The Current Icing Product (CIP) and the Forecast Icing Product (FIP) that freely available on the Aviation Weather Center website are both utilized in the EZWxBrief progressive web app. Shown below is the icing depiction in the EZRoute Profile. So let's take a look at the differences between CIP and FIP.
Both CIP and FIP consist of three products, namely, Probability, Severity and Supercooled Large Drop (SLD) icing. SLD is a "potential" field whereas the icing probability field for both CIP and FIP is a calibrated probability. You can read more about the importance of using a calibrated probability here.
First of all, it is important to understand that CIP is an analysis. In other words, it's a glimpse of the recent past. Therefore, CIP is not intended to tell you about the future. That’s the job of FIP. CIP is produced on a grid that roughly matches the Rapid Refresh model output. Each grid box is roughly 13 km or about 8 miles from side to side. When CIP runs, it produces output every thousand feet from the surface to FL300.
On the Aviation Weather Center website, CIP images like the one show above are provided every 2,000 feet MSL starting at 1,000 feet MSL and continuing up to and including FL290. However, using the EZImagery in EZWxBrief, the charts are shown every 1,000 feet through FL300. A composite image that combines all levels through FL300 is also available.
Both CIP and FIP get updated once per hour. The big advantage of CIP is that it doesn’t rely on just one data source. Instead, it combines model guidance with data from numerous observational sources to produce the hourly analysis. As shown below, CIP combines a 3-hour Rapid Refresh model forecast along with observations from satellite, NEXRAD, METARs, lightning data and PIREPs of icing to assess the probability that icing was present at the valid time on the chart. There's valuable icing information in all of these sources. Rather than rely on just one source that may have serious limitations, CIP takes advantage of the strengths of each data source and tries to minimize its weaknesses by leveraging off information provided by the other data sources.
The CIP analysis runs hourly and is available about 20 minutes after each hour.
CIP uses several model parameters that are not typically available through sensors. This includes relative humidity and temperature since they give you a good indication of where the moisture exists and whether or not it’s at a good temperature for icing. CIP also uses forecasts of vertical velocity and liquid water content, but such fields are extremely difficult for a model to predict accurately on a consistent basis. For now, their not quite good enough to stand on their own, but the information is definitely useful. So, CIP uses it, but does not rely upon it to say where the icing is. Like any of the datasets and parameters used by CIP, they each provide valuable information, but they are most powerful when used together.
For the most part, CIP sets the temperature limits to -25°C on the low end and 0°C on the high end. This is because icing becomes increasingly rare with decreasing temperature, and its rather unusual for it to occur when the temperature is colder than -25°C. So, to avoid over warning for icing at cold temperatures, CIP’s lower limit was set to -25°C.
There is one important exception, which is deep, moist convection, when the temperature threshold is pushed down to -30°C. It is in these situations that icing has the best chance to occur at such temperatures. That limit could be even be extended down to -35°C to capture the very cold end of the spectrum for icing in convection, but for now, it's set at -30°C.
The model forecasts of relative humidity give an indication of where the moisture is likely to be. Most pilots are aware that without clouds present, there’s no chance for icing. Even though the model forecast of relative humidity may be high, in reality, there may or may not be clouds at that location. Observations from satellite and surface stations provide a great deal of information to allow CIP to limit the locations where it indicates icing to those where clouds exist.
CIP starts with the satellite imagery. It uses imagery from multiple wavelengths and derivatives thereof to make its first guess at where the clouds are located. The combinations are a bit complicated, but clouds have specific properties that make them show up well at certain wavelengths. In particular, CIP looks at the signals in the visible, long-wave and short-wave infrared channels to determine whether or not it’s cloudy during the daytime.
At night CIP focuses on the long- and short-wave infrared channels to make its decisions. This still works quite well. The cloudy/clear determination is toughest right around sunrise and sunset, when both the visible and short-wave infrared channels have problems. During these narrow time windows, CIP falls back on a combination of the long-wave infrared, ceiling reports from METARs and some model fields. It’s still very good, but this is CIP’s weakest time of the day.
From a pilot’s perspective, PIREPs are like gold when it comes to getting a feel for the icing conditions that are out there. If there were 50 or more reliable pilot reports along your route of flight, knowing where to expect icing might be an easier task. However, pilot reports are often sparse especially during the overnight hours.
From CIP’s perspective, PIREPs play a critical role as well. Pilot reports are only considered “point observations,” that given an indication of the presence or absence of icing at a particular time, location and altitude or range of altitudes. The report normally indicates the severity and type of icing as well. If a pilot report is given in a region where CIP suspects that icing exists, the confidence that CIP is correct increases, which can increase the icing probability.
