Hurricane tracking technology matures with the help of unsung heroes

Except for Ernesto, hurricane season has been mostly quiet so far this year. There are a couple brewing right now, but they aren't expected to pose much of a threat.

It's interesting how tracking and prediction technology has advanced in the past few years, and how it enables experts to better assess threats. And here's a fascinating article about the technology:

It was the night of Aug. 26, and the National Hurricane Center had some grim news for Mississippi, Alabama and Florida's western Panhandle: Tropical Storm Ernesto appeared to be heading their way.

But a series of squiggles on a computer screen was predicting a different path: straight up South Florida's spine.

That electronic dissent came from a massive computer program known as the Global Forecast System, run by a little-known federal agency and housed in an IBM supercomputer northwest of Washington, D.C. The program, known as the GFS, is part of a growing arsenal of cyber-forecasts that meteorologists at the hurricane center and elsewhere consult when predicting when the next storm will arise and where it will go.

And this time, the GFS happened to be right - a fact soon confirmed by other weather-predicting programs, the hurricane center's subsequent forecasts and Ernesto's eventual path toward its landfall Wednesday near Cape Sable.

The article has some interesting background on the evolution of hurricane tracking, from a proposal in the 20s for a "forecast factory" manned by 64,000 number-crunching mathematicians to today's GFS system which can perform 1.5 trillion calculations per second.

Perhaps the greatest advancement, though, is the ability to integrate global atmospheric data:

The most sophisticated models from the National Weather Service, the U.S. Navy and agencies in Europe amount to swirling, three-dimensional simulations of the Earth's atmosphere, based on millions of satellite measurements of temperature, air pressure, moisture and wind movements at points around the globe

They're a far cry from the pre-computer, pre-satellite era, when hurricane forecasts could be wrong by more than 1,000 miles. [..] Last year, the hurricane center's five-day forecasts were more accurate than its three-day forecasts were in 1990.

Despite these advances, the models aren't perfect and human judgment plays a big part in forecasts and advisories:

The models are especially poor in forecasting a storm's strength, as shown by the way Ernesto fizzled instead of intensifying as it crossed Cuba and entered the Florida Straits.

Models do much better in predicting the storms' paths. But the hurricane center says the models are still not as accurate on average as its own staff's forecasts, which combine the model results with the meteorologists' knowledge of the programs' biases.

Of course, anything involving geek-sexy technology on a scale of this magnitude is going to attract the interest of geeky amateurs. The article and sidebar list numerous sources of raw data, but experts warn that the data has to be taken as a whole and analyzed by trained experts before making decisions on it.

I'll admit to being one of those geeky amateurs when we lived in Florida. There were several PC based hurricane tracking shareware programs available. Some were simple tracking programs that generated maps, others incorporated fairly sophisticated (for the times) forecasting algorithms.

The one I settled on had built-in tracks of 800 storms going back to 1886. Its forecasting model used knowledge of these past storms to predict the track and intensity of approaching storms. It had built-in tracking locations or you could enter your own latitude and longitude, and it would forecast when the storm would arrive at that location and at what intensity. It was remarkably accurate, especially as the storm got closer.

Nowadays, this type of information, complete with impressive maps and charts, is only a click away on the internet. But back then (circa 1997) it was a little harder to come by. As the storms approached, we anxiously awaited TV news or weather radio updates with the latest readings of the storm's latitude/longitude, pressure, maximum sustained winds, and its direction and speed of movement. You entered this in to the program and got a new 12 hour forecast and a longer range "cone" map of probability.

Eventually, NOAA and the National Hurricane Center started posting real-time "dropsponde" and "vortex" reports on their reconnaissance website. I wrote a program to import these readings into the tracking software, automating the data collection and entry.

In the process, I learned a lot about NOAA's aircraft reconnaissance operations. Data for these reports are generated by measuring devices dropped into the eye of the hurricane from aircraft manned by courageous (or crazy?) pilots and meteorologists.

As sophisticated as the software and satellite data collection has become, someone still has to fly into the eye of the storm to get the raw data used by forecasters and their software models. These pilots and meteorological aviators are the real heroes of hurricane tracking, and their work has saved countless lives. It's taxpayer money well spent.

Now if FEMA would just pay attention to them...