Probabilistic Forecasting of Ceiling and Visibility: Blending LAMP, NWP and Regression

The physical processes controlling ceiling and visibility (for example, the formation, evolution and motion of low cloud, precipitation, fog and haze) and the diverse seasonal, diurnal and geographic influences that modulate these controls yield a challenging forecast problem that impacts the overall efficiency of the national airspace and the safety of general aviation.

Current and emerging ceiling and visibility (C&V) forecast resources comprise an increasingly diverse and skilled population of probabilistic forecast tools – LAMP, WRF Rapid Refresh, SREF, the North American Rapid Refresh Ensemble (NARRE), time-lagged ensembles built out of today's HRRR and NAM, plus observations-based statistical forecast models. While this diversity is desirable, it raises the question how can the strengths within this population of resources be most effectively harnessed and used?

This paper describes the status of current work toward CVF, a 1-10 hour C&V forecast guidance product. The product's projected role is to capture and integrate the skill of current forecast capabilities, to deliver probabilistic forecast skill greater than that of its input components, and to evolve over the years as its input forecast components evolve and advance. We outline the key elements of our development approach and lessons learned to date, focusing on

• A methodology for advantageous integration of LAMP, Rapid Refresh time-lagged ensembles and regression-based forecasts;

• A methodology for real-time determination of optimal probability thresholds for use in generating deterministic forecasts from probabilistic forecast input;

• Verification results comparing LAMP, CVF and WRF Rapid Refresh ceiling and visibility forecast skill.

Finally, we summarize an assessment of the skill improvements achieved through our integrated approach and the outlook for operational application of the CVF methodology developed.