Doug,
I created those plots on ADDS (also visible via my Real-Time Weather Data
Pages). The amount of supercooled liquid DOES NOT directly translate to
any probability of icing. It is just a depiction of the amount of cloud
water produced by the RUC microphysics in locations with temp less than 0C
(subcooled or supercooled using an improper term). Please do not use it as a
sole predictor of icing locations: the RUC microphysics (or any numerical
model for that matter) are not yet that good. Besides ADDS my other main duty
here at NCAR is development of those microphysics for MM5 and then transition
to RUC (with John Brown's help).

If you want an "icing likelihood" product, use the Icing product on ADDS
(IIDA - Integrated Icing Diagnostic Algorithm) which combines model data,
METARs, satellite and radar data into analyses of where icing is likely to
be happening (in a nowcast sense). This 0-100 likelihood scale is not a true
probability in a statistical sense. It also does not imply more severe icing
for higher likelihood although that relationship does bear out somewhat in
RAP's statistical evaluations.

As an integral member of the Icing Product Development Team here at NCAR-RAP,
I can tell you that IIDA is the best the science has to offer at this time.
Of course we'd like to tell you where icing is, how severe, and what type with
high accuracy, but we're simply not there yet. A forecast version of IIDA is
being created right now. By winter 00-01, we hope to release it on ADDS.

--Greg Thompson
 NCAR-RAP