Extrapolation of observational icing data Looking for icing scenarios Using numerical model outputs Crossing observations with model outputs (data fusion, icing index...) Towards a conclusion
+WMO
1. 2.
Extrapolation in space and time of icing observations. Documenting icing situations with the help of observational data and standard meteorological forecasts. Using this information to identify scenarios known to be favourable for icing. Using an icing index produced by a numerical model.
3.
+WMO
If one has a reliable source of icing observations (accretions) coming from airspace users, they can be used to depict the phenomenon and to anticipate it by appropriate extrapolation in time and space. When you have frequent observations (PIREP or SPECIAL AIREP) nicely spaced, one can consider an operational follow-up of the phenomena (very short range forecast) (*) This has the advantage that the users are implicated in the forecasting process which can help in detecting the flaws in the other forecasting techniques. This can also be used to document extreme situations.
In any case, this approach is often biased by the difference of the observations.
+WMO
This picture displays the inflight icing observations coming from PIREPs over the US. They have been superimposed on a numerical model output of SLD icing potential. (*)
+WMO
With good knowledge of the meteorological situations favourable for icing conditions, each time one finds such a scenario (*) in the observational data (**) or in the numerical model output, one has to inform the users.
Drawback: it is a difficult method to put in place over large areas and needs a costly local expertise.
+WMO
If you have numerical models with a fine enough mesh (mesoscale), one could consider an output of an icing index (*). This method has to be supported by a regular evaluation through comparison with observational data, covering all the classic scenarios. This method has the advantage that it is easy to put into operation for large areas.
The quality of the forecast is correlated to the models ability to reproduce complex processes on a small scale: realistic microphysics, sophisticated but robust schemes (**).
+WMO
water vapour condensation-evaporation Bergeron-Findeisen process, homogeneous nucleation melting riming, dry and wet growth
This is an explicit scheme (*) which works within the MESO NH model (**). Cloud LWC is an output of such a process.
+WMO
For forecasting icing conditions, research is ongoing to automatically combine the model output indices with combined observational data (satellite, radar, PIREP). ADWICE (Germany) combines the output of the LM model with conventional observations (SYNOP, METAR) and radar, but does not use satellite data. CIP (US) is an index of icing severity that combines the output of the RUC model with all possible icing observations, PIREPs and electrical detection included. SIGMA (France) combines the output of the Aladin model (air and surface temperature, humidity) with satellite and radar observations (*).
+WMO
FZDZ high risk under cold clouds tops FZRA light risk under cold clouds tops The use of these combined products needs a good knowledge of the resources and used algorithms and a validation for local use (**), for example in a mountainous region.
+WMO
In general, it seems that the latest automated products have the ability to distinguish (*) the icing cases from the non icing ones (risk/no risk), but the results are marginal when identifying the situations with severe icing. Moreover, convective situations are usually badly captured by numerical models. It is therefore that when a local study (**) is added, the scores of detecting severe icing will improve to the satisfaction of all aeronautical users (***).
But the constantly improving remote sensing techniques and numerical models leave us with a good margin for improvement.
+WMO
10
+WMO
11