Creating Weather Impact Scores from Current Weather Forecast Products for a Probabilistic Airport Capacity Model

Current aviation weather forecasts, such as the LAMP and the TAF, provide information on a multitude of sensible weather factors. In some cases, this wide assortment of factors can be useful, but in others it can actually detract from the overall value of the forecast. For example, this forecast data is beneficial for human forecasters as it gives them an increased understanding of the future state of the atmosphere and, as a result, the ability to produce a more accurate forecast. However, when using these products in our probabilistic airport capacity model, Weather Translation Model for GDP Planning (WTMG), this large amount of weather information actually hinders the overall performance of the model due to a weakness in the underlying regression tree methodology. To address this deficiency, the hourly forecasts from the LAMP and the TAF were replaced with a set of numeric impact scores. These scores combine multiple operationally significant forecast fields into a single score that quantifies the anticipated level of impact at a given airport. In this research three types of impact scores were developed: ceiling/visibility impact (combines ceiling and visibility forecasts), wind impact (combines wind direction and wind speed forecasts), and significant weather impact (combines precipitation and thunderstorm forecasts). By using these scores instead of the raw weather forecasts, the number of weather factors fed into the WTMG is reduced from as many as 10 to 15 to three. These impact scores were used within the WTMG at eight different airports, including BOS, EWR, ORD, and PHL, with favorable results. In this paper we present a review of previous work completed on the WTMG, an overview of the processes used to create each of the impact scores, and the results of the WTMG for each of the airports when using the impact scores as the model's primary source of weather information.