American Scientist

Over the past two decades hurricane forecasters have made steady and significant improvements in predicting North Atlantic cyclone tracks. Errors have been reduced by at least half out to 72 hours before landfall, and only since 2001 have predictions even been made 96 hours and 120 hours out.

Unfortunately, the same cannot be said for predictions of hurricane intensity, which have improved little if at all over the same period. Tracking forecasts have the advantage of direct observations from aircraft and satellites, allowing the location of the center to be identified fairly accurately. Intensity forecasting, on the other hand, suffers from the difficulty of measuring, let alone forecasting, wind speed, combined with the inability of models to capture detailed air-sea interactions and cloud-rain processes.

Recently, however, an international team has presented data that may lead to a significant step forward in intensity forecasts. Leo H. Holthuijsen of the Delft University of Technology, along with Mark D. Powell of the National Oceanographic and Atmospheric Administration and Julie D. Pietrzak (also at Delft) published a paper in September in the Journal of Geophysical Research titled “Wind and Waves in Extreme Hurricanes.” In it they explore several factors that have not previously been taken into account in intensity models.

The most important of these will probably prove to be refinement of the influence of wind speed on the drag coefficient (C_D) between the atmosphere and the sea surface. Most wave models to date have assumed that C_D rises linearly with wind speed to 25 to 35 meters per second 10 meters above the water surface (the standard altitude in the following, unless otherwise noted), above which it either remains constant or continues to increase. Holthuijsen, his coauthors and some previous investigators dispute this assumption. This matters because hurricane intensity is a function of the amount of energy that can be transferred from water to air (called the enthalpy coefficient) divided by the surface drag. Below a certain value, a hurricane is not sustainable.

The Delft and NOAA scientists analyzed both photographs of the sea surface taken from reconnaissance aircraft and 1,149 wind profiles from GPS drop sondes. (The photographs, taken mostly in the 1970s, were made 100–1,500 meters above the water and are now a precious resource, as such low-altitude flights were deemed too dangerous by the early 1980s and were discontinued.) The drop sonde measurements were used to compare wind speeds at 20 and 160 meters, which provided a way to calculate C_D. Photographs were scanned and processed to look at the transition from white caps (breaking wave crests) to a foam, spray, bubble emulsion layer organized in long filaments called “streaks.”

As Mark Powell, the team’s meteorologist, relates, “The existence of a foam layer at high wind speeds has been known for ages by sailors, of course. In a 2003 paper my colleagues and I speculated that a foam layer was contributing to a slip layer and concomitant reduction in the drag coefficient.… However, now we have more data to substantiate the reduction.… Existing theories and experiments do not predict the drag coefficient dropping to nearly zero as surface wind speeds approach 60 meters per second, nor do they predict the formation of a high-velocity near-surface jet at still higher wind speeds (for which we cannot estimate a drag coefficient).”

The photographs also revealed differences in how white caps and streaks formed in different sectors of the windfield. In the area just right of and ahead of the center’s direction of travel, streaks formed at comparatively low speed, whereas streak formation was delayed to the left and in front of the direction of travel. This is because wind-driven waves interact with swell generated at some distance. According to Holthuijsen, “The onset [of streaks] occurs at about 25–30 meters per second for wind-driven waves with or without following or opposing swell. It occurs at about 35–40 meters per second in the presence of cross swell.”

Including a more nuanced understanding of how the drag coefficient is influenced by wind speed and sea state should make it possible to improve intensity forecasts. Powell concludes, “Right now none of the hurricane prediction models really consider that the C_D should depend not just on the wind speed but also on the sea state, and that the sea state varies depending on the wave and current behavior, which can vary substantially around the storm. Our results are a wake-up call to the modelers that we really need to consider this as a coupled problem with interacting atmospheric, wave and ocean models.”