Christopher W. Landsea
Landsea, C. W., 2000: El Niño-Southern Oscillation and the seasonal predictability of tropical cyclones. In press in El Niño : Impacts of Multiscale Variability on Natural Ecosystems and Society, edited by H. F. Diaz and V. Markgraf.
"Tropical cyclone" is the generic term for a non-frontal synoptic scale low-pressure system that develops over tropical or sub-tropical waters with organized convection and a well-defined cyclonic surface wind circulation. Its energy source is primarily derived from evaporation and sensible heat flux from the sea in the presence of high winds and lowered surface pressure. These energy sources are tapped through condensation and fusion in convective clouds concentrated near the cyclone's "warm-core" center (Holland 1993). Tropical cyclones with maximum sustained surface winds of less than 18 ms-1 are called "tropical depressions". Once the tropical cyclone reaches winds of at least 18 ms-1 they are typically called a "tropical storm" and assigned a name. Names are decided upon by representatives from countries in the basins affected at annual World Meteorological Organization regional meetings (Neumann 1993). If winds reach 33 ms-1, they are then called: a "hurricane" (the North Atlantic Ocean, the Northeast Pacific Ocean east of the dateline, or the South Pacific Ocean east of 160°E); a "typhoon" (the Northwest Pacific Ocean west of the dateline); a "severe tropical cyclone" (the Southwest Pacific Ocean west of 160°E or Southeast Indian Ocean east of 90°E); a "severe cyclonic storm" (the North Indian Ocean); and a "tropical cyclone" (the Southwest Indian Ocean) (Neumann 1993). Additionally, the category of "intense (or major) hurricane" has been utilized for the Atlantic basin for those tropical cyclones obtaining winds of at least 50 ms-1, which corresponds to a category 3, 4 or 5 on the Saffir-Simpson hurricane intensity scale (Simpson 1974, Hebert et al. 1996).
It should be pointed out that such definitions are quite arbitrary ones
and that nearly all intensity wind values at the surface are an estimation
(by satellite pictures) or an extrapolation (from aircraft reconnaissance
downward to the surface). Thus by the nature of the tropical cyclone, by
the limited data available and by the way that meteorologists have defined
intensity thresholds, the strength of individual tropical cyclones can
be difficult to pinpoint with certainty. Also, changes in observational
platforms available to monitor tropical cyclones can produce as much or
greater change in the cyclone record as can actual climate fluctuations.
Studies of interannual (and especially interdecadal) changes of tropical
cyclones must carefully consider both the relative arbitrariness of the
intensity record of the storms and the dependency of intensity on the observations
Before tropical cyclogenesis and development can occur, there are several precursor environmental conditions that must be in place (Gray 1968, 1979):
1. Warm ocean waters (of at least 26.5 °C) throughout a sufficient depth (unknown how deep, but at least on the order of 50 m). Warm sea surface temperatures (SSTs) are necessary to fuel the heat engine of the tropical cyclone1.
2. An atmosphere which cools fast enough with height such that it is potentially unstable to moist convection. It is the precipitating convection typically in the form of thunderstorm complexes which allows the heat stored in the ocean waters to be liberated for tropical cyclone development.
3. Relatively moist layers near the mid-troposphere. Dry mid levels are not conducive for allowing the continued development of widespread thunderstorm activity because entrainment into the thunderstorms dries and cools the rising parcel, reducing buoyancy.
4. A minimum distance of around 500 km from the equator. For tropical cyclogenesis to occur, there is a requirement for non-negligible amounts of the Coriolis force to provide for near gradient wind balance to occur. Without a substantial Coriolis force, inflow into the low pressure is not deflected to the right (to the left in the Southern Hemisphere) and the partial vacuum of the low is quickly filled.
5. A pre-existing near-surface disturbance with sufficient vorticity and convergence. Tropical cyclones cannot be generated spontaneously. To develop, they require a weakly organized system with sizable spin and low level inflow.
6. Low values (less than about 10 ms-1) of vertical wind shear between the 850 and 200 mb levels. Vertical wind shear is the magnitude of wind change with height. Large values of vertical wind shear disrupt the incipient tropical cyclone and can prevent genesis, or, if a tropical cyclone has already formed, large vertical shear can weaken or destroy the tropical cyclone by interfering with the organization of deep convection around the cyclone center (DeMaria 1996).
Having these conditions met is necessary, but not sufficient, as many
disturbances that appear to have favorable conditions do not develop. Recent
work (Velasco and Fritsch 1987, Chen and Frank 1993, Emanuel 1993) has
identified that large thunderstorm systems (called mesoscale convective
complexes [MCCs]) often produce an inertially stable, warm core vortex
in the trailing altostratus decks of the MCC. These mesovortices have a
horizontal scale of approximately 100 to 200 km, are strongest in the mid-troposphere
and have no appreciable signature at the surface. Zehr (1992) hypothesizes
that genesis of the tropical cyclones occurs in two stages: stage one occurs
when the MCC produces a mesoscale vortex and stage two occurs when a second
blow up of convection at the mesoscale vortex initiates the intensification
process of lowering central pressure and increasing swirling winds.
Variations in environmental conditions that affect tropical cyclone
Seasonal variations of tropical cyclone activity depend upon changes in one or more of the above parameters. Many studies have focused upon the variations in these values both before and during the tropical cyclone season. While the bulk of these studies has been centered upon the Atlantic basin, all of the global basins have been analyzed to some degree for interannual predictability.
Globally, tropical cyclones are affected dramatically by the El Niño-Southern Oscillation (ENSO). ENSO is a fluctuation on the scale of a few years in the ocean-atmospheric system involving large changes in the Walker and Hadley Cells throughout the tropical Pacific Ocean region (Philander 1989). The state of ENSO can be characterized, among other features, by the sea surface temperature (SST) anomalies in the eastern and central equatorial Pacific: warmings in this region are referred to as El Niño events and coolings are La Niña events. The Southern Oscillation Index (SOI), the standardized difference in sea level pressure between Tahiti and Darwin, Australia, also describes the state of ENSO with high (low) pressures at Darwin and low (high) pressure at Tahiti corresponding to El Niño (La Niña) events. ENSO greatly alters global atmospheric circulation patterns and it is able to affect tropical cyclone frequencies primarily by altering the lower tropospheric source of vorticity and by changing the vertical shear profile.
The various basins do not respond identically to ENSO. Some show changes in frequency of cyclogenesis, while others show shifts in the genesis locations. These variations are due to the time of year that the basin reaches its peak in activity versus the annual cycle of ENSO, the location of the basin with respect to the central equatorial Pacific and the background climatological flow features within the basin. Basins within the Pacific can be partially forced by direct alterations of the SSTs in the genesis regions, however, most basins experience remote forcing through alteration of the tropospheric flow features. It is the combination of spatial, temporal and climatological factors that determine how individual tropical cyclone basins will be altered by ENSO.
Nicholls (1979) first identified that the tropical cyclones In the vicinity of Australia (90°E to 165°E), are reduced in number during the warm phase (or El Niño) of ENSO. Revell and Goulter (1986) and Hastings (1990) demonstrated that the reduction of Australian region tropical cyclones is compensated by an increase in the South Pacific east of 165°E (Fig. 1), because of a shift in the center of action in tropical cyclone genesis. There is also a smaller tendency to have the tropical cyclones originate a bit closer to the equator (Revell and Goulter 1986). The opposite is observed in La Niña events. This appears to be due to a weakening of the Australian monsoon trough (e.g. the boundary between the cross-equatorial near surface westerlies and the tradewind easterlies - see McBride 1995) in the western portion of the basin and an extension of this trough well to the east of its usual location during an El Niño event, thus changing the availability of lower tropospheric large scale cyclonic circulation and convergence for the storms to develop (Fig. 2 - Evans and Allen 1992). Evans and Allen (1992) also identified a regional change for the Northern Territory of Australia that is opposite to the general tendency for the entire basin. They found fewer tropical cyclones (and fewer landfalls) during La Niña than in El Niño years because of a stronger - though landlocked - monsoon trough. Such an overland positioning of the monsoon trough, while allowing for large rainfall production over northern Australia, is not conducive for tropical cyclone formation because genesis of tropical cyclones requires an oceanic moisture and heat source.
Likewise, the Northwest Pacific basin experiences a similar change in the location of tropical cyclone genesis without a total change in frequency. Pan (1981), Chan (1985), and Lander (1994) have detailed that west of 160°E there are reduced numbers of tropical cyclones forming and from 160°E to just east of the dateline an increase in the amount of genesis occurring during El Niño events (Fig. 3). The opposite occurs during La Niña events. Changes in the monsoon trough location and strength again appear to dictate the tropical cyclone variations, though there has been no documentation of this possible effect. Additionally, Lander (1994) uncovered a mid-season increase in tropical cyclones forming in subtropical latitudes (20 to 30°N) during La Niña events, which he hypothesized to be tropical cyclogenesis forced by the Tropical Upper Tropospheric Trough (TUTT; a persistent, summer-autumn, "cold-core" trough with maximum amplitude at the tropopause that occurs primarily over the tropical and subtropical mid-oceans - see Fitzpatrick et al. 1995) within the tradewind belt.
The western portion of the Northeast Pacific basin near Hawaii (140°W to the dateline) has been suggested to experience more tropical cyclone genesis during an El Niño year and more tropical cyclones tracking into the sub-region in the year following an El Niño (Schroeder and Yu 1995). The opposite effects of La Niña have yet to be analyzed and the mechanism for such changes is unclear at this time.
While the previously mentioned studies have focused upon the ability to change conditions locally in altering the tropical cyclogenesis frequencies, the Atlantic basin feels the effects of ENSO remotely through changes in the vertical shear wind profile. During El Niño events, the vertical shear increases primarily due to increases in the climatological westerly winds in the upper troposphere (Fig. 4) and reduced 200mb westerlies and shear during La Niña events (Gray 1984a, Shapiro 1987). The larger (smaller) vertical shear accompanying El Niño (La Niña) events lead directly toward decreased (increased) numbers of Atlantic hurricanes. Goldenberg and Shapiro (1996) identified that the area between 10 and 20°N from North Africa to Central America (hereby known as the Atlantic "main development region") shows the largest sensitivity toward changes in the vertical shear, with weakly opposite conditions occurring in the subtropical latitudes of 20 to 35°N (Fig. 5). This tendency for weaker (stronger) vertical shear in the subtropical latitudes during El Niño (La Niña) events may account for increasing (decreasing) the number of subtropical forming tropical cyclones, though these changes in the subtropical latitudes are weaker in magnitude to the changes occurring in the main development region. Additional impacts of ENSO on Atlantic climate can be found in Enfield and Mayer (1997) and in the Enfield and Mestas-Nuñez (1997) chapter in this book.
