada. With 5-6 hurricanes forecast to recurve along that general longitude, it is reasonable to presume that at least a few tropical cyclones will threaten or bypass the coastline this year. Moreover, tropical cyclones that develop in the west or northeast Atlantic are likely to take similar paths. This season, Newfoundland has a higher risk of tropical cyclone brushes and possibly even 1-2 landfalls than the maritime provinces of New Brunswick, Nova Scotia, and Prince Edward Island. Most or all tracks should remain south and just east of the maritime provinces.
Tropical cyclones occasionally hit the Azores. In fact, category 1 Hurricane Gordon passed through the islands in 2006. A similarly strong tropical cyclone is unlikely to impact the Azores this year. This past spring has featured a strong mid-level ridge centered in close proximity to the islands that extends west to 55-60ºW. Thus, the multiple tropical cyclones that form in the Mean Development Region are expected to recurve west of 40ºW, a position far enough west to avoid the Azores. Moreover, the ridge will prohibit tropical cyclones from forming near the Azores. However, 1-2 named storms are expected to form between 40-60ºW in the northeast Atlantic. The recurvature trajectories of such storms are more difficult to forecast, thus some impact from these storms cannot be ruled out.
Although Cape Verde is often impacted by developing tropical depressions, actual named storm hits are uncommon since the time of passage over water from Africa is limited. However, the few stronger systems that have hit Cape Verde resulted in significant damage. With a strong ridge in place over the east Atlantic this past spring, such a direct hit appears unlikely this year. Thus, any early tropical cyclone development off the west coast of Africa is expected to occur south of Cape Verde.
Puerto Rico, Leeward Islands, and Windward Islands
Following the 1998 and 1999 hurricane seasons, Puerto Rico and the Leeward Islands have managed to escape a direct hit from a significant hurricane. Unfortunately, there are indications that the eight-year streak of no landfalls is about to be shattered. There is a high risk of 1-2 significant tropical cyclone landfalls in Puerto Rico and the Leeward Islands.
Two ingredients are necessary for this area to be considered a high risk zone: favorable conditions in the Mean Development Region and a steering mechanism that will keep any Mean Development Region-originating tropical cyclones on a westward heading toward the islands. Obviously, the Mean Development Region is the primary source region of tropical cyclones that impact the northeast Caribbean. This year, the area is expected to be highly favorable for tropical cyclogenesis. A combination of warm AMO, La Niña, and positive springtime NAO conditions favors a strong Cape Verde hurricane season, with 7-8 named storms, 4-5 hurricanes, and 1-2 major hurricanes forecast to originate in the Mean Development Region. More importantly, stronger than normal mid to upper-level ridging has dominated the subtropical Atlantic in recent months. The ridging often extends as far west as 60ºW, which is where an overall weakness in the synoptic pattern has taken shape. The majority of tropical cyclones that do form in the Mean Development Region will harmlessly recurve northeast of the island chain. However, the probability that all 7-8 tropical cyclones will recurve east of the islands is very low given the degree of ridging that has been in place. Furthermore, the 1-2 tropical cyclones that continue westward into the northeast Caribbean while underneath the large subtropical ridge will likely be intense due to synoptic reasons.
The one bit of good news is that all Mean Development Region-originating tropical cyclones should remain north and east of the Windward Islands. The degree of ridging along 20ºN decreases significantly by 60ºW. Thus, all tropical cyclones that threaten the east Caribbean should begin to recurve no later than 55-60ºW. Moreover, all parameters indicate that tropical cyclone formation immediately east of the Windward Islands is unlikely. Thus, no tropical cyclone landfalls on the Windward Islands are expected.
Bahamas and Turks and Caicos Islands
Tropical cyclones frequently impact the Bahamas and Turks and Caicos Islands, but a number of parameters argue against a direct hit in 2007. First, a weakness centered between two high pressure systems, one over the continental United States and the other over the northeast Atlantic, should permit all west Atlantic and Mean Development Region-originating tropical cyclones to recurve east of the Bahamas and Turks and Caicos Islands. Such tracks are also supported by a combination of positive springtime NAO, warm AMO, and La Niña conditions. Second, no tropical cyclones are forecast to originate in the east Caribbean Sea (see section 6). Third, the anomalous ridging centered over the continental United States should prevent any late-season, west Caribbean tropical cyclones from recurving into Cuba and the Bahamas. Fourth, the eastern periphery of the ridge in tandem with La Niña and warm AMO conditions will hinder west Atlantic development. Finally, the pattern also suggests that any in-situ west Atlantic tropical cyclones would likely recurve away from the Bahamas and Turks and Caicos Islands. No tropical cyclones are forecast to directly impact this region.
Cayman Islands, Cuba, Jamaica and Hispaniola
Tropical cyclones that have hit these islands in the past have resulted in significant damage, but it looks as if the general area will escape the wrath of the 2007 hurricane season despite being located directly between the two “highest risk” regions of the Atlantic Basin. First, a weakness centered between two high pressure systems, one over the continental United States and the other over the northeast Atlantic, should permit all west Atlantic and Mean Development Region-originating tropical cyclones to recurve over or east of Puerto Rico, thus sparing Hispaniola and points westward from a direct landfall. Second, the combination of La Niña, warm AMO, and positive springtime NAO conditions argue against in-situ east Caribbean Sea formation. Finally, any late season tropical cyclone that may form in the west Caribbean Sea is likely to continue westward into Honduras or the Yucatán Peninsula rather than recurving due to the abnormal ridging noted over the southeast United States last spring. No tropical cyclones are forecast to directly impact the Cayman Islands, Cuba, Jamaica, and Hispaniola.
Located in the middle of the Atlantic Basin, Bermuda is often brushed by tropical cyclones. Because of its small size, however, direct landfalls do not occur often. In 2007, the mean trough position will put Bermuda in danger of getting sideswiped by several recurving hurricanes that originated in the Mean Development Region or west Atlantic. It is difficult to say if any will directly hit the island, but residents and tourists should be prepared to deal with multiple hurricane threats during August and September.
A moderately warm AMO combined with La Niña conditions should allow the upswing in Atlantic Basin tropical cyclone activity that began in 1995 to continue in 2007. Above average activity is forecast in the Mean Development Region, with at least one major hurricane striking the northeast Caribbean before recurving into the west Atlantic. A weakness between two intense ridges over the east Atlantic and continental United States will act as a protective barrier of tropical cyclones for the United States East coast. No tropical cyclones are expected to undercut the weakness and continue westward through the Caribbean Sea. Meanwhile, ridging over the continental United States should keep any Gulf or Caribbean-originating tropical cyclones away from the United States Gulf coast. However, the Mexican Gulf coast and east shore of the Yucatán Peninsula run a risk of landfalls stemming from homegrown tropical cyclone activity in the southwest Gulf of Mexico and west Caribbean Sea. The featured homegrown development in the west Caribbean Sea is expected to be a late-season major hurricane. Finally, the forecast total number of tropical cyclones in June-November is presented below. While the 2007 season is expected to be somewhat above average, the level of activity should be far less than that observed in 2004 or 2005.
IWIC 2007 Atlantic Basin Hurricane Season Forecast
Off-season formations not included.
Regardless of seasonal expectations, interests in hurricane-prone areas should always be prepared for a tropical cyclone landfall. Even if 2007 unexpectedly features below normal tropical cyclone activity, it would only take one intense landfalling hurricane to make the season a devastating one. Such a scenario occurred 15 years ago. All measures of tropical cyclone activity were below normal in 1992. However, it was the year that category 5 Hurricane Andrew slammed south Florida and Louisiana, which resulted in 65 fatalities and over 30 billion dollars of damage.
No changes will be made to the 2007 seasonal forecast over the course of the season, though updates will be posted within the IWIC daily tropical weather discussions. A post-review of the 2007 season will be underway by December. Such timing will enable the forecasters to analyze how well the latest methodology fared, and then make adjustments for 2008 if necessary. Preliminary research on the 2008 Atlantic Basin hurricane season will ensue January.