shows the static tree. It is a symmetrical tree with four potential entry points, each equidistant from two
downstream merge points and from the meter fix. Aircraft enter the tree at any of these points temporally
spaced so they would reach the meter fix appropriately spaced if they followed the static tree. The
Dynamic and Dynamic Adjusted conditions both used Metron’s simple tree planner algorithm with fixed
quadrant size, fixed RNP, and a single TRACON entry meter fix (Prete, Krozel, Mitchell, Kim & Zou,
2008). As can be seen in Figure 1, the tree planner warps the routing, with the goal of a route that will
allow aircraft to avoid the weather while in the tree. The tree adapts to the predicted weather every 15
minutes so the controller will have as many as three different simultaneous tree structures. The difference
between the Dynamic and Dynamic Adjusted conditions arises from the way aircraft are initially spaced
as they enter the scenario. This factor was introduced because creating dynamic trees has interesting
consequences for arrival spacing. Due to the warping of routes in a dynamic tree, and the fact a
preceding aircraft may enter on any of the four entry points, an entering plane might have a very different
distance to fly to the meter fix than the plane preceding it, making the spacing problem at merge points
and the meter fix more difficult. In principle, under TBO such path stretches need not cause problems.
Aircraft could know of weather deviations for the plane entering the tree just ahead of them and adjust
their speed to arrive at the entry points appropriately spaced. As a result, we implemented the dynamic
trees in two ways: “Dynamic” and “Dynamic Adjusted.” In the Dynamic condition, aircraft arrive at the
outer leaves of the tree at intervals that are unaffected by the path length to the meter fix. In the Dynamic
Adjusted condition, aircraft on shorter paths are delayed relative to those on longer paths. Because the
difference between Dynamic and Dynamic Adjusted was subtle, participants were only informed as to
whether the condition was Dynamic to avoid biasing them.
Figure 1: Examples of Static and Dynamic tree structures, with weather. Notice how the STAR is
adjusted in the Dynamic tree to avoid the weather.
when weather impacted the path at the moment the aircraft passed. Note that no attempt was made to
synchronize the weather with individual aircraft during scenario development.
Figure 2: “Tukey” boxplot (Tukey, 1973) of the distance flown to the runway by the experimental planes
in each condition. For each condition the solid line indicates the median; the box indicates the inter-
quartile range (IQR); the whiskers indicate the range excluding “outliers”; and the circles indicate outliers.
points more than 1.5 IQR from the box. The horizontal line indicates the distance flown (183 nm) on the
static tree.