Estimates of the economic value of weather forecasts for use in air traffic management (ATM) would significantly bolster current verification information, allowing for better decisions regarding the type and quality of data required for NextGen. The complexity of ATM, however, demands an approach beyond the traditional cost/loss model, which clearly fails to capture a meaningful measure of value for this domain. In an attempt to understand the economic outcome of automated ATM planning based on weather forecast information, NOAA's Earth System Research Laboratory (NOAA/ESRL) has collaborated with Boeing Corporation to examine the use of convective forecasts in a flow model.

The National Flow Model (NFM), a queue-based air traffic flow simulator, utilizes projected schedules and calibration settings from historical cases to create ATM plans based on forecast weather, and execute those plans against actual weather for a particular scenario. Importantly, it provides output that includes flight delays and cancellations, a direct measure of cost. The value of the forecast can be measured by comparing its associated cost with that of utilizing no forecast information and that of utilizing perfect knowledge of future weather.

Initial runs of the NFM, focused on five study days in the 2007 convective season, failed to provide meaningful output, primarily because the forecasts of convection indicated far too much impacted airspace. In each of the simulations, the over-forecast of hazardous weather forced the planner to essentially stop the flow of aircraft entirely. Through subjective evaluations, however, forecasters and air traffic managers have indicated that the forecasts have value for the planning process.

Motivated to understand how the forecasts might be made useful as input to the NFM, ESRL studied various post-processing techniques to condition the forecasts for air traffic planning. This paper will describe the conditioning of the forecast products and promise for measuring economic value through the outlined approach.