On January 18, 2011, the Forecast Icing Potential (FIP) found on the operational Aviation Weather Center (AWC) website received a much-needed face lift. For this update, the Forecast Icing Potential was transitioned to the Forecast Icing Product (FIP). This next generation FIP is following in the footsteps of its analysis counterpart, the Current Icing Product (CIP). A little over two years prior, the Current Icing Potential also got a face lift and was transitioned to a suite of three (3) products, namely, CIP Probability, CIP Severity and CIP SLD - SLD stands for Supercooled Large Drop icing. FIP will underwent this same transformation in 2009, but the new calibrated version of FIP carries more than just a simple name change.
Just like CIP, FIP now consist of a suite of products, namely, FIP Probability, FIP Severity and FIP SLD. FIP severity also includes the 25-percent and 50-percent masked FIP severity products. With the addition of these products, FIP is also getting its turn to be calibrated. While the FIP severity field and FIP SLD field will look similar to the operational CIP counterpart, the FIP probability field will look a bit different especially at the forecasts beyond 6 hours.
From uncalibrated to calibrated
If you were paying close attention over a decade ago, you may have noticed that CIP went through this same transition from a potential field to a probability field. The old potential field isn’t calibrated, meaning that it isn’t a statistical probability. The colors representing the potential field were contoured at 95, 85, 75, 65, …, etc. However, for CIP's transition from an icing potential to an icing probability, the developers of CIP found that when calibrating it, they could not statistically justify probabilities in excess of 75 percent. As a result, the highest probability shown in red (below) indicates a 75- to 85-percent probability of icing.
What's the difference between calibrated and uncalibrated?
CIP is an analysis. An analysis is not an icing forecast, but a glimpse of the icing environment in the recent past. CIP builds the icing environment based on observational data such as METARs, PIREPs, NEXRAD, satellite and lightning in addition to the various numerical weather prediction parameters supplied by the Rapid Refresh (RAP) model’s three-hour forecast. Of course, we’re rarely 100-percent certain that icing exists, so the product is shown as a statistical probability. FIP, on the other hand, is strictly model-based and must take all of its icing diagnostics from the model alone making for more uncertainty. Consequently, it will never be as good as CIP, but does provide valuable guidance for many pilots even out to 18 hours.
Here’s the crux of the difference between the forecast potential (uncalibrated) and forecast probability field (calibrated); as the forecast increases in length, the probability of icing will generally decrease due to uncertainty. If you look at a FIP probability forecast starting at two hours you will notice that the icing probabilities will generally decrease through three, six, nine hours and beyond. It may appear to the unwary eye as if the icing chances are dissipating over time. While that may indeed be what is forecast to happen, the most likely reason for the decreasing probability is uncertainty in the forecast, not a dissipation of icing. In other words, with a model-based solution, can you really be sure of an icing event that’s nearly 18 hours in the making with any certainty? The old potential field did not take this uncertainty into account. The result of the new calibrated field is a forecast at 18 hours that doesn’t generally exceed a probability of 35 percent.
This will also affect the FIP masked severity. With a 18-hour forecast, the 50 percent masked severity field will rarely show anything more than grey, which reflects the increase in uncertainty with forecast time. Even at 6 and 9 hours, the lack of any forecast probabilities above 60 percent will be evident.
The 12-hour maximum icing potential field (left) shows a marked decrease in values after the product became a calibrated icing probability field (right). With the potential field there are lots of locations with values of 75 and greater. However, the calibrated probability field doesn’t exceed 45-percent which demonstrates the uncertainty with forecast length.
What other improvements were implemented?
Other “under the hood” improvements in the new FIP include a better scheme for icing in convection, more accurate cloud top estimation, and enhanced supercooled large drop (SLD) forecasts. These won’t change the look-and-feel of the product, but should provide an added margin of safety for the flying public.
Most pilots are weatherwise, but some are otherwise™
Dr. Scott Dennstaedt
Founder, EZWxBrief™
Weather Systems Engineer
CFI & former NWS meteorologist