The remaining basins - the eastern portion of the Northeast Pacific (the North Pacific Ocean from 140 °W to North America), the Southwest Indian and the North Indian - appear to have little ENSO-forced variations (i.e. Jury 1993, Dong and Holland 1994, McBride 1995), though there may be ENSO relationships produced in these areas that have not yet been identified.
Beside the El Niño-Southern Oscillation, there is another global factor that appears to force changes in tropical cyclones: the stratospheric Quasi-Biennial Oscillation (QBO), an east-west oscillation of stratospheric winds that encircle the globe near the equator (Wallace 1973). This oscillation has a distinct effect upon Atlantic (more activity in the west phase [Fig. 6] - Gray 1984a, Shapiro 1989), Southwest Indian (more activity in the east phase - Jury 1993) and Northwest Pacific (more activity in the west phase - Chan 1995) tropical cyclones. While the exact mechanism of the stratospheric QBO's influence on tropical cyclones is uncertain, it has been hypothesized that upper tropospheric to lower stratospheric vertical shear variations (Gray et al. 1992b) and/or upper tropospheric static stability changes (Knaff 1993) may be responsible.
In addition to the global effects of ENSO and QBO, there are also local effects that appear to directly impact tropical cyclone frequency within individual basins. These include variations of local sea level pressures, SSTs and tradewind and monsoon circulations.
Sea level pressures act to directly impact the strength of the vertical wind shear. For example in the Atlantic basin because of a relatively invariant sea level pressure field near the equator, above (below) normal SLP in the main development region from 10 to 20°N between Africa and the Americas tightens (loosens) the local pressure gradient and strengthens (weakens) the easterly tradewinds by 1 to 3 m s-1, thereby contributing to increased (decreased) vertical shear (Gray et al. 1993, 1994). Additionally, Gray et al. (1993) have suggested that abnormally low SLP indicates a poleward shift and/or a strengthening of the Intertropical Convergence Zone (ITCZ). Both situations contribute to less subsidence and drying in the main development region through which easterly waves move. Knaff (1997) indicates that low SLP is accompanied by a deeper moist boundary layer and a weakened tradewind inversion. Moreover, an enhanced ITCZ provides more large-scale, low level cyclonic vorticity to incipient tropical cyclones, thereby creating an environment that is more favorable for tropical cyclogenesis (Gray 1968). In contrast, above normal SLP tends to be associated with opposite conditions which are unfavorable for tropical cyclogenesis. Ray (1935), Brennan (1935), Shapiro (1982), Gray (1984b) and Gray et al. (1993, 1994) have discussed the relationship between sea level pressure anomalies and Atlantic basin activity, while Nicholls (1984) has analyzed Australian tropical cyclones and local pressure values.
Sea surface temperatures in the genesis regions of tropical cyclone basins have a direct thermodynamic effect on tropical cyclones through their influence on moist static stability (Malkus and Riehl 1960). Pacific SSTs also indirectly influence the vertical shear through its strong inverse relationship with surface pressures in some regions (Shapiro 1982, Gray 1984b, Nicholls 1984). (These direct and indirect effects of local SST variations are considered separately from the remote forcings of the SST modulations directly due to ENSO.) In particular for the Atlantic basin, warmer than average waters are usually accompanied by lower than average surface pressures, and thus, weaker tradewinds and reduced shear. Cooler than average waters are usually accompanied by higher pressure, stronger tradewinds and increased shear. Somewhat surprisingly, interannual SST variations have relatively small or negligible contributions toward increasing the tropical cyclone frequency in most basins. Only the Atlantic, Southwest Indian and Australian regions have significant though small, positive associations in the months directly before the tropical cyclone seasons begin (Raper 1992, Shapiro and Goldenberg 1997). However, Saunders and Harris (1997) provide substantial evidence that both preceding and during the hurricane season that Atlantic SSTs in the main development region contribute a large percentage of the variance explained (over 30% during the height of the season) with the number of hurricanes generated in that area. Indeed they argue through a partial correlation analysis that these Atlantic SSTs are the dominant physical modulator of tropical Atlantic hurricanes. In addition to these studies, Ray (1935), Carlson (1971), Wendland (1977) and Shapiro (1982) have also examined the Atlantic basin, Jury (1993) has investigated the Southwest Indian, and Nicholls (1984) and Basher and Zheng (1995) have analyzed the Australian/Southwest Pacific for SST associations.
One aspect that has recently been uncovered is the association of a tropical cyclone basin with its generating (or nearby) monsoon trough. As previously discussed, Evans and Allen (1992) identified that variations in the Australian monsoonal flow can be associated with changes in tropical cyclone activity such that a strong (weak) monsoon circulation during a cold (warm) phase of ENSO is accompanied by many (few) tropical cyclones. Bate et al. (1989) also suggested that variations in the Australian monsoon could alter the tropical cyclone activity, independent of any pronounced ENSO events. Over the Atlantic basin, June through September monsoonal rainfall in Africa's Western Sahel has shown a very close association with intense hurricane activity (Fig. 7 - Reed 1988, Gray 1990, Landsea and Gray 1992, Landsea et al. 1992). Wet years in the Western Sahel (e.g. 1988 and 1989) are accompanied by dramatic increases in the incidence of intense hurricanes, while drought years (e.g. 1990 through 1993) are accompanied by a decrease in intense hurricane activity. Variations in tropospheric vertical shear and African easterly wave intensity have been hypothesized as the physical mechanisms that link the two phenomena (Gray 1990, Landsea and Gray 1992), although Goldenberg and Shapiro (1996) have demonstrated that changes in the vertical shear probably dominate. They note that wet (dry) years are associated with reduced (increased) wind shear, due to both weaker (stronger) than average lower tropospheric tradewinds and upper tropospheric westerlies throughout the main development region.
