4.4 Spatial-temporal Mapping of Modified Surface Observations for the Terminal Area Icing Weather Information for NextGen (TAIWIN) Diagnostic Icing Capability

Monday, 29 January 2024: 5:15 PM
317 (The Baltimore Convention Center)
Ben Bernstein, Leading Edge Atmospherics, LLC, Longmont, CO; and A. Gaydos, S. D. Landolt, S. DiVito, J. Lave, D. Jacobson, and S. Faber

For decades, surface weather observations have provided highly-relevant information to the diagnosis and short-term forecasting of icing conditions at the surface and aloft. Operationally, this information is provided in the form of METARs and SPECIs, primarily from the ASOS and AWOS networks. When interpolating or extrapolating this information to locations away from observation sites, it is imperative to account for icing-relevant features, including those associated with topography, fronts, clouds, precipitation, and spatial and/or temporal variations thereof. For example, a METAR observation of freezing fog made in a river valley may be quite meaningful within that valley, but less so in the higher terrain that surrounds it. A different METAR observation of snow (SN) and low ceilings made in a lake-effect snow band may be highly relevant within parts of that band where characteristics are similar, but less relevant just outside of the band, where the snow and clouds may be weaker or even non-existent.

It is well understood that the features reflecting a report’s relevance are often evident in data from other other weather data sources, such as satellite, radar and models. In the case of the lake-effect snow band, a distinct, elongated streak of enhanced radar reflectivity may be present and colocated with a streak of cool cloud top temperatures. Such distinct features can and should be used to spatially “map” the point observations of snow and low ceilings, which are much more applicable within those features than outside of them.

Beyond spatial mapping, it is also well understood that the band and the features associated with it may move and/or change over time. If in 15 minutes, the band moves a few miles to the south, away from the station reporting SN, yet maintains its intensity and signatures, snow is still very likely to be present within that band. Even if the station is no longer reporting SN, the previous METAR observation of SN is still applicable and its value and implications essentially “follow” and evolve along with the clouds on satellite and echoes on radar. If that band weakens, and the character of the signals associated with it have changed, then that observation has become less relevant. Thus, it is equally important to account for both movement and temporal changes when mapping and applying METARs and other point observations.

Finally, reasonable limits should be applied to the age and distance of a point observation from the features that are associated with it. A very recent report made in very close proximity to features that it is matched to is more likely to be relevant than an older report that was made farther away.

Meteorologists and developers are well aware of these factors when applying point observations to icing diagnoses and short-term forecasts, yet METAR observations have traditionally been used differently in icing algorithms. Icing diagnoses valid at a certain location and time typically use only the most recent observation from a station, as long as it was made within a certain distance and was reasonably recent. This application often remains consistent, regardless of the presence of topographic features, shorelines, and relevant meteorological features that may have evolved and moved since the reports were made. While this approach works reasonably well for certain applications and situations, as user needs evolve and icing algorithms produce output at finer space and time scales, it becomes increasingly imperative to place the point observations they employ into proper context if icing conditions are to be effectively represented on such fine scales.

The TAIWIN (Terminal Area Icing Weather Information for Nextgen) program has developed an initial icing capability, producing diagnoses, nowcasts and forecasts for terminal areas in select domains in the CONUS. The terminal airspace is defined as having a radius of 30-nm around the airport and covering from the surface to 12,000 ft AGL. The TAIWIN Capability produces high-resolution, high-frequency icing information for each domain, producing output every 15 minutes at spacing of 1-km horizontally and ~500 ft vertically, as well as singular output for select terminal areas, indicating the predominant icing conditions for the entire terminal area. To accomplish this, its icing diagnosis merges model output from the 3-km HRRR with native-resolution GOES-16 satellite observations, MRMS (Multi-Radar/Multi-Sensor System) radar observations, and point observations (METARs and SPECIs) from the U.S. and Canadian surface networks. In an effort to represent the meaning and value of these observations on frequently-updating, high-resolution grids, TAIWIN’s icing diagnosis attempts to map them spatially and temporally using meteorological context evident in the satellite, radar and model datasets, as well as surrounding topography.

This presentation is intended to describe the first-cut approach used, as well as its benefits, shortcomings, and some potential upgrades. Example cases from the recent FAA In-Cloud ICing and Large-drop Experiment (ICICLE) field program will be used for demonstration.

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.

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner