Probabilistic Forecast of Visibility at Gimpo, Incheon, and Jeju International Airports Using Weighted Model Averaging

In this study, weighted model averaging (WMA) was applied to calibrating ensemble forecasts generated using Limited-area ENsemble prediction System (LENS). WMA is an easy-to-implement post-processing technique that assigns a greater weight the member forecast exhibits better performance; it used provide probabilistic visibility forecasting in form of predictive probability density function for ensembles. The mixture discrete point mass and two-sided truncated normal distribution components. Observations were obtained at Gimpo, Incheon, Jeju International Airports, 13 LENS, period December 2018 June 2019. Prior applying WMA, reliability analysis conducted rank histograms diagrams identify statistical consistency between ensembles corresponding observations. method then each raw model, proposed. Performances evaluated mean absolute error, continuous ranked score, Brier integral transform. results showed proposed provided improved performance compared with ensembles, indicating well calibrated predicted function.