Convective weather avoidance with uncertain weather forecasts

This paper describes simulations of an automated planning system that routes flights around airspace impacted by forecasted convective weather. If the system predicts that a flight will enter a weather-impacted airspace within a predefined time horizon, it generates a new route. Because the forecasts are uncertain, the system periodically generates, using updates of the weather forecasts and radar tracks, new reroutes. The simulations included convective weather in the northeastern quadrant of the United States over a 24-hr period. Multiple simulations investigated the system performance as the planning horizon and planning frequency varied. As the planning horizon and frequency increased, the system successfully routed more traffic around weather but with more route changes. For a planning horizon of 20 to 120 minutes and a planning frequency of four cycles per hour, the reroutes increased flight time by 3.3% and avoided 79% of the weather-impacted airspaces that were detected. Most flights required one to three reroutes to pass by a weather-impacted airspace, while the worst case flights required six reroutes.