Army Aircraft Icing
45
forecasts are the typical method pilots use for determining whether to launch and
where and when not to fly in icing weather. However, icing forecasts are often in
error with regard to timing, location, and intensity of icing conditions.
Forecasting icing conditions is difficult because icing conditions are typically
not directly observed by weather observers, but are determined indirectly from
other atmospheric parameters. Air temperature, which must be near or below 0C
for ice to form, is measured directly. However, supercooled cloud liquid water
content, the most important variable in addition to temperature for assessing in-
flight icing conditions, is typically derived from other measurements, such as
dew point or relative humidity. Modeling explicit cloud microphysics from
temperature and relative humidity is difficult because upper air temperature and
relative humidity measurements are made by radiosondes only twice per day, at
locations hundreds of kilometers apart. Thus, predicting the location of icing
clouds, which are transient in both space and time, is extremely difficult. Vertical
atmospheric motion is enhanced or suppressed by warm or cold fronts, low
pressure, or topography, and varies on temporal and spatial scales finer than the
radiosonde observing network. For example, because of poorly understood
dynamics, icing forecasts may under-forecast icing frequency in mountainous
areas, increasing the possibility that aircraft will encounter dangerous conditions
when forecasts indicate that it is safe to fly (Stanley et al. 2002). This is con-
sistent with comments from the commander of the 1-501st (ATK), in the com-
mander's questionnaire, that accurate icing forecasts are difficult to obtain in
mountainous terrain of the Balkans. In addition, poorly understood processes can
enhance or suppress in-cloud icing. For example, supercooled cloud droplets can
exist indefinitely in their supercooled state. However, they can also spontane-
ously freeze, or glaciate, and cause their neighboring supercooled drops to also
freeze. Therefore, liquid clouds can remain supercooled for many hours and
present a hazard to aircraft for the entire period. Clouds that freeze to ice, or
glaciate, are not generally dangerous to aircraft.
Older icing algorithms are based principally on radiosonde observations of
temperature, humidity, wind speed, and wind direction with height. Because of
the infrequent and spatially distant measurements, forecasters use models to sim-
ulate atmospheric physics, typically on a fine-scale grid a few tens to hundreds of
kilometers across, at a 1- to 3-hour time frequency. This improves the temporal
and spatial quality of icing forecasts. In addition, surface and satellite observa-
tions add information about the time and location of cloud cover and precipita-
tion. Since icing avoidance can be accomplished by avoiding cold air, or by
avoiding clouds and liquid precipitation, high resolution observations of these
variables are a valuable asset to icing forecasters.