A final factor that has been considered for forcing interannual variations
of tropical cyclone activity is changes in the "steering flow" in which
the storms are embedded. (To a first approximation, tropical cyclones can
be considered to be steered by the surrounding deep layer [the ocean surface
to 100 mb] atmospheric flow features [Franklin et al. 1996].) Namias (1955)
and Ballenzweig (1959) first suggested that interannual variations in the
mid-tropospheric flow fields could help account for both variations in
Atlantic basin tropical cyclogenesis and in the tracks of the storms once
formed. While their ideas regarding genesis have not borne out, the hypothesis
regarding changes in steering have held up. Shapiro (1982) confirmed that
mid-tropospheric flow features can account for sub-regions within the Atlantic
basin experiencing more or less activity in any particular year.
Currently, the only feasible methodology for seasonal tropical cyclone forecasting is by the use of statistical regression models. Eventually, the use of numerical models (or global circulation models - GCMs) to produce seasonal forecasts may also be possible. Indeed, there have been a couple of encouraging steps forward (e.g. Wu and Lau 1992; Watterson et al. 1995) that have shown that - either directly through the number of tropical cyclone-like vortices or indirectly through measurements of crucial environmental fields - there may someday be skill with such models. However, real-time skill today is unattainable because of a) the inability in some GCMs to produce a realistic representation of tropical cyclones in the coarse grid spacing available; b) the complete lack of a stratospheric QBO - shown earlier to be a crucial component in the tropical cyclone variability of many regions - in the GCMs; and c) the inability to forecast the oceanic boundary conditions including the timing, phase and magnitude of the El Niño-Southern Oscillation phenomena as well as local SST anomalies. However, as detailed below, statistical forecasting schemes have already and are continuing to provide skilled and useful predictions of tropical cyclone activity around the world.
With the completion of the 1996 hurricane season, Prof. William Gray and colleagues at Colorado State University (U.S.) have issued real-time seasonal hurricane forecasts for thirteen years. The original forecasting procedures are described in Gray (1984a,b), but have since been substantially redeveloped and improved. Forecast techniques have been developed from the analysis of data going back to 1950. Instead of an ordinary least squares (OLS) regression technique, Gray et al. (1992a, 1993, 1994) have utilized a linear regression model based upon the least absolute deviations (LAD). LAD creates regression lines that are fitted to the data by minimizing the actual distance between hindcasted values and the observations. This differs from the more traditional OLS regression approach that is based upon the unphysical square of the same distance. Thus all observations are weighted equally in LAD rather than an undue emphasis on the outliers that is seen in OLS. Complimentary with LAD is the use of the agreement coefficient, r, which provides a measure of the fit of hindcasted and observed tropical cyclone values. The agreement coefficient (Mielke 1991) measures skill by comparing the absolute differences between hindcasted and observed values versus a random assortment of these absolute differences: a r = 0 indicates absolutely no agreement between hindcasted and observed values and a r = 1 indicates perfect agreement between the two. Values of r that range from 0 to 1 can be considered the amount of variability that the hindcasts can explain in the observations.
Forecasts issued at the end of the previous year's hurricane season are a fairly recent endeavor. The 1 December forecast is based upon five predictors (Gray et al. 1992a). These predictors include those based upon the extrapolated state of the stratospheric QBO through the zonal winds at 50 mb, 30 mb and the vertical shear of the zonal winds between the two levels and previously measured North African rainfall - August and September precipitation within the western Sahel and August through November precipitation along the Gulf of Guinea. Table 1 lists these predictive groupings and Fig. 8 shows the location of these various predictors.
Because of the consistency of the QBO, successful long range extrapolations of the mean stratospheric zonal winds can be made almost a year in advance. For this 1 December forecast time, an extrapolation of mean following-year September QBO conditions is made based upon November information. The two West African rainfall indices are needed for Atlantic tropical cyclone forecasting because of the intimate link between concurrent seasonal amounts of intense hurricane activity and seasonal rainfall in the Sahel of West Africa (Landsea and Gray 1992). Gray et al. (1992a) identified that rainfall along the Gulf of Guinea and in the Sahel itself provides a somewhat dependable indication of future Sahel rainfall (and thus Atlantic hurricane activity). The Sahel rainfall correlation to its previous year rainfall is reflected in the strong tendency for anomalies of precipitation to continue from year to year. This persistence is likely due to a combination of global sea surface temperature forcing (Lamb 1978; Folland et al. 1986) and changes in the land surfaces including desertification which may reinforce drought conditions (Nicholson 1988; Xue and Shukla 1993). The positive feedback between the Gulf of Guinea rainfall in August through November to Sahel rainfall/Atlantic hurricanes the following year appears to result from changes in available moisture for the North African monsoon through long-term storage in the soil and biosphere (Gray et al. 1992a). While the previous year Sahel rainfall can be used to forecast for only about 5% of the intense hurricane variability, the Gulf of Guinea rainfall anomalies provide a much stronger predictor of around a third of the variability hindcasted in the intense hurricane activity.
Overall, the 1 December hindcasts were able to explain about 40% to 50% of the variability of the tropical cyclone activity. Because of the tendency of overfitting of statistical regressions with large numbers of predictors relative to the number of datapoints (e.g. greater than around one to ten) in a non-cross validated approach (Elsner and Schmertmann 1994), true independent forecasts will have a substantial degradation in skill. Thus the skill estimated to be available for future independent predictions is at the level of 20-35% of the variability according to methodology described in Mielke et al. (1996). This can be compared to climatology, which provides none of the variance by definition, and to year-to-year persistence (i.e. an auto-regressive model with a one year lag), which only can account for about 5% of the variability. Fig. 9 demonstrates the observed differences in intense hurricanes for the ten hindcasts for the most active tropical cyclone seasons and the ten hindcasts for the quietest seasons. Note the very large differences in observed intense hurricane tracks indicating a substantial amount of skill present in these hindcasts. This is an impressive result considering that this forecast is issued six months before the start of the "official" hurricane season and eight months before the active portion of the hurricane season. The latter forecasts of early June and early August make substantial use out of physical parameters which affect the Atlantic hurricanes (e.g. ENSO conditions, sea level pressure anomalies, upper tropospheric zonal winds, etc) and which also have the tendency to persist from the forecast date through the peak of the season. This is not feasible for the early December forecasts with such a long lead time, especially for ENSO's upcoming state because of the difficulty in obtaining skill across the March-May "predictability barrier"2 (Wright 1985, Wright et al. 1988).
The 1 June seasonal tropical cyclone forecast incorporates elements from the 1 December forecast as well as adding in more timely information from the most recent few months (Gray et al. 1994), most importantly being an indication of ENSO's evolving state. There are 13 predictors in five groups as listed in Table 2 used in this forecast. Fig. 8 shows the locations of these various predictors. As with the 1 December forecast, three of the predictors are for extrapolating the state of the QBO expected during September - zonal winds at 50 mb, 30 mb, and the vertical shear between the two layers. Four predictors involve North African surface parameters. Two of these, the Gulf of Guinea and western Sahel rainfall, were described in the previous section. The other two North African predictors are the anomalous surface temperature and sea level pressure gradients from February through May of the current year. The remaining six predictors involve conditions over the Caribbean Sea (April to May sea level pressure anomalies and 200 mb zonal wind anomalies) and current information regarding the strength and trend of ENSO.
The new predictors include two North African surface predictors which relate to the pre-rainy season conditions over sub-Saharan North Africa. When zonal surface temperature and sea level pressure gradients during February through May are relaxed as the monsoon onset begins, the Sahel rainfall and Atlantic hurricane activity are stronger than normal. Conversely, when the surface temperature and sea level pressures have tightened gradients from the west coast to the interior, the Sahel rainfall is reduced and the Atlantic hurricane activity is quieter than normal. These surface conditions act to alter the strength of the southwesterly monsoon flow into the Sahel. Over the Caribbean, April and May sea level pressure and 200 mb zonal wind anomalies - a reliable measure of the crucial vertical wind shear variations - are utilized as predictors for the hurricane season. The pre-season sea level pressure anomalies and the 200 mb zonal winds over the Caribbean have a tendency to persist into the heart of the hurricane season and thus are useful as predictors of the hurricane activity. The last four predictors give indications of the current strength of ENSO and its trend in the previous few months: the April and May equatorial eastern Pacific SSTs and the SOI and their change between January/February to April/May. These values provide reliable indications about how ENSO will likely behave during August through October, the peak crucial Atlantic basin hurricane months.
With the use of these 13 predictors, the hindcast testing is able to anticipate between 50% and 70% of the variability by 1 June. This should degrade to 25-55% of the variability in independent real-time (operational) forecasts, demonstrating a substantial improvement over the skill levels that are suggested for our 1 December forecasts. If these atmospheric and oceanic relationships are stable, then substantial independent real-time forecast skill is available.
For the final initial time forecast of 1 August, information is utilized that extends right up to the start of the active portion of the hurricane season (Gray et al. 1993). This forecast may appear to be more of a "nowcast" than a prediction when one recalls that the "official" Atlantic hurricane season extends from June through November. However, an inspection of the seasonal variation of named storms and hurricanes reveals that only 11% and 6% of the annual named storm and hurricane activity (as measured by days in which these cyclones are present) respectively, occurs before 1 August on average (Landsea 1993). Less than 2% of the intense hurricane activity is observed on average before 1 August and 95% occurs just in the three months of August through October. Additionally, the small amount of activity that does occur in June or July has shown no predictive value for the entire season: a busy (e.g., two or three named storms) June and July can precede a very active year (such as 1990 when 14 named storms occurred) or a very quiet year (such as 1986 when only six named storms were observed). Alternatively, quiescent (e.g., with no named storms observed) June and July years can either precede very active years (such as 1988 with 12 named storms) or very quiet seasons (such as 1983 with only four named storms observed).
Nine predictors in four predictor groupings (listed in Table 2 and the locations of which are shown in Fig. 8) are used in the 1 August forecast (Gray et al. 1993); all but one of which are simply updates of predictors described earlier. The QBO measures of 50 mb and 30 mb zonal winds and the vertical shear between the two levels through July are extrapolated two months forward to September. The Caribbean Sea sea level pressure anomalies and 200 mb zonal wind anomalies are again utilized, but now updated for the months of June and July. In addition, the June and July values of SSTA and SOI are used for a current indication of ENSO's phase and strength. The Caribbean Sea and ENSO predictors are useful as a consequence of their strong tendency to persist through the remainder of the hurricane season. The previous year August through November Gulf of Guinea rainfall is utilized, but in combination with the one additional predictor - the rainfall anomaly in the western Sahel during June and July. Since the rainy season usually commences during these two months, this rainfall index provides a reliable idea of the early summer strength of the monsoon in its effect on the Sahel. Typically, the use of June and July rainfall provides a useful indication of how rainy the remaining two months of August and September of the rainy season will be (Bunting et al. 1975, Gray et al. 1994). Because of the strong concurrent correlation between Atlantic tropical cyclone activity and seasonal Sahel rainfall, this June and July western Sahel rainfall provides an excellent precursor signal for the hurricane activity from August until the end of the hurricane season, particularly for the expected intense hurricane activity. Note that these more recent rainfall measurements replace the previous year August and September western Sahel rainfall anomalies.
The skill levels based upon hindcast testing range between 45 and 60% of the variability explained by 1 August. In real-time independent forecast testing, the amount of skill likely to be available will be in the range of 25-40%. While this is an improvement over the hindcast skill available by 1 December, it is somewhat lower than what may be possible by 1 June, two months earlier. This is due to an improvement of the forecast scheme for the 1 June lead time to which (Gray et al. 1994) have allowed it to perform better than the older version (Gray et al. 1993a) of the 1 August scheme. Current work is underway to reduce the number of predictors for all of the lead times, to include the years of the early 1990s and to only select those predictors that contribute a reasonable amount of variability toward the regression equation. One particular change will to be to utilize the "Niño 3.4" region (5°N-5°S, 120-170°W) in place of both the SOI and the original ENSO SST index (Niño 3), which was farther to the east. The Niño 3.4 index has been identified as the SST region having strongest concurrent association with mid-latitude and tropical ENSO-forced circulation variations (Barnston et al. 1997).
Regardless of the exact performance of the published regression schemes in Gray et al. (1992b, 1993, 1994) in the future, there are now thirteen years of forecasts that have issued in real-time by Prof. Gray and his collaborators at Colorado State University. As in any real-time forecasting situations, the seasonal forecasts have not solely relied upon the quantitative regression results in Gray (1984b) and Gray et al. (1992a, 1993, 1994). The forecasts issued also give some weight to consistency between predictands, predictive factors not explicitly in the regression model, and forecaster intuition. Thus the forecast results presented below are the final "official" forecasts and are not strictly the regression model results. A full independent verification of LAD regression results presented in Gray et al. (1992a, 1993, 1994) will need to wait until a larger sample (at least 10 years) is available.
In most of the prior real-time forecasts from 1984-1996, predictions
have beaten climatology and persistence, which were previously the only
way to estimate future hurricane activity. Table 2 presents the real-time
(operational) seasonal forecasts for named storms, hurricanes, intense
hurricanes and hurricane days (a measure of the duration of the season)
from various starting times and their verification. The eight early June
seasonal forecasts for 1985, 1986, 1987, 1990, 1991, 1992, 1994, and 1995
were more accurate in general than climatology for both named storms and
hurricanes (1950-1990 mean value of 9.3 named storms and 5.8 hurricanes).
The forecasts for 1984, 1988 and 1996 were about as successful as climatology,
while the two seasonal forecasts for 1989 and 1993 were failures. To quantify
the amount of skill available, the agreement coefficient, r,
is utilized to compare the real-time forecasts against the observations.
Table 2 shows that the early June predictions have explained about
25% of the variability for named storms and hurricanes, significantly greater
than that available by persistence (2 and 15%, respectively) and by climatology
(0%). The early June intense hurricane forecasts have yet to show significant
skill, however, seven years is too small a test database to say anything
definitive. The early December forecasts have too small a sample size (five
years) of independent data to come to any conclusions as of yet. On the
other hand, the early August forecasts for all of the tropical cyclone
parameters show increased, significant skill over and above the early June
predictions - up to 55% of the named storm and 40% of the hurricane variability.
These values of variability explained for the early June and early August
named storm and hurricane forecasts are in the range, and even higher than
for the 1 August forecasts, of the expected independent skill discussed
earlier. Real-time forecast and verification reports for all of these forecast
dates during the last several years are now available via the
World Wide Web:http://tropical.atmos.colostate.edu/.
In addition to the contributions by Gray et al., substantial progress has also been made toward seasonal forecasting of Atlantic basin tropical cyclones including U.S. landfalling hurricanes in research led by Prof. James Elsner at Florida State University (U.S.). In Elsner and Schmertmann (1993), they utilized the predictors from Gray et al. (1992a) to derive a fully cross-validated Poisson regression model that outperforms the LAD model for intense hurricanes. In Hess et al. (1995), a subjective stratification is performed to remove those hurricanes that had a mid-latitude baroclinic influence sometime during their genesis or development to hurricane force. With these removed from the database, it was found that the predictors from Gray et al. (1992a, 1993) show a stronger relationship with the "tropical-only" hurricanes (Fig. 10), though there are concerns regarding the subjectivity inherent in removing the baroclinically-influenced hurricanes from the entire database. The baroclinically-influenced hurricanes were not found to be predictable by available predictors. Hess et al. (1995) then employ an OLS multiple linear regression to provide for tropical-only hurricanes (to which they add a climatological value of baroclinically-influenced hurricanes) forecasts at both a 1 December and 1 August initial time. Again this methodology shows a modest, but significant improvement over Gray et al. (1992a, 1994) in the hindcast dataset.
