The global forecast icing product

An algorithm that forecasts icing conditions across the globe is under development. Output fields from the Global Forecast System (GFS) numerical weather prediction model are combined to produce the icing forecasts. Initially, a cloud detection algorithm from the model output fields is used to identify the subfreezing cloud locations along with their top and base heights. Next the vertical cloud structure of the atmosphere is examined using the GFS model soundings, and several possible meteorological icing scenarios are identified. The GFS output is uniquely combined, depending on the scenario, using various decision tree and fuzzy logic methods. These methods then identify the potential icing conditions and their severity. Since cloud microstructure differs across the world, the globe is broken into several climatic regimes including the tropics, mid-latitudes, and polar regions. Regional versions of the cloud scheme, decision tress, and fuzzy logic membership functions are used. This paper presents the global icing forecast product, describes the algorithm used to create this product, and shows the final output.