Global icing forecast verification using CloudSat data
Icing, rarely occurring at upper levels, has little impact on aviation at typical enroute flight levels for trans-oceanic routes; consequently, in this context the forecasts have little use in routing decisions, strategic or tactical. If adopted, the icing forecasts will primarily be used in the operational fuel certification process, currently reliant on a relatively naïve icing forecast based only on relative humidity and temperature. As dictated by formal operating procedures, flights must carry contingency fuel to manage circumstances such as cabin depressurization or engine failure that would force the flight into levels where icing might be of concern (e.g., FL100). If, during planning, the forecast indicates that an encounter with icing in the contingency scenario is a possibility, then additional fuel must be carried. Otherwise, the flight can proceed without the additional fuel, obviously the preferred answer from an efficiency perspective.
To assess the quality of the icing forecast products in the context of the operational planning process, the study clearly had the challenge of determining actual icing conditions in vast domains where in-situ observations are, at best, very sparse. Pilot reports, the traditional observation of icing used in CONUS verification studies, have many shortcomings in the global context. The reports are sparse in space and time; they poorly sample the flight levels relevant to the operational decision, as most flights remain at the upper levels, relatively free from icing conditions; and they sample ‘yes' and ‘no' icing in a very imbalanced fashion, limiting the metrics that can be used.
The overall assessment strategy was to build an icing diagnostic based on the polar-orbiting CloudSat cloud profiling radar and a global temperature analysis field. For clouds at the flight levels of concern to operational planning, CloudSat detects cross sectional cloud extent very well, a critical strength of the approach. Within each detected cloud extent, the algorithm combines the cloud type and the temperature from the GFS analysis to determine areas of icing.
This paper describes the development of the icing diagnostic based on CloudSat; how the algorithm was tuned with respect to an existing operational CONUS icing product; and results of comparison with the WAFS icing forecasts.