Abstract
The effective management of traffic flows during convective weather events in congested air space requires decision support tools that can translate weather information into anticipated air traffic operational impact. In recent years, MIT Lincoln Laboratory has been maturing the Convective Weather Avoidance Model (CWAM) to correlate pilot behavior in the enroute airspace with observable weather parameters from convective weather forecast systems. This paper evaluates the adaptation of the CWAM to terminal airspace with a focus on arrival decision making. The model is trained on data from five days of terminal convective weather impacts. The performance of the model is evaluated on an independent dataset consisting of six days of convective weather over a variety of terminal areas. Model performance in different terminal areas is discussed and the sensitivity of prediction accuracy to weather forecast horizon is presented. I. Introduction future air traffic system capable of predicting convective weather impacts and proactively issuing TMIs will more effectively use the available airspace, and in turn mitigate the effect of convective weather on the system. The Convective Weather Avoidance Model (CWAM) is a probabilistic model of pilot decision making in the presence of convective weather. CWAM is based on the correlation of spatially filtered weather observations with trajectories of aircraft that penetrated or avoided areas of convective weather in the en route flight regime [1]. The output of the en route CWAM is a three-dimensional {cloud tops, flight altitude, precipitation intensity} Weather Avoidance Field (WAF) that provides the likelihood that a pilot will deviate at a specific position and time given the current and forecasted weather. Outside of the en route phase (e.g. during departure and arrival), aircraft are commonly below the tops of most convection and are subject to different decision mechanisms, both of which are not modeled in the original CWAM. Therefore, in order to model impacts over an entire flight trajectory, CWAM should be adapted to include low-altitude flight phases such as arrival and departure [2]. This paper presents an evaluation of the adaptation of CWAM for arrival operations. Arrival CWAM is trained on approximately 11,000 flights and 1,900 terminal weather encounters over five convective weather days [3]. The training database includes multiple types of weather avoidance decisions that occur during arrival operations to four major metroplex areas (ORD, DFW, CLT, DEN). The decisions types distinguish between strategic and tactical time horizons and encompass both pilot and air traffic management decisions. Additionally, unlike pilots in en route airspace who may have an option to fly at higher altitudes over storms, pilots in arrival airspace are constrained to follow descending trajectories that are typically below the cloud tops. For this reason, the output of the arrival CWAM is a two-dimensional WAF {precipitation intensity, cloud tops}. The performance of arrival CWAM is evaluated by an independent dataset, where the sensitivity of the model to terminal airspace structure and weather forecast horizon are investigated. The independent dataset contains weather decisions from six convective weather days in a variety of terminal areas (ORD, DFW, DEN, CLT, BOS, JFK/LGA/EWR, DCA/IAD). The most descriptive features of pilot avoidance of convective weather are precipitation intensity and storm height, where a 4 km spatial filter on the 90 th percentile value of each feature corresponds to the best tradeoff between probability of detection and false alarm rate.The performance of the model
Original document
The different versions of the original document can be found in:
Document information
Published on 01/01/2012
Volume 2012, 2012
DOI: 10.2514/6.2012-5500
Licence: CC BY-NC-SA license
Share this document
Keywords
claim authorship
Are you one of the authors of this document?