Scoring of Convective Precipitation Forecasts Used for Air Traffic Flow Decision Making

 

James Evans*, Senior Staff

MIT Lincoln Laboratory, Lexington MA

 

Convective forecasts for use in FAA/airline traffic flow management have become increasingly important in the past few years as a result of the dramatic increase in aviation delays in the summer months.  Much of the scoring of these forecasts to date has used conventional grid based approaches with binary comparisons using conventional measures such as POD, FAR and skill.

 

However, the aviation application is quite different from the conventional weather service application of precipitation forecasting where one is typically interested in precipitation at a point.  The key operational needs for forecasts in traffic flow management applications include:

 

1.      choice of routes (both before takeoff and when in flight)

2.      traffic flow management actions including

a.      number of aircraft allowed to use en route sectors and/or terminals at various time intervals in the future (i.e., capacity estimation)

b.      miles in trail (MIT) spacing on routes

c.      ground stops for departures

 

The route choice and capacity estimation decisions cannot be easily related to point location precipitation statistics.   Hence, there is a need to develop new criteria to be used to evaluate aviation forecast accuracy. 

 

Additional complicating factors for the aviation application are:

 

(1)   the growing use of probabilistic forecasts,  and

(2)   handling of forecasts which provide as a part of the forecast estimates of the forecast accuracy

 

In this paper, we:

 

1.      show examples of contemporary aviation use of such forecasts,

2.      discuss some of the differences between point probabilities and route probabilities, and

3.       review past and ongoing work at evaluating forecasts in the context of the aviation route selection and capacity estimation applications

 

The paper will conclude with some suggestions for near term focused research to develop measures for these forecasts that are much more closely tied to the operational usage.



* email JIME@LL.MIT.EDU