Surface observations or METARs also provide point observations in space and time, like PIREPs do. CIP uses METARs in several ways. Sometimes they are used to help determine whether or not clouds are present, especially when satellite fields are at their weakest. Their primary uses in CIP are to define the height of cloud base and to help determine whether or not precipitation is occurring over a given location and what the precipitation type is.
When it's cloudy over an airport, we all know that the METAR reports the cloud cover and the ceiling height. The cloud covers include obscured, overcast, broken, scattered, few and clear. Past studies have shown that the vast majority of icing occurs when at least broken sky cover is reported, but some also still happen when only scattered clouds are reported. Some of the time, this may be because more solid cloudiness is present nearby, but not right over the airport. To be a little on the conservative side, CIP will use the lowest reported ceiling heights among all cloud layers indicated to have scattered or greater cloud coverage.
When there’s no precipitation falling, the cloud base represents the lowest altitude at which structural icing is possible. So, the METAR cloud base sets that lower bound and no icing will be indicated beneath it, as long as there’s no precipitation occurring. If precipitation does exist below the cloud bases, now it's possible to have an icing hazard at those altitudes. If freezing drizzle or freezing rain is reported, then that threat is pretty high, because supercooled large drop (SLD) icing may be present below the cloud base.
Ice pellets, rain and drizzle also give some clues that icing might be present below cloud base, but it’s a little trickier to figure out if it's there or not. Regardless of the precipitation type, CIP puts these reports into the context of the information that it has from the model and other observations to assess whether or not there’s an icing threat below cloud base. Snow falling from the base of the lowest cloud deck isn’t considered to present an airframe icing hazard. In general, snow doesn’t stick to the aircraft. When snow is the only precipitation type reported in a METAR, the implication is that there is no airframe icing threat below cloud base.
METARs can only provide point observations of precipitation. Ground-based radar data provides more refined information on where precipitation is occurring, as well as its reflectivity. These data not only help to fill the gaps between stations, but also provides some indications of the precipitation rate and the size of the particles that are falling.
They give you some idea about how strong or heavy the precipitation is. For example, assume you know that you are in a freezing rain situation and the reflectivity is 20 dBZ in one area and 40 dBZ in another. Where the reflectivity is higher, the rain drops should be larger and/or more plentiful. In a snow situation, higher reflectivity can imply larger and/or more snowflakes, which may decrease the risk of icing, especially near cloud base. Therefore CIP leverages the composite of the NEXRAD data valid at the top of the hour.
Lightning, of course, indicates deep convection. When lightning is around, you’ve got a situation where supercooled liquid water is very likely to be present, perhaps in copious amounts. Large drops or SLD can be present, too. When an observation of lightning is ingested by CIP, it immediately raises a red flag that deep, moist convection is present.
CIP is only considering lightning strikes that have occurred in the last 15 minutes, and they are all treated the same. Let’s say the temperature aloft is -27°C, but lightning is in the vicinity. CIP uses -25°C as a cutoff, but when deep, moist convection is identified by CIP, it will allow for icing to be diagnosed all the way down to -30°C, so it will indicate icing to higher altitudes in the deep convection than it would if convection was not found. And it's also one of the most common reasons you will see CIP show the potential for SLD.
The basic concepts and physics employed by FIP are the same as they are in CIP. The main difference is that every bit of information has to be derived from the model itself, because you don’t have things like a satellite image for 6 hours from now. So, FIP has to depend on the model to emulate all of the observations that CIP normally has at its disposal. For example, forecasts of relative humidity are used to derive the locations of clouds, their cloud base heights, cloud top heights and cloud top temperatures…things that primarily came from satellite and METAR observations used by CIP.
But it can’t possibly be as good as CIP. There’s nothing like good observations when it comes to icing, but FIP makes good estimates of the cloud parameters as mentioned, as well as things like where precipitation will be and what type it will be. Everything is a surrogate, so its not quite like the real thing, though. For example, because FIP is dependent on the model moisture fields, it can forecast clouds in places where they didn’t actually end up occurring. So, it might forecast icing in cloud-free areas at times, while CIP will never indicate icing in a place where it did not first diagnose clouds or precipitation. In the end, FIP is still quite accurate, even out to 12 hours.
Most pilots are weatherwise, but some are otherwise™
Dr. Scott Dennstaedt
Weather Systems Engineer
CFI & former NWS meteorologist