But so much misinterpretation and misunderstanding surround climate forecasts that many managers are reluctant to use them.

University of Arizona hydrologists hope that will change when their new software package goes online later this summer. The software evaluates climate forecasts and is designed to give users more confidence in temperature and rainfall predictions by tailoring forecasts to each user's needs and assessing a prediction's reliability.

Land managers may come to think of the software as a personalized Consumer Reports-style evaluation of long-range forecasts.

"Itıs a web page where, among other things, you enter information about your decision-making calendar," says hydrology Ph.D. candidate Thomas Pagano. "For instance, an Arizona water manager would start making decisions in September about precipitation that's going to fall from November to April."

If heavy precipitation is forecast, the manager will want to release water from reservoirs to prevent spring flooding. But what if the forecast is wrong? That could lead to a water shortage and heavy demands on groundwater pumping.

Pagano and Hydrology Associate Research Scientist Holly Hartmann say their software will improve user confidence by ensuring that the forecasts are easy to understand and that people can reliably interpret them.

Hartmann is the lead author on a paper about their software and website that is the cover story in this month's issue of the Bulletin of the American Meteorological Society.

"One of the major reasons that natural resource managers don't use climate forecasts is that they don't know how good they are," Pagano says.

Seasonal forecasts are based on data from satellites, computer models and ocean buoys. Climate forecasters use this data to study the interactions between atmospheric pressures and ocean temperatures and then compare these interactions with historical weather patterns to make their predictions.

"We're evaluating official forecasts as they are issued by the Climate Prediction Center (CPC)," Pagano says. "The CPC has a team of about 20 forecasters who use their expertise to blend together the results of more than six classes of models to produce a new set of forecasts each month."

The UA software then compares the forecast to a set of earlier forecasts and observations of what actually occurred.

"There are many different ways to compare the forecasts and the observed outcomes, and matching the user with the right score is part of the art of forecast evaluation," Pagano explains.

For instance, simply converting the forecast to "it's going to be above normal" and then counting the number of times that above-normal was predicted and above-normal was the observed outcome will show how often the forecasts were correct.

"Then we compare the historical performance of the forecast to random guessing," Pagano says. The user can look through increasingly complex sets of statistical results -- depending on his or her level of statistical savvy -- to understand how closely the prediction is likely to match the actual outcome.

TV weather news makes a good analogy, Pagano says. "If the weatherman says there's an 80 percent chance of rain, you could compare that to earlier 80-percent predictions to determine whether it really did rain 80 percent of the time."

With millions of dollars -- and sometimes human life -- at stake, resource managers have been reluctant to use today's sophisticated climate forecasts.

By explaining and evaluating these forecasts, the UA software will help managers benefit from the forecasts, regardless of their training.

The software tool allows users to customize evaluations to consider lead times, seasons and locations most relevant to their specific needs. Ranchers could use seasonal forecasts, for instance, to predict winter and summer forage yields. If conditions look like they will produce low yields, ranchers can stockpile hay ahead of time.

Similarly, those who manage wildfires on public lands can use long-range temperature and rainfall forecasts to evaluate fire danger, decide where to allocate resources, and determine when to conduct prescribed burns.

"Since seasonal forecasts aren't as consistently reliable as short-term weather forecasts, it's important that we communicate their performance characteristics in a way that users can understand and exploit," says Robert Livezey, a former senior scientist at CPC and now chief of the National Weather Service Office of Climate, Weather, and Water Services. "This work makes substantial progress in this regard for specific users."

The software also includes several features that allow users to interactively explore historical climate data for their region to get a better idea of what forecasters mean when they say "above normal" or "below normal."

"This is particularly relevant to residents of the Southwest whose experience may be limited to the recent past and who may have only a foggy notion of what's 'normal' in their region," Pagano says.

But the software isn't limited to the Southwest, he adds. The web pages can evaluate forecasts for any location in the nation.

Although the software targets resource managers, anyone who uses forecasts or is interested in them can benefit from the web pages. Many people use forecasts, from resource managers to public health officials (those who manage valley fever in Arizona, for instance) to energy companies that may be setting contracts to buy heating oil.

The web pages and software have been used in controlled demonstrations in which the researchers work with decision makers to ensure that the information is understandable and on-target for their needs. Now beta testers are providing feedback on the website.

"We are planning for a late summer/early fall release to the general public, after we incorporate comments from the beta testers," Hartmann says. The web pages will become public through UA's Hydrological Data and Information System (HyDIS) website at this URL.

Funding for the project came through grants from NASA's Earth Observing System and from the National Oceanic and Atmospheric Administration (NOAA).


[Contact: Holly Hartmann, Thomas Pagano]

28-May-2002