The Terminal Area Icing Weather Information for NextGen (TAIWIN) Icing Nowcast - An Initial Approach

The FAA TAIWIN (Terminal Area Icing Weather Information for NextGen) Program has developed an aircraft icing Capability that provides diagnoses and forecasts of four icing categories (small drop, freezing drizzle, freezing rain and no icing) from the current time to 12 hr in the future for select terminal areas across the CONUS (Continental US). The capability was rapidly developed, tested, and began running in real time in November 2021. Its diagnosis and forecast components both employ a time-lagged ensemble of HRRR model output (“HRRR-TLE”) to produce high-resolution grids of icing. While the HRRR-TLE is the sole input for the forecast, the diagnostic combines the HRRR-TLE grids with information from satellite, radar, and surface observations to produce a more comprehensive assessment of the icing environment. This difference in data sources can result in harsh transitions between diagnostics and forecasts that could impact operational decisions. In an effort to improve this transition, the TAIWIN program is developing a nowcast component as part of its icing capability.

The TAIWIN icing nowcast provides short-term forecasts at 15-min intervals between time t+15 min and t+105 min in the future, bridging the gap between the diagnostic, valid at time t=0 (the present) and the forecast, which begins at t+120 min (2 hr) in the future. Several nowcast tools and combinations thereof are being considered for use, including persistence, feature tracking/advection, and short-term model forecasts, with a goal of providing a high-quality, smoother transition between diagnostics and forecasts across the 0-2 hr time horizon.

As with many meteorological phenomena, persistence has proven to be an effective short-term icing forecast. The value and performance of persistence and the temporal degradation thereof relates to how dynamic the icing environment is. Radar feature tracking has historically proven skillful in the short-term prediction of convective storms, and there has been great interest in whether this skill may translate to icing-relevant features in radar and satellite data. The TAIWIN team has tested feature tracking and advection techniques on cool-season radar echoes and satellite features for a wide variety of icing events. Short-term forecasts of icing-relevant features from all of these methods have proven to be reasonably skillful out to 2+ hr, giving developers confidence that their inclusion in icing nowcasts may prove beneficial.

The initial approach for the TAIWIN icing nowcast is a fairly simple, weighted combination of persistence icing grids from the diagnostic and forecast icing grids from the HRRR-TLE. The weight applied to the diagnostic grids decreases linearly from 100% at t=0 min to 0% at t=+120 min, while the weight applied to the forecast (HRRR-TLE) grids increase linearly from 0% to 100% across that same forecast window. The resulting, merged icing grids provide a relatively smooth transition between the diagnostic and forecast in nowcast grids valid at +15, 30, 45, 60, 90 and 105 min. The inclusion of tracked or advected icing-relevant features in satellite and radar data is expected to make the initial TAIWIN nowcast more realistic.

In this paper, the initial approach to TAIWIN’s icing nowcast will be described. Each component that is currently used and those under consideration for future use will be discussed. Example output from a variety of cases from the ICICLE field program will be presented, including events that were dominated by freezing drizzle, freezing rain, and small drop icing.

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.