Database Reference
In-Depth Information
expert, a part of them are flights without passengers used for relocating aircraft
between big airports, such as Charles de Gaulle and Orly. Short-distance flows
between small airports correspond to training and leisure flights of private pilots.
Focusing on the long-distance flows (100 km and more) reveals a mostly radial
connectivity scheme with a center in Paris.
To investigate the temporal dynamics of the flows, we use the table display
as shown in Figure
12.6
e. The columns of the table correspond to the hourly
time intervals and the rows to the flows. The lengths of the colored bar segments
in the cells are proportional to the flight counts for the respective flows and
intervals. The colors correspond to the eight compass directions. The table view
is linked to the flowmap. Thus, clicking on the vectors connecting Paris Orly and
Marseille on the map, we get two rows highlighted. The yellow one corresponds
to the northwestern direction, that is, fromMarseille to Paris, and the orange one
to the opposite direction, from Paris to Marseille. There are one or two flights
fromMarseille to Paris every hour in the intervals 07-14h and 15-18h and three
flights per hour from 22h to midnight. The traffic in the opposite direction has a
different profile: three flights per hour from midnight till 02h and several flights
in the morning, at noon, and in the evening. The complementary link from the
table view to the map can be used to locate flows with particular dynamics.
12.5.2 Validation of the Findings
First, to assess the validity of the extracted areas of takeoffs and landings, we
compared them with the known positions of the airports and found that the areas
include the airports. Furthermore, the areas have elongated shapes (Figure
12.6
d)
whose spatial orientations coincide with the orientations of the runways of the
respective airports. Next, the results of data aggregation by the areas (i.e., counts
of takeoffs, landings, and flights between airports) correspond very well to the
common knowledge about the sizes and connectivity of the French cities and
airports. The discovered patterns have been also checked and interpreted by a
domain expert who confirmed their plausibility.
12.6 Complex Pattern Extraction Using a Moving Object
Database System
Moving object database systems are another good candidate for air traffic anal-
ysis. This section demonstrates a concrete example of using the SECONDO MOD
system in order to extract complex spatio-temporal patterns from the flight trajec-
tories. The task is to extract the
missed approach
and the
stepwise descent
events
that occurred in the ATC data set described in Section
12.3
.The
spatio-temporal
pattern
(
STP
)
algebra
in SECONDO brings a generic set of query operations acces-
sible through the SECONDO query languages to let the user express arbitrarily