In an attempt to go from the entire Atlantic basin to more regional
scales including landfalling U.S. hurricanes, Lehmiller et al. (1997) split
portions of the Atlantic basin up into four threat regions: the U.S. Northeast
Coast (from northern North Carolina to New England), the U.S. Southeast
Coast (from eastern Florida to southern North Carolina), Gulf of Mexico,
and Caribbean Sea. For the first two regions, they considered U.S. landfalling
hurricanes and intense hurricanes and for the second two regions, they
considered hurricanes and intense hurricanes that occurred anywhere within
these water boundaries. The predictors utilized are those from Gray et
al. (1992a) for a 1 December of the previous year forecast and from Gray
et al. (1993) with some additional ones for use in a 1 August forecast.
The additional 1 August predictors (July 700-200 mb vertical shear at Miami/West
Palm Beach (U.S.), July sea level pressure in Cape Hatteras (U.S.), and
July averaged U.S. East Coast sea level pressure) assist in determining
the regional vertical shear and the steering flow strength. Similarly to
the other 1 August predictors, these are suggested to work through a persistence
of anomalous conditions through the height of the hurricane season. With
a multivariate discriminant analysis, Lehmiller et al. were able to successfully
hindcast the occurrence or non-occurrence of storms at least three-quarters
of the time (versus a climatological accuracy of nearly 50%) for the following
forecasts: 1 December Caribbean Sea hurricanes, 1 August Gulf of Mexico
intense hurricanes, 1 August Caribbean Sea intense hurricanes and 1 August
U.S. Southeast Coast hurricanes. The other hindcasts, including all efforts
for the U. S. Northeast, were unable to significantly improve upon climatology.
The predictions of regional hurricanes, as well as the basinwide measures
of Atlantic hurricane activity, issued by Elsner et al. can be found in
the issues of the Experimental Long-lead Forecast Bulletin (Barnston
The use of steering flow variations is also key for predictions of year-to-year tropical cyclone movement over regions of the Northwest Pacific (Chan 1994). In this forecasting methodology, Chan utilizes the zonal winds at both 850 and 500 mb before the season commences in a cross-validated linear OLS regression model to forecast the tropical cyclone numbers moving through designated regions. This allows for predictions of the annual number of westward moving tropical cyclones through the Philippines by early April (Table 3) and the annual number of northward moving tropical cyclones through the region just south of Japan by early March. Both methods appear to provide skill in predicting the number of tropical cyclones substantially above that of climatology. However, the physical link - presumably persistence of steering flow conditions before the season begins in March to the end of the season in late autumn - has not yet been convincingly shown.
Research into seasonal tropical cyclone forecasting in the Australian region (5-32°S, 105-165°E) began with Nicholls' (1979) report on the usage of an index of ENSO to predict the upcoming October-May cyclone season. Nicholls utilized the Darwin surface pressure (one-half of the SOI) to hindcast about 40% of the variance of activity, skill coming primarily through the early and mid-season tropical cyclones. In Nicholls (1984), this work was extended to show that local SSTs and SSTs in the equatorial eastern Pacific (a direct measure of ENSO) could also be utilized in a forecast mode (Fig. 11). Holland et al. (1988) showed that most of the local SST predictive signal was due directly to ENSO effects.
Recently, Nicholls (1992) noticed a temporal change in the behavior of the tropical cyclone numbers versus the ENSO predictors. The predicted values have been consistently too high compared to what has been observed since the season of 1986/87. The values in the correlation for independent data dropped to only about 20% of the variability. He suggests that most of this bias is artificial due to change in tropical cyclone monitoring policies. Instead, utilizing the year-to-year differences in cyclone numbers and the SOI gives a predictive association explaining about 65% of the variability. Thus with just one predictor (the SOI, or more recently, the one year change in SOI), a regression scheme is able to capture almost two-thirds of the interannual variability in Australian basin tropical cyclone activity by early December and slightly less by early September. Nicholls' real-time forecasts are posted in November of each year and can be found at the following World Wide Web site:http://www.bom.gov.au/bmrc/.
While Nicholls has utilized El Niño's effect of reducing tropical
cyclone activity in the Australian basin, Basher and Zheng (1995) use the
tendency for El Niño increases of South Pacific tropical cyclones
for forecasting purposes in that region. Again making use of an OLS multiple
regression, Basher and Zheng use SOI and local SSTs during September to
forecast the number of cyclone occurrences in 20°
longitude by 12° latitude boxes
extending across the South Pacific. Using just the September SOI, they
find that they can explain 52% of the variance for the entire region extending
from 170°E to 130°W.
The local SSTs do not contribute significantly above this value. It is
just in the region from 150°E
to 170°E where the local SSTs
are able to forecast 35% of the variance, and the SOI provides no additional
information. This is due to ENSO's influence reversing between inhibiting
to enhancing of tropical cyclogenesis (for El Niño events) around
the longitude of 160°E.
In section II, discussion centered on the need for favorable synoptic-scale variations of the environment to allow genesis and development of tropical cyclones to proceed. Since the El Niño-Southern Oscillation (ENSO) produces such large, widescale changes in the tropical circulation, perhaps it is not surprising that tropical cyclones are strongly altered by ENSO. However, these changes are not uniform throughout the global tropics. In some regions, an El Niño event would bring increases in tropical cyclone formation (e.g. the South Pacific and the North Pacific between 140°W to 160°E) while others see decreases (e.g. the North Atlantic, the Northwest Pacific west of 160°E, the Australian region). Las Niñas typically bring opposite conditions. These alterations in tropical cyclone activity are due to a variety of ENSO effects: by modulating the intensity of the local monsoon trough, by repositioning the location of the a monsoon trough and by altering the tropospheric vertical shear. Thus based on the global nature of ENSO alone, assumptions of independence between different tropical cyclone basins would be incorrect.
