On Wednesday last week, Jonathan Moskaitis (senior-meteorology) and his roommate biked four miles to a small valley outside State College to check a thermometer.

The two heard the area known as "Barren Valley" was the coldest spot in the state, so they decided to investigate by setting up a special thermometer that measures the minimum and maximum temperatures reached in an area in the valley. They plan to check it monthly, hoping to get the state record-low temperature.

"It is going to be hard to check it in the winter," Moskaitis said. "We might have to ski or something."

Putting thermometers in remote valleys is just one of Moskaitis's meteorological pastimes. He described forecasting as a hobby and said that every winter he and his friends each guess what the seasonal forecast will be, although no money is involved.

"It's all about pride," Moskaitis said.

In January of his sophomore year, Moskaitis began working with Craig Bishop, associate professor of meteorology, doing research for the thesis he is required to complete as a student in Schreyer Honors College.

"I taught him dynamics in the 422 class (Meteo 422-Dynamic Meteorology II) and was very impressed by his work," Bishop said.

Moskaitis became involved in a project aimed at improving forecasters' predictions of winter storms that Bishop started in 1996 in coordination with the National Centers for Environmental Prediction in Washington, D.C.

"He was helping us put into operation a program that would reduce errors in forecasts of winter storms," Bishop said.

Bishop, along with Brian Etherton, graduate assistant in the department of meteorology and Sharanya Majumdar, a former Penn State researcher, developed a computer program called the Ensemble Transform Kalman Filter that enables storm trackers to better collect data over the Pacific Ocean where winter storms originate.

"The idea is to improve forecast over the United States by taking observations two, three, four days in advance, and then put that information in forecast models," Majumdar said.

In its experimental stage the ensemble program, which came to be known as Winter Storms Reconnaissance in 1999, would compile data from various forecasts taken over the Pacific Ocean. It would then pick out regions for "targeted observation" where weather planes flying flight tracks from Anchorage, Alaska, and Honolulu should fly to make the most accurate predictions.

Bishop said the National Weather Service found the computer program's observations were helping forecasts and decided to put more money into its development. Bishop invited Moskaitis to help with that development in January 2000. He began learning about the program by observing Majumdar run code for identifying target areas from which weather planes collect data.

After observing Majumdar, Moskaitis spent the summer at NCEP in Washington, D.C., working on making Bishop's program user friendly.

"His main contribution is making the code operational," Majumdar said.

Before Winter Storms Reconnaissance, the process used to determine appropriate flight tracks for weather planes was much more subjective, Majumdar said. Researchers at Penn State would corroborate with researchers at NCEP to find "interesting weather events" then based on those discussions would call flight planning to tell the planes where to take their observations.

"What Jon has done has directly affected the successfulness of forecasting," Majumdar said.

With Winter Storm Reconnaissance, a computer determines the areas for targeted observation and suggests flight tracks. Moskaitis expressed some surprise at playing such a large role in helping NCEP operate the program without Penn State's involvement.

"I had no idea I'd be spending my summer doing this stuff," Moskaitis said.

Majumdar and Bishop both expressed how pleased they were with Moskaitis's contributions.

"It shows that an undergraduate can make a contribution in the real world," Majumdar said.

Last summer, Moskaitis was paid by Penn State to go to the University of Miami and work with Majumdar to see how the predictions improved from the Winter 2000 to Winter 2001.

"We want a dependable correlation between what we said would happen and what actually did happen," Moskaitis said. "In two-thirds of the cases, the forecast improved."

At the moment, Moskaitis is deciding where he wants to apply for graduate school, although he has no idea exactly where he will end up.

"I want to go somewhere far away," Moskaitis said.

He is interested in the challenge of meteorological research and hopes to travel in the future.