The weather at the junction of the Tigris and Euphrates Rivers is certainly different from the weather at the junction of the Schuylkill and Delaware.

But the perils of weather forecasting are not.

The same flaws in computer models and the worldwide observation network, combined with nature's wayward behavior, make forecasting for Baghdad and the vicinity as problematic as forecasting for Philadelphia and vicinity.

The fundamental difference is that in Baghdad and the vicinity, a blown forecast can be a matter of life and death.

"The way the atmosphere behaves here isn't any different from how it behaves halfway across the world," said Todd Miner, a meteorology professor at Pennsylvania State University.

And despite the extreme efforts of combat weather forecasters, in all likelihood nature will stage an ambush or two if the war is prolonged.

It hasn't happened with the storm now raging across Iraq - at least according to the U.S. Air Force. Although the weather yesterday slowed troops and grounded scores of aircraft, the Air Force said its battlefield forecasts had been right on target throughout the war zone.

However, that contention isn't easily verified: No surface-weather observations have been available publicly from Iraq since 1980, one of the many regions of the world from which weather data are scant. To get readings and live, short-term forecasts, an undisclosed number of combat-trained meteorologists are on the ground, packing weather instruments along with their rifles.

Their direct observations and short-term forecasts from the field are critical; otherwise, the military would rely almost solely on computer models used by forecasters around the world, which are far from infallible.

The military is using them, but only in an advisory capacity, said Air Force Lt. Col. Tom Frooninckx, who is in charge of wartime forecasting operations.

Computer models have revolutionized forecasting - and created a whole new set of expectations and problems. They do a great job of seeing a storm coming next week but might have a heck of a time once it arrives.

"The models are the models. They're going to be wrong," said Bernie Rayno, a Mideast specialist at AccuWeather Inc.

Government and commercial meteorologists consult short-term and mid-range models every day. The main ones are generated by Canada, the United States, the United Kingdom and the European Union. They all have their own takes on the weather - sometimes slightly, sometimes radically, different from one another.

In the case of the Iraq storm, "various computer models came up with various guidance and forecasts," said Frooninckx, who heads a team of 50 forecasters at South Carolina's Shaw Air Force Base. In the end, he left it to human beings to make the calls. "The models are not the final say," he said.

The models can be marvels: The general features of the Iraq storm were seen at least a week ago, Rayno said. But some meteorologists complain that all the options actually have made forecasting harder.

"Do you think the weather forecasts are any better now than they were 10 years ago?" asked Fred House, a physics professor at Drexel University. "I don't think they're as good as they used to be."

The models depend on the quality of observations fed into them. Those observations give information on the state of the atmosphere, and using the laws of physics, the computers try to figure out what the atmosphere will look like in six-hour increments on out to several days, all over the world.

But they have big problems right from the start. There are huge gaps in the observation network - over oceans, over less-developed nations, over war-torn countries such as Iraq.

"Our technology has actually exceeded our ability to measure the initial state of the atmosphere," said Pete Manousos, of the government's Hydrological Prediction Center.

Even if the observations were perfect, the models wouldn't be. No model captures all the physical processes at work. If they could, the calculations would have to be carried out to infinite decimal places to be accurate, and computers aren't powerful enough.

Overlooking the limitations, some forecasters have gone awry by mistaking the model for the thing itself, House said.

"Models are advisory, but because of the graphics and the beautiful shows-and-tells, models become truth. Nothing replaces plotting and analyzing your own weather map," he said.

"There's a general overreliance on computer models," said AccuWeather's Rayno. "If you could trust a particular model, you wouldn't need forecasters."