Hurricane Prediction: Progress and Problem Areas

Science  07 Sep 1973:
Vol. 181, Issue 4103, pp. 899-907
DOI: 10.1126/science.181.4103.899


The subjective approach to hurricane prediction of yesteryear has given way to the automation of many facets of prediction, leaving the forecaster with more time to reason dynamically in making the final adjustments to the official track derived from a hierarchy of computer-generated guidances. The most important task immediately ahead is to develop and apply better diagnostic procedures to help the forecaster identify the guidance sources which are most likely to contain the truth, and to help assure that his final decision will either succeed or fail safely.

The ultimate prediction method for both hurricane development and movement will probably need to simulate not only the circulation of the hurricane but the interaction of the cloud clusters in its vortex with the largescale environment. Assuming that this can be effectively done, the most important further handicap will be the problem of describing the initial conditions.

The development of a model which will dependably simulate the behavior of the three hurricanes in ten that turn out to be bad actors will not likely become operational in the near future. The most troubling problem in applying such a model—and it would have to be tested for years over a range of many storm types and conditions before becoming fully operational—is that the more complex the model is the more inscrutable the results are and the more difficult it becomes to identify the condition in which computational failure is probable. It would appear, therefore, that engineering approache to prediction and diagnostic procedures of evaluation will dominate the methodology of hurricane forecasting for a number of years to come.