The Future of Global Airframe Icing Forecasting at the UK Met Office: Exploring Probabilistic Options and Exploiting Machine Learning.

The Met Office is one of two World Area Forecast Centres (WAFC) responsible for providing global weather forecasting services for Aviation. The WAFCs are guided by Annex 3 to the International Civil Aviation Organization (ICAO) Convention on Civil Aviation which includes the requirement to provide global gridded forecasts of inflight airframe icing. Since November 2020, the WAFCs have been required to provide global forecasts of icing severity at 0.25° resolution, in support of the ICAO Global Air Navigation Plan (GANP) and associated Aviation System Block Upgrades (ASBU). ASBU block 1 requirements include development and delivery of a probabilistic icing severity forecast in the coming years, of which multiple prototypes have been trialled and tested. The outcomes of this work are presented, with the implementation of new WAFS probabilistic gridded hazard data to follow in the coming years. Looking forward, on behalf of the UK Civil Aviation Authority (CAA), options have been explored at the Met Office to present forecast inflight icing data exploiting machine learning techniques. Using 40 years of reanalysis data from the WeatherBench dataset (processed ERA5), a convolutional neural network (CNN) has been developed and trained to predict the likelihood of icing at a low spatiotemporal resolution, out to a forecast range of 5 days. Using a basic index and limited input parameters, the machine learned forecast has been compared to a persistence forecast and has shown to demonstrate skill. The methodology and outcomes of this proof-of-concept research are presented, along with explorations to translate these output into probabilistic information.