Stratus occurs regularly over the San Francisco Bay Area from late spring to early fall and is a primary cause of delays at San Francisco International Airport. A database of meteorological measurements has been accumulated from around the Bay Area by the Marine Stratus Initiative, which endeavors to provide forecast guidance to optimize airport capacity for arriving flights. A data set derived from this database will be analyzed using the See5 machine learning system to create an algorithm for one-hour real-time forecasts of Bay Area stratus burn-off. This research will demonstrate machine learning analysis as a practical tool for operational forecasting of stratus burn-off, indicating a twofold benefit in applying these methods. Operationally, the algorithm is capable of properly forecasting up to 80% of the burn-off cases. Diagnostically, the analysis identified 10 meteorological variables and corresponding threshold values key to stratus burn-off over the San Francisco Bay.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner