HIGHLIGHT: An Operational Evaluation of a Ground Delay Program (GDP) Decision Support Tool using an Integrated Forecast of Stratus Clearing at San Francisco Airport

The primary challenge to air traffic management in the San Francisco area is the low altitude cloud layer that develops overnight in the San Francisco Bay area. This layer is called marine stratus and has a tremendous impact on San Francisco Airport (SFO) arrivals since it precludes simultaneous arrival operations on closely spaced parallel runways, reducing the arrival capacity from 60 to 30 flights per hour. Frequently, the stratus layer fails to burn off before the first bank of arrivals are scheduled to land in the morning, and a Ground Delay Program (GDP) must be implemented. During periods when arrival demand exceeds capacity, a GDP defers the excess demand to later periods with available capacity by assigning ground delays to flights at their departure airports. If a GDP is issued with an end time later than the actual stratus burn-off time, unnecessary delay is absorbed on the ground at the departure airports, and arrival capacity is wasted. On the other hand, if the GDP end time is earlier than the stratus burn-off time, airborne holding and diversions may occur, and safety may be compromised. The latter is a risk that air traffic controllers prefer to avoid, thus leading to current practices that tend to result in overly-conservative GDP end times. The frequency of marine stratus occurrence, combined with the tremendous aviation impact, motivated the National Weather Service (NWS) to develop the SFO Marine Stratus Forecast System, which is designed specifically to predict the important stratus clearing time in the SFO approach zone.

Analysis showed that despite the deployment of the SFO Stratus Forecast System in 2004, GDP practices continued as before, with no measurable benefits. The conclusion was that the probabilistic nature of the forecast product was difficult for human interpretation, and to take advantage of the forecast, a model was needed that could aid in the translation the probabilistic forecast into a Traffic Flow Management (TFM) decision.

The GDP Parameters Selection Model (GPSM) was thus motivated by research to explore the integration of probabilistic weather forecasts in traffic flow management (TFM) decision making. Operational decision-making in the National Airspace System (NAS) today is primarily based on deterministic information of traffic and weather, despite the fact that there is inherent uncertainty in both of these elements. Attempts at integrating probabilistic weather forecasts into operational decision-making have proven to be challenging. Convective weather forecasts in the en route environment include uncertainty in multiple dimensions, including time, space, and severity. This requires complex models to integrate probabilistic weather forecasts with TFM decision making. In contrast, the SFO Marine Stratus Forecast System provides an opportunity to explore the integration of probabilistic weather forecasts into TFM decision-making with the reduced complexity of one dimension. This system provides a forecast of a single weather parameter (the time at which the stratus will burn off) at a fixed geographical location (the SFO approach zone).

The GPSM integrates the probabilistic forecast of stratus clearing into the current process of modeling and issuing GDPs at SFO. By utilizing the probabilistic nature of the forecast, the model can select the best GDP parameters given the uncertainty in the time of the stratus burn-off, addressing both the objectives of minimizing delay and managing risk. By using the probabilistic forecast of stratus burn-off at SFO more effectively, GDPs in today's environment can be issued less conservatively, minimizing the overall ground delay, unused arrival slots, unnecessary delay issued, and the number of aircraft affected by the GDPs. This model is an important step toward integrating probabilistic weather forecasts with TFM decision support tools.

On May 15, 2012, the FAA, along with the wider operational community, began an operational evaluation of GPSM. The tool was used during the GDP decision process during the five-month stratus season in 2012 and was available to all involved parties, including the Air Traffic Control System Command Center (ATCSCC), Oakland Center (ZOA), Northern California TRACON (NCT), SFO Tower, Center Weather System Unit (CWSU), Monterey National Weather Service (NWS), and any operators that fly into SFO airport. In this paper we will present the outcome of this evaluation, both in terms of quantitative measurements and qualitative assessments. Lessons learned from the development and operational testing of this tool will be shared, the first of its kind in terms of fully integrating a weather forecast with air traffic management decision support.