The forecast: more precise forecasts
Vincent Larson and his research group aim to sharpen weather forcasting models clouded by low resolution.
If you've ever seen TV meteorologists braving the harsh elements from a "weather deck," don't be fooled into thinking that's where they do their work. The foundation of forecasting weather today lies in mathematics.
"The computer codes that are used to forecast weather or climate are tens of thousands of lines long," says Vincent Larson, assistant professor of atmospheric sciences at UWM. "They all encode mathematical equations. So the subject is naturally and inherently mathematical."
Mathematical models maintained by such groups as the National Weather Service and the European Center for Medium Range Weather Forecasting use current data such as temperature, moisture, wind, and air pressure to predict weather in the future.
Then, human forecasters study the model's output, comparing its prediction of current weather to actual conditions.
"If the numerical forecast is already in error to some extent, then the human forecaster will modify the forecast for tomorrow and the next day based on the deviations that have already occurred," Larson says.
A major limitation of any model is resolution. "A numerical model might have grid points that are 100 kilometers apart," Larson explains. "Even with the most powerful computers in the world, we can't resolve all the details of the atmosphere. We can't represent all the processes that are occurring at all the locations of interest."
That's where Larson's work comes in. He's working to improve models by studying important contributors to weather that often fall in between the grid points: clouds.
He and his research group—a postdoctoral fellow and graduate and undergraduate students—have developed a cloud parameterization. "It's a way of estimating the effects of clouds, particularly in the boundary layer—the lowest kilometer or two of the atmosphere."
"The boundary layer is where we live," Larson continues. "It's where pollution is emitted. It's where we feel the effects of weather, so representing the boundary layer is of practical interest. But also, big storms often are triggered, or they develop after small clouds in the boundary layer have evolved and grown to some degree. So forecasting when a storm will erupt or when severe weather will erupt depends often times on how rapidly the small clouds have developed. It's an important precursor to severe weather."
The Larson group's cloud parameterization computes atmospheric vertical transport of heat and moisture.
Larson credits the parameterization's mathematical algorithm for its unique capabilities. "Our parameterization estimates the full probability of encountering a particular value of liquid water and updraft speed at a given location and time," Larson wrote in a successful grant application earlier this year, "whereas other parameterizations only estimate limited information, such as fractional cloudiness."
Larson hopes to have his model integrated into the Weather Research & Forecasting Model, the next-generation model used by the National Weather Service.
Indicators of future climate?
Clouds may also hold answers to questions of future climate, Larson says, citing an estimate from the 1980s that a 4 percent increase in the area covered by stratocumulus clouds would offset the expected effects of a doubling of carbon dioxide in the atmosphere.
"In a future climate, it may be that clouds look slightly different than they do today," Larson says. "There might be more of them, they might be higher, there may be more or less liquid. It's very hard to predict those effects. But they're thought to be very important to the average temperature we experience. So one of the most difficult but important questions is how will clouds look and behave in a future climate and how will those clouds affect a future climate."
Larson is also working with a collaborator at the Geophysical Fluid Dynamics Laboratory at Princeton University to integrate his parameterization into one of the three major U.S. models that predict climate change.
"I think we have no really rigorous, reliable, solid way of estimating the errors in future climate forecasts," he says, "in large part because we don't understand how clouds behave or how they will behave in a future climate.
"So therefore we can make forecasts—there have been lots of simulations of future climates—but we really have no quantitative way of saying what the probability is that those forecasts will actually occur."
Members of Larson's Cloud Research Group are research interns Michael Falk and David Schanen, graduate students Kurt Kotenberg and Adam Smith, and undergraduate Brian Griffin. In 2006 Falk won the Tournament of Champions on the TV game show Jeopardy.