Improving Probabilistic Forecasts of Aviation Weather Hazards

The next generation of ICAO-sanctioned significant weather (SIGWX) hazard forecasts will be automated grids of probabilistic forecasts derived from a combination of numerical weather prediction (NWP) model outputs (e.g., from the World Area Forecast Center (WAFC) London and WAFC Washington), replacing the current, manually generated guidance products. The Ensemble Prediction of Oceanic Convective Hazards (EPOCH), a multi-model, ensemble-based system was designed, in part, to address these next generation requirements. The forecast guidance will support strategic planning for transoceanic flights with lead times of up to 36 hours, which is needed for the longest commercially available flights today. Currently, EPOCH produces calibrated probabilistic forecasts of convection, and cloud-top heights exceeding 30 kft.

We report on recent R&D activities related to refining the current methodologies for calibrating and combining forecast guidance from multiple global ensemble models, and we take a closer look at approaches that may improve measures of convection forecasting performance (e.g., reliability, resolution, sharpness, coverage) using the Convective Diagnosis Oceanic (CDO) product for verification. More specifically, we expand on previous studies and provide a characterization of the trade-offs between bandwidth and performance using the logistic regression approach in calibration and computation of probabilities in comparison to a relative frequency approach. Additionally, in light of independent forecast quality assessment feedback, we discuss potential benefits of a regional, differential weighting scheme to optimize the pooling of probabilities. Finally, we will touch on some of the challenges encountered when attempting to make use of probabilistic predictions for improved decision making in situations where there is a cost asymmetry between false positive and false negative cases.

Disclaimer: This research is in response to requirements and funding by the Federal Aviation Administration (FAA). The views expressed are those of the authors and do not necessarily represent the official policy or position of the FAA.