Extended Abstract (384K)J6.6
Integration of terminal area probabilistic meteorological forecasts in NAS-wide traffic flow management decision making
George Hunter, Sensis Corporation, Campbell, CA; and K. Ramamoorthy
Many of the advanced air traffic management and traffic flow management concepts envisioned for future implementation require advanced meteorological data products. Weather has several important effects on system capacity so it is important to integrate these effects into the air traffic management decision making. This means that a wide range of automated meteorological forecast products, with forecast horizons ranging up to several hours, will be required for use in air traffic management tools.
Meteorological forecasts are uncertain and while improved forecast accuracy is desired, it is important for the uncertainty of the forecasts to be described for proper use in air traffic management forecasting and decision making. This is the subject of on-going research and key questions include: How should forecast uncertainties be described? What update rates and forecast horizons are required or desirable? How sensitive is the performance of the air traffic management to forecast accuracy? In this paper we focus on meteorological effects in the terminal area and present our recent research results.
The air traffic management research community has developed detailed methods and practices for modeling terminal area and airport capacity. These methods model airport capacity in both visual and instrument conditions. Whether visual or instrument conditions prevail depends on the local ceiling and visibility distances. Beyond this, there are several meteorological phenomena that can reduce airport capacity.
High head winds can reduce final approach ground speed, and therefore inter arrival spacing and capacity. High cross winds make airport operations more difficult and reduce capacity. Very high winds can shutdown the airport altogether. Similarly, very low ceiling and visibility can shutdown the airport arrival capacity.
Heavy thunderstorms can also shutdown the airport if overhead, and otherwise can reduce the terminal area airspace capacity, or interfere with arrival or departure traffic streams. And local lightning can slow or shutdown ramp activities such as refueling. Degraded airport surface conditions, such as wet pavement, slows braking and turning, tends to increase runway occupancy time and so can reduce runway capacity. En route icing conditions causes departure delays for deicing operations. On the positive side, emerging wake vortex sensing and tracking technologies may be able to increase airport capacities in the future, in favorable conditions.
While all these effects are important, in most cases the most persistent effects are the local ceiling and visibility, and surface winds. In this paper we describe how we have used forecasts of these phenomena in our probabilistic traffic flow management experiments. And we provide preliminary results.
There are two key questions that must be addressed when using meteorological forecasts, such as ceiling, visibility and winds, in an air traffic management planning and decision support tool. First, the impact of the meteorological phenomena on the system capacity must be understood. Models must be developed that translate the meteorological data to capacity data. These models are useful both for understanding the capacity under known conditions, such as the current capacity of an airport, as well as predicting the capacity under forecasted conditions.
Second, the uncertainty of the meteorological forecast must be understood. Models must be developed that characterize the forecast as a random variable rather than a deterministic value. These models should account for all available information that is relevant and influences the forecast accuracy. These may include, for example, a forecast accuracy, confidence or skill parameter, meteorological descriptions of the structure and intensity of the forecasted weather, the forecast look ahead time, seasonal and regional effects, and so forth. The end result is a stochastic version of the meteorological forecast, such as in the form of a probability distribution function.
In addressing these two key questions we use both rational (model-driven) and empirical (data-driven) approaches because both have strengths and weaknesses. The weakness of purely rational approaches is that they are oblivious to real-world mechanisms that are not captured in the model. The weakness of purely empirical approaches is that they rely on the particular scenarios in which the data were collected. Those scenarios may not be generally descriptive. For instance, empirical approaches cannot measure capacity, they can merely measure traffic throughput, and it is difficult to know whether the throughput was capacity-limited or demand-limited. Throughput may appear to be low merely because demand was low. Also, it is likely that pilots sometimes avoid weather not because they must, but because they can. Sometimes avoiding the weather is a preference rather than a requirement. This suggests that weather impacted capacity can be fuzzy, or alternatively that it is higher than the measured throughput indicates when alternatives are easily available.
The impact of final approach headwinds on airport capacity serves as an example of how we use both rational and empirical approaches. Headwinds cause the aircraft groundspeed to decrease. Therefore, for aircraft flying at a given airspeed profile, with given in-trail spacing on final approach, the presence of a headwind increases the inter-arrival time spacing, and thus reduces the runway throughput. This model compares well with the empirical throughput data, though it predicted slightly higher capacity reduction. This could be explained by several possible mechanisms not captured in the model. For instance, pilots sometimes increase airspeed in the presence of high headwinds. The model is easily adjusted to better fit the data.
In this paper we present our research in modeling ceiling, visibility and surface winds effects and forecasts. We present how we use these models in our probabilistic traffic flow management decision support tool with results and performance sensitivities.
Joint Session 6, Next Generation Air Transportation System (NextGen) Part II
Wednesday, 23 January 2008, 10:30 AM-12:00 PM, 226-227
Previous paper