FORECASTING TROPICAL CYCLONE INTENSITY CHANGES:
AN OPERATIONAL CHALLENGE.
Lixion A. Avila, OAR, Miami Florida
1. INTRODUCTION
The problem of forecasting tropical cyclone intensity change continues to be a real challenge for
tropical meteorologists despite the recent advances in numerical weather prediction. The
OAR (NHC) 24 and 72 hour official forecast intensity errors for the Atlantic
basin since 1990 are shown in
Fig.1. An inspection of the figure clearly shows that no significant
improvement has been made during that period and, in general, the average errors are about 10
and 20 knots for 24 and 72 hour period respectively. Overall, these errors appear to be small and
indeed they are. However, the real challenge is to forecast those cases of rapid intensification or
decay, particularly if they occur just prior to or at
landfall.
Fig. 2
Hurricane Andrew in August 1992 intensified significantly twice. First, when it was approaching the Bahamas and then again just before making landfall over South Florida. Hurricane Opal in October 1995 jumped from a Category 3 to 5 on Saffir-Simpson hurricane scale (Saffir and Simpson, 1974) in a matter of a few hours. Fortunately, Opal's rapid intensification occurred in the central Gulf of Mexico and not near the coast Hurricane Bertha in July 1996 suddenly strengthened at landfall near Wilmington N.C., after a period of weakening. It is an important goal to produce the best possible intensity forecast at landfall. Actions taken by the emergency management community, mainly with respect to the different levels of evacuations along the coast, are closely tied to the forecasts. Extratropical transformation is another difficult problem related to intensity changes and in general, little is known about it. Hurricane Erika in September 1997 is a typical example. Erika was already moving northeastward embedded within a mid- latitude trough and was forecast to become extratropical in a period of 36 to 48 hours. Instead, the hurricane kept its warm core longer than anticipated and passed by the Azores producing hurricane force wind gusts and heavy rain in some of these islands.
2. DISCUSSION
2.1 Diagnosis of the cyclone.
One of the primary tasks of the forecaster is to be able to detect the genesis of a tropical cyclone and the potential for intensification, dissipation and extratropical transformation. Once it has been determined that a tropical cyclone has is already formed, the forecaster is required to determine its maximum intensity, which is defined as the maximum 1-min sustained wind at the 10-m level. In spite of advanced satellites, aircraft and other new technology, the measurement of that parameter is highly uncertain. Even in rare cases, when a tropical cyclone moves over a wind gage, in general, there are problems with the wind averaging period, as well as the exposure, elevation, survival and calibration of the instrument. Assuming that a reconnaissance aircraft has sampled the area of maximum winds, the flight-level wind data has to be adjusted to 10-m level. Powell et. al. (1996) have worked extensively on the wind adjustment to the surface. However, there is no unique formula or methodology available for the adjustment. Data from the new Global Positioning System (GPS) sondes dropped within the inner core of the hurricane will shed new light on the problem.
In most of the world's basins, the intensity of tropical cyclones is estimated from satellite using the Dvorak (1984) technique. This method is primarily based on cloud pattern recognition and is highly subjective.
2.2 Forecasting the cyclone and its environment.
Once the forecaster has estimated the initial state of the tropical cyclone and its environment, in general, there are only a few parameters which the forecaster uses to determine the potential for intensity changes. Most common are sea surface temperature (sst), variation in surface pressures, and vertical shear of the horizontal wind. However, many of these parameters are in general difficult to measure due to the lack of observations. Upper-tropospheric trough interactions, which can be traced back to Riehl (1954), are features which are also taken into consideration during the preparation of the intensity forecast. This theory is once again becoming rejuvenated within the hurricane community (See Molinari and Vollaro, 1989). The problem here is that it is hard to determine subjectively and operationally when the trough interaction will produce a favorable or unfavorable environment for the tropical cyclone to intensify. However, there is hope with the recent improvements in numerical modeling.
Forecasters can also infer short-range changes in intensity from the cloud pattern observed in satellite images or radar. For example, the shear pattern, which is a general sign of steady state intensity or weakening, is determined by the location of the low-level cloud circulation with respect to the deep convection. Eyewall replacement cycles are commonly associated with intensity fluctuations as indicated by Willoughby et al, 1982. It is difficult to detect such small scale eyewall cycles unless there is a constant radar or aircraft surveillance. Moreover, the time scale of these cycles appears to vary from one hurricane to another. For example, the time period in between eyewall replacements during Hurricane Allen in 1980 was a day or two while in Hurricane Andrew it was less than 12 hours.
There are numerous models available to forecast the track of tropical cyclones. However, only a few models are operationally available to address the intensity problem. For intensity up to 72 hours, the SHIFOR model, which is based on climatology, is the most commonly used. SHIPS model predicts intensity up to 72 hours and uses climatology and persistence as well. It also includes predictors such as the sst, the vertical shear of the horizontal wind, the 200 mb eddy flux convergence of relative angular momentum and upper-tropospheric temperature. Details of this model can be found in DeMaria and Kaplan, 1994.
Several Global models and the Geophysical Fluid Dynamic Laboratory (GFDL) model are also used as guidance to predict changes in intensity associated with tropical cyclones. In general, the GFDL model overestimates the intensity of tropical cyclones and global models, in most cases, do not represent correctly the tropical cyclone in its initial stage. At least in the Atlantic, the models tend to underestimate the intensity of the mid-oceanic upper-level troughs. Instead, the models incorrectly forecast an anticyclonic upper-level and reduced environmental shear (in comparison to the verification) which appears to be favorable for tropical cyclones to intensify.
3. CONCLUSIONS
1. An accurate intensity forecast is extremely important in the warning process, primarily because emergency management decision-making is closely tied to the intensity of landfalling tropical cyclones.
2. It is difficult to determine the initial intensity of a tropical cyclone due to the lack of reliable observations.
3. Very little guidance is currently available to forecast intensity changes. Extrapolation is basically the most commonly used rule.
4. The process that lead to changes in intensity of a tropical cyclone appears to be closely related to changes in structure, primarily of the inner core. Most of the time, there is no information available. Even when there is information, it has not been translated into forecasting tools.
5. It is recommended that the development of intensity forecast tools be elevated to a high research priority.
ACKNOWLEDGMENTS: The author is grateful to Dr. Richard Pasch, hurricane specialist at the NHC and Dr. Mark DeMaria for many valuable discussions.
4. REFERENCES
DeMaria, M., and J. Kaplan, 1994: A statistical hurricane intensity prediction scheme (SHIPS) for the Atlantic basin. Wea. Forecasting, 9, 209-220.
Dvorak, V. F., 1984: tropical cyclone intensity analysis using satellite data. NOAA Tech. Memo. NES 11, 47 pp.
Molinari, J. And D. Vollaro, 1989: External influences of Hurricane Intensity. Part I: outflow Layer Eddy Angular Momentum Fluxes. J. of Atmos. Sc., 46, 1093-1105.
Powell, M.D., S. H. Houston, and T. A. Reinhold, 1996: Hurricane Andrew's landfall in south Florida. Part I: standardizing measurements for documentation of surface wind fields. Wea. Forecasting, 11, 304-328.
Riehl, H., 1954: Tropical Meteorology, Mc Graw-Hill, 392pp.
Saffir, H., and R. Simpson, 1974: The hurricane disaster potential scale. Weatherwise, August 1974, 169-170.
Willoughby, H. E., J. A., Clos and M.G. Shorebah, 1982: Concentric Eye Walls, Secondary Wind Maxima, and the Evolution of a Hurricane Vortex. J. Appl. Sci., 39, 505-514.