Science has advanced to the stage where predictions of weather and climate variability are issued with lead times of days, months and years. There is, however, uncertainty in all forecasts, and weather and climate forecasts are no exception. Weather prediction currently uses Monte Carlo (ensemble) techniques to describe the potential realizations of the future system state. Such ensemble forecasts can be used to estimate the probabilities of particular events, as well as uncertainty in the forecast. However, current decision making using weather predictions does not fully exploit advances in operational ensemble methods and frequently does not use quantitative decision models to identify and evaluate options, tradeoffs and risks. Probabilistic forecasts allow decision makers to make rational judgments based on their own cost/loss sensitivity to the possible outcomes.
Workshop instructors, led by Dr. Leonard Smith, include scientists from the London School of Economics (LSE) who are actively involved in the NOAA THORPEX Research Program, the European DEMETER project on seasonal forecasting, and the EU ENSEMBLES project on forecast reliability on all time scales. The instructors have applied their forecasting approach to numerous problems in the industrial and commercial sector, including utilities, transportation and the medical systems sector.
Technology is emerging from these projects and from other work at CATS, namely probabilistic forecasting, which can place environmentally induced risk in the user domain to aid decision-making. While there exist a number of important impact studies, the existing plethora of environmental data and emerging forecasting tools remain underutilized by government and commercial organizations. There is a need to "translate" this technology into society and commerce in general.
Excerpt from an article in USA Today, June 19, 2001:
The annual cost of electricity could decrease by at least $1 billion if the accuracy of weather forecasts improved 1 degree Fahrenheit. [The Tennessee Valley Authority] generates 4.8 percent of the USA's electricity. Forecasts over its 80,000 square miles have been wrong by an average of 2.35 degrees the last 2 years, fairly typical of forecasts nationwide. Improving that to within 1.35 degrees would save TVA as much as $100,000 a day, perhaps more. Why? On Monday at 5:30 a.m., TVA's forecast for today called for an average four-city high of 93 degrees in Memphis, Nashville, Knoxville and Chattanooga, rising from 71 degrees at 6 a.m. TVA has scheduled today's power generation based on this forecast and will bring on line a combination of hydro, nuclear, coal, wind, natural gas and oil plants as temperatures rise. It will buy wholesale electricity if that costs less than generating its own power. Gas plants are more expensive to operate than nuclear or coal, so TVA will fire up its "peakers" only when it expects demand to be very high. If the average temperature comes in 1 degree hotter, rising to 94, TVA's customers will demand 450 more megawatts. There would be no time to fire up an idle gas plant, and the cost of last-minute wholesale electricity could skyrocket. "There are times when electricity is $80 (per megawatt hour) a day ahead and $800 to $8,000 24 hours later," says Robert Abboud of RGA Labs, which helps utilities with complex decisions. On the other hand, if the four-city temperature comes in a degree cooler than forecast, TVA may have fired up a plant unnecessarily, or bought electricity a day in advance that will go wasted. Temperature is most important, but utilities can also benefit from accurate forecasts of cloud cover and humidity.