A Fuzzy Logic Fog Forecasting Model for Perth Airport

Perth Airport is a major airport along the southwest coast of Australia. Even though, on average, fog only occurs about twelve times a year, the lack of suitable alternate aerodromes nearby for diversion makes fog forecasts for Perth Airport very important to long-haul international flights. Fog is most likely to form in the cool season between April and October. This study developed an objective fuzzy logic fog forecasting model for Perth Airport for the cool season. The fuzzy logic fog model was based on outputs from a high-resolution operational NWP model called LAPS125 that ran twice daily at 00 and 12 UTC, but fuzzy logic was employed to deal with the inaccuracy of NWP prediction and uncertainties associated with relationships between fog predictors and fog occurrence. The outcome of the fuzzy logic fog model is in one of the four categories from low to high fog risk as FM0, FM5, FM15 or FM30, intended to map to approximate fog probability of 0, 5, 15 and 30%, respectively. The model was found useful in its 5 year performance in the cool seasons between 2004 and 2008 and required little recalibration if mist was treated as if it were also a fog event in the skill evaluation. To generate an operational fog forecast for Perth Airport, the outcome of the fuzzy logic fog model was averaged with the outcomes of two other fog forecasting methods using a simple consensus approach. Fog forecast so generated is known as the operational consensus forecast. Skill assessment using frequency distribution diagram, Hansen and Kuiper skill score, and Relative Operating Characteristic curve showed that the operational consensus forecast outperformed all three individual methods. Out of the three methods, the fuzzy logic fog model ranked second. It performed better than the other objective method called GASM but worse than the subjective method which relied on forecaster's subjective assessment. The skills of the fuzzy logic fog model can be further improved with the tuning of fuzzy functions. In addition, similar models can be customised for other airports. The study also suggested the use of the simple consensus approach to enhance forecasting skills for other stations or weather phenomena if there were two or more independent forecasting methods available.