Today, Hurricane Irene is heading for the southeast coast of the U.S., and the current predictions of the National Hurricane Center suggest that the storm will make landfall either on the Outer Banks of North Carolina or New England (or both). The graphic below shows the NWS track forecast. You notice they use a cone shape that increases in width in time, consistent with the loss of track accuracy in time.
For three years the U.S. mainland has been spared a direct hit and we have become complacent. When I was back in D.C. for a meeting at the NWS National Centers for Environmental Prediction there was lots of talk about massive cuts to the hurricane research budget.
Perhaps, things are about to change.
The interesting thing about hurricane forecasts is that we have gotten much better in some ways (predicting the track of the storm) and have barely improved in others (forecasting the intensity of the storm).
Want to see some sobering figures? Here are the changes of errors over the past several decades for hurricane track and intensity. A steady and impressive improvement in track forecasts. For 48h forecasts the errors have gone from 300 nautical miles in 1970 to 80 nm today. Huge improvement. But intensity forecasts have hardly changed--even gotten worse at 24h.
We understand why this discrepancy occurs. During the past decades we have gotten much more observational data AROUND hurricanes from satellites and other sources and our numerical models have gotten better. If you have a better handle on the environment of hurricanes you can secure a much better prediction of their track, because they are steered by the larger scale flow. But to predict changes in their intensity, you have to know the details of their innards; you have to be able to specify their internal structure and have computer models with the very high resolution needed to get the fine-scale structure (eye wall, rainbands) correct. Historically we have had neither. Thus, our ability to forecast intensity is poor.
But the situation could be changed for the better. Today, we have the computer power and models capable of doing the job--we are talking about running our computer models with grid spacings of 1-2 km (today we use 12 km). Here is an example of what we are capable of--a simulation at 1.3 km grid spacing:
One approach is to send manned aircraft, like the NOAA P3, with sophisticated radars and dropsondes (weather instruments that are dropped into hurricanes.) This is good, but it is very expensive and we can't afford to have enough planes out there at all times to do the job.
A better idea would be to use unmanned aircraft to do the reconnaissance. Today NOAA and NASA are experimenting with the use of Global Hawk drones that could fly high above hurricanes with downward-looking radars and the ability to release dropsondes into hurricanes.
A local Washington company may have the solution: moderately expensive (tens of thousands of dollars) unmanned aircraft that could get the critical observations. The company- Aerovel, located in the Columbia Gorge, is led by a brilliant Stanford-trained engineer named Tad McGeer. He designed another unmanned aircraft that was the first to cross the Atlantic (I know about it, because I was the meteorologist for the mission. The aircraft, Laima, is now in our Museum of Flight). Aerovel has developed an aircraft that can take off like a helicopter, transition into level flight, fly thousands of km, return to base and land itself.
Don't believe me? Check out a video of one of their recent test flights (click on image or link):
Imagine a swarm of these unmanned aircraft densely sampling the lower structures of hurricanes, while the Global Hawks are observing from above. The information would be sent back in real time to the National Weather Service, where high-resolution models use the information to make highly skillful hurricane predictions...far more skillful than we have today.
A dream? No. It is quite possible. But it will take several million dollars to test this out. That is where we need someone or some group with sufficient resources to provide the seed money that will give this idea a real-life evaluation.
