Vinsamlegast notið þetta auðkenni þegar þið vitnið til verksins eða tengið í það: http://hdl.handle.net/1946/44905
For the aviation industry, safety is of utmost importance. To ensure safety the selection of suitable alternate airport is crucial with regards to their location and existing infrastructure. This is crucial for airlines, particularly in Iceland, where rapidly changing weather can significantly impact air transport operations, leading to inconvenience for passengers and increased costs for airlines.
This research project marks the initial steps toward addressing the main question which is to predict the unavailability of alternate airport in Iceland within a 48-hour timeframe, specifically with regards to adverse weather conditions. However, due to the unavailability of a dataset with a 48-hour weather forecast, the study is based on the currently available 24-hour weather forecast data. It is imperative to find ways to develop a 48-hour meteorological forecast that incorporate weather phenomena affecting flights, such as cloud ceiling, visibility and wind.
The availability ratio was found to be lowest at Akureyri Airport while Reykjavik Airport and Egilsstaðir exhibit availability ratio between 94-100%. Additionally, the analysis reveals that the duration of unavailability at Akureyri Airport is longer compared to other airports in the project. The occurrence of total closure of the four alternate airports in Iceland due to weather is found to be minimal, accounting for only 0,1% of cases. Moreover, three alternate airports remain open 87% of the time in the period studied here and importantly, during the critical arrival hours at Keflavik International Airport between 05:00Z-08:00Z respectively, the availability is around 83%. Interestingly, the findings obtained using a logistic regression model reinforce the conclusion derived from the historical data demonstrating that such model may be used for estimation in future work. They reveal that the highest likelihood lies in accurately predicting the unavailability at Akureyri Airport when the weather conditions outlined in the project (cloud ceiling, visibility and wind) dip below the airport operating minima (AOM). Moreover, it is shown that if unavailability is predicted by the 24-hour weather forecast data, it is most likely to occur precisely 24-hours later with a 1 to 3-hour buffer. Finally, future perspective and recommendations will be presented to guide future projects in addressing the primary research question.