WIND-3 is an analog forecasting system based on fuzzy logic that produces probabilistic predictions of cloud ceiling height and horizontal visibility at 198 airports in Canada. For data, it uses historical observations (climatology), current observations (METARs), and model-based guidance (GEM). It has run continuously for the past two years and produces useful guidance for aviation forecasters. It is described in detail by Hansen (2007). The present article provides an update on three areas of subsequent research: 1) verification of probabilistic forecasts using ROC diagrams, including comparison with the benchmark technique of conditional persistence climatology (CPC); 2) experiments with alternate scenarios, for example, predicted ceiling and visibility with rain versus without rain; and 3) applicability of analog forecasting and CPC to prediction of fog.

Supplementary webpage: http://collaboration.cmc.ec.gc.ca/science/arma/aram