In addition to ENSO, three basins (the Atlantic, Southwest Indian and Northwest Pacific) show systematic alterations of tropical cyclone frequency by the stratospheric Quasi-Biennial Oscillation (QBO). This intuitively is unexpected given that tropical cyclones are primarily a tropospheric phenomenon, but may be due to alterations in the static stability and dynamics near the tropopause. Certainly more research is needed to provide a thorough explanation of these relationships. However, given the robustness of these alterations in tropical cyclone activity that match the QBO phases, it appears unlikely that the association is purely a chance correlation.
Interannual tropical cyclone variations have also been linked to more localized, basin-specific features such as sea surface temperatures (SST), monsoon strength and rainfall, sea level pressures and tropospheric vertical changes. These regional factors can be as large as the forcing due to ENSO, though most are not. Together with ENSO and the QBO, these factors produce changes in the frequency, intensity, formation region and track of tropical cyclones in all basins. However, understanding how tropical cyclone variability relates to the surrounding environmental conditions is hampered by having only a few decades worth of reliable tropical cyclone records. The emerging field of "paleotempestology" - the study of pre-historic tropical cyclones (e.g. Liu and Fearn 1993, Keen and Slingerland 1993) - may be able in coming years to assist in analyzing how and why tropical cyclones change from year to year.
Despite the current lack of accurate long-term tropical cyclone records, some of the tropical cyclone-environmental associations have led to methodologies in seasonal forecasting by the onset of the tropical cyclone season. Given that the typical lifecycle for a particular phase of ENSO of a year or greater, one can utilize this for the basis of seasonal predictions with lead times of up to several months. These are now being done (or can be performed) for the Atlantic basin tropical cyclones, those near Australia and in the South Pacific basin. These forecasts assume no knowledge of the future state of ENSO, except that the current ENSO phase will persist for at least the next few months in the future.
Over the last fifteen years, seasonal forecasting in various basins has evolved to the point that up to 50% (or more) of the tropical cyclone variability can be predicted at the start of the cyclone season. In addition to the frequency and intensity of the all basin storms, statistical methodologies have also introduced to forecast track frequency and the likelihood of hurricane landfall in certain coastal zones. Currently, such statistical models are the only feasible methodology for seasonal tropical cyclone forecasting, because the lack of skill in global circulation models (GCMs). Because of the complexity of difficulties faced in utilizing these numerical models, it may take a decade or two of concerted research effort before such GCM forecasting is feasible, if ever. However, for the current time, creative use of statistical regression schemes can provide and will continue to provide skilled and useful predictions of tropical cyclone activity around the world.
A sensible question would be how can seasonal forecasts of tropical cyclones be used when they are for large geographic regions such as the entire North Atlantic Ocean, Caribbean Sea and Gulf of Mexico. There are a number of reasons for issuing such predictions. Practically, most people in the general public cannot - and should not - utilize the forecast directly. As an example, it would be foolish if John Q. Public decided to ignore hurricane preparedness and mitigation plans because the year was one predicted to be below average. 1992 serves as an excellent warning against such actions: the Atlantic hurricane season was very successfully forecasted by Prof. Gray (see Table 2) as a quiet year with only four hurricanes, yet one of those was Hurricane Andrew - the most destructive U.S. hurricane on record (Mayfield et al. 1994). Strong wording should be added to any seasonal hurricane forecast to discourage an individual from using the forecast in such a way.
Corporations and governments, however, because of their size and scope of operations, are starting to make reasonable use of such forecasts each year. One large, private manufacturing company with interests all along the U.S. coastline serves as an example of the many utilizations that have been made with these predictions: decisions on the amount of "hurricane" liability insurance that covers preparations, damages and repair costs; determinations of annual budgets and preventative maintenance schedules; an aid in the production and inventory storing planning; plans for data processing disaster recovery; and schedules for workers' shifts in the upcoming year. Such usage can be better enhanced by increases in skill, by providing longer-range accurate predictions, and by regionalizing the forecast for smaller locales (such as being pursued by Lehmiller et al. ).
Another justification for such forecasting efforts is that it leads to further advances in our understanding of linkages in the climate system. It was the failure of the 1989 seasonal Atlantic hurricane forecast (Table 2) that led to the discovery that the West African monsoon is intimately tied to the occurrence of Atlantic hurricanes on an interannual basis. Successes in seasonal tropical cyclone activity may also provide insight into forecasts of other tropical phenomena such as droughts/flooding associated with anomalous changes in the strength and location of the ITCZ and the monsoons (e.g. Hastenrath 1995).
Finally, the seasonal tropical cyclone forecasts at times generate public
and media interest, whether or not individuals can actually make use of
the predictions. A beneficial side effect of such interest is that it heightens
the awareness of the public to the danger of hurricanes and hopefully prompts
more people to take precautions and make preparations.
2 Some forecasting groups have claimed ENSO predictive skill through the spring season (e.g. Chen et al. 1995, Penland and Sardeshmukh 1995) primarily through hindcast runs on dependent data. However, independent tests show that no ENSO model exists - statistical or numerical - that exhibits skill in real-time predictions relative to a simple ENSO climatology and persistence (ENSO-CLIPER) model (Knaff and Landsea 1997). Thus until there are available truly skillful ENSO models for lead times of several seasons, improvements in forecasts issued in early December for Atlantic hurricanes the following year may be slow to occur.