Pilot Decision Making on Weather Avoidance

Effective management of air traffic in congested airspace during episodes of convective weather requires decision support tools that translate weather forecasts into air traffic impacts and then use those ATC impact forecasts to suggest air traffic management strategies. A critical first step in the translation process is a validated model for airspace that pilots will seek to avoid.

Convective Weather Avoidance Model (CWAM1)

In research funded by NASA Ames Research Center, Lincoln Laboratory developed an enroute Convective Weather Avoidance Model (CWAM1 – see Figure 1) that outputs three-dimensional weather avoidance fields.

pilot view of thunderstorm cloudPilot view of thunderstorms.

The probabilistic Weather Avoidance Fields identify regions of airspace that pilots are likely to avoid due to the presence of convective weather (Figure 2). The model used Corridor Integrated Weather System (CIWS) Vertically Integrated Liquid (VIL) and echo top fields and National Lightning Detection Network (NLDN) data to predict aircraft deviations and penetrations. The statistical results showed that the difference between flight altitude and the radar storm top was the most important factor in explaining pilot deviations. The second most important factor was the precipitation intensity.


A second study (CWAM2) extended the analysis to additional Centers (ZDC, ZID and ZOB – see Figure 3) and included several additional deviation predictors. The additional predictors captured information about storm growth and decay, vertical structure and weather type (convective or non-convective). Even with all the additional information, the difference between flight altitude and radar storm top was again the top predictor of pilot deviation to avoid convective weather.

Deviation Prediction Errors

Most deviation prediction errors occurred for flights that encountered echo tops near the flight altitude, for which pilots can legitimately make different choices. Figure 4 illustrates two flights on the same route at roughly the same flight altitude, 10 minutes apart. The first pilot deviates widely to avoid the weather, while the second flies extremely close to the high storm tops. The model correctly predicted the second pilot’s behavior, but not the first.

Ongoing Research

A third CWAM study (CWAM3) is currently being planned. It will be the first to include operational information, such as time of day (daylight, twilight, night), aircraft type, airline, airspace congestion, etc., as potential predictors of deviation. Lincoln is also investigating the visual cues available in the cockpit to gain a better understanding of which aspect of the weather the pilot considers hazardous.

top of page