89th American Meteorological Society Annual Meeting

Tuesday, 13 January 2009: 2:30 PM
Analysis of Probabilistic Weather Forecasts for Use in Air Traffic Management
Room 132A (Phoenix Convention Center)
Kapil Sheth, NASA Ames Research Center, Moffett Field, CA; and T. Amis and S. Gutierrez-Nolasco
Poster PDF (2.1 MB)
Based on air traffic delay results from the Federal Aviation Administration's (FAA) Operations Network (OPSNET) data, more than 70% of the National Airspace System (NAS) reportable delays were attributed to severe weather. The Weather-ATM Integration Working Group of the Joint Planning and Development Office has suggested that probabilistic convective weather forecasts be included in the System-Wide Information Manager for the Next Generation of Air Transportation System. At the current time, deterministic products do not provide accurate weather data in the two to six hour time frame, which creates inefficiencies in the decision-making process for that longer term. However, probabilistic convective weather forecasts are available about three times each hour for up to a six-hour period, but additional understanding is desired and significant scientific development is required for use of these data for air traffic management purpose.

In earlier research conducted at NASA, probabilistic convective weather forecasts were studied for applications of flight routing decisions in air traffic management. The previous work was conducted over the entire continental United States, which was beneficial for understanding the aggregate behavior. The objective of that study was to derive the threshold of probability that could be used by decision makers for circumventing severe weather. By synchronizing air traffic data and probabilistic convective weather forecasts, it was observed, as expected, that a majority of aircraft were avoiding certain probability contours. The analysis showed that for a one-hour forecast, the 80th percentile value for all aircraft traversing the entire probability field was around 35%. The same probability cut-off value for a two-hour forecast was around 25%. The values for three and four-hour data were around 18-20%, while the five and six-hour data fell below 10%. A simple case of routes for one aircraft avoiding a 10%, 35% and 60% probability contours for a one-hour forecast was presented in terms of the additional length and time of flight. With such a process, a decision-maker could assess the risk associated with traveling in the vicinity of forecasted severity of weather contours.

The objective of current research, the probabilistic weather data are similarly analyzed with a more focused scope in the vicinity of the Fort Worth Air Route Traffic Control Center (ZFW ARTCC) to quantify benefits in a localized region. Air traffic data from the FAA's Enhanced Traffic Management System (ETMS) and probabilistic convective weather forecasts from the National Center for Atmospheric Research's (NCAR) Consolidated Storm Prediction for Aviation (CoSPA) product for the week of July 23, 2007 were used for this analysis. The Future ATM Concepts Evaluation Tool (FACET), a simulation and modeling system for analysis of air transportation concepts, developed at NASA was utilized for this purpose. It was observed that in this small data set, the aircraft demonstrate similar behavior in terms of probability cut-off values, as in the previous research. For the data analyzed, it was observed that the 80th percentile value for all aircraft in the one-week period between flight level (FL) 300 to 350 and between FL 350 to 400 was about 28% (a slight reduction from the earlier observations). These data were examined for different altitude levels, users of airspace and aircraft types for this small data set. A histogram of the frequency of probability of forecasted weather data will be presented in the final paper, along with extending this analysis to a larger data set consisting of several months of data. The objective is to assess the probability cut-off parameter for various temporal forecasts in the ZFW region for flight routing decisions process.

Once these threshold values of probability are available, what-if evaluation analysis will be performed for various route options, in the FACET simulation environment. For example, if a fix for arrival traffic (e.g., Ardmore, ADM) for Dallas/Ft. Worth airport or overflight traffic transitioning through the ZFW Center is forecasted to be impacted by convective weather in the next one through six hours, what kind of rerouting schemes could be employed (e.g. reroute flows from Texarkana, TXK, through Wichita Falls, SPS to Childress, CDS) for alleviating the weather effects, including airport closure or playbook routes from the FAA published Severe Weather Avoidance Plan. Thus, another contribution of this paper would be to provide a decision-support algorithm for ranking of options for flight routing decisions.

Supplementary URL: