A safety hazard for many aircraft is the accumulation of ice which can result from flying through areas of supercooled liquid droplets. Drizzle droplets pose the greatest hazard by freezing behind the deicing boot on the aircraft wing thereby destroying proper airflow and potentially causing significant loss of lift. Traditionally the only indication of this hazard has been from reports by pilots flying through such areas. While these pilot reports have been used in conjunction with model data diagnostics to produce icing forecasts, there remains a need for better remote sensing capability of icing potential.
The WSR-88D Nexrad radar was designed to detect a precipitation drop size of 1 mm but has sufficient power to detect drizzle droplets between 50 and 150 microns in size. In its clear air operational mode the WSR-88D can detect reflectivities from -30 to +30 dBZ. In precipitation mode detection is from 0 to +80 dBZ. Drizzle droplets typically give returns between -10 and +10 dBZ. The Aviation Weather Center (AWC) has attempted to construct a freezing level composite from the WSR-88D data to be used as an aid in its icing forecasts. A 4-km resolution composite of reflectivity at the -10 C temperature level is generated four times each hour using ETA model data, IR satellite data and reflectivity data from the lowest four elevation scans of each Nexrad radar site. For each pixel in the 4 km digital area the program determines the height of the -10 C temperature level from the Eta model and the closest radar. Using the distance from the radar to the pixel it calculates the elevation scan within which the temperature level falls and checks for any reflectivity reported at this location. The "cone of silence" over each radar site is filled with data from the neighboring radar site if it is available. Areas with no data are flagged, and a final quality control check is done with satellite data to ensure that there is a cloud present at the location of the radar echo.
Over the winter of 1997-98 AWC forecaster experience showed that the composite was quite useful, especially when used in conjunction with pilot reports and satellite data, but did exhibit some deficiencies. There is no means to distinguish between frozen and liquid particles, and in precipitation mode the radar frequently misses drizzle droplet returns. The limited number of elevation scans available causes a "cone of silence" over each radar but access to all scans would greatly reduce the size of this area. Also, instances of significant icing could go undetected where the freezing level fell below the bottom of a radar's lowest elevation beam. In spite of these limitations the freezing level composite has become a primary data set used by forecasters in trying to delineate areas of potential icing hazards.