BEFORE THE APALS CAN be used, an approach database must be created. This is accomplished by flying the approach with the APALS in data-collecting mode. The radar collects SAR images of the terrain either side of the approach path. These are combined with aerial photographs and ground surveys to identify unique clusters of ground features, which can be used for correlation.

Each cluster consists of 150-200 datapoints corresponding to radar reflectors such as buildings, bridges and culverts. The database for an approach - which can be straight, segment or curved, Lockheed Martin says - consists of around 60 clusters, and occupies some 200kbytes of memory on the APALS hard disk.

Typically 30-35 correlations are performed during an approach. The system can withstand two miscorrelations, but will advise the pilot of degraded accuracy if three fail in succession. Up to 10% corruption of the database - caused by the construction of a building or removal of a hill, for example - can be tolerated.

Once the APALS is operational, each approach will be recorded and the data downloaded regularly to check the correlation performance. If a database appears to be giving problems, it will be updated. Lockheed Martin is working to interest database suppliers Jeppesen and Swissair in maintaining the APALS databases.

Viewed on the consoles in the GII cabin, the system's performance is impressive. On a deliberately botched approach, the system detected when the aircraft reached the boundaries of the 95% required navigation-performance tunnel used to define the acceptable limits of the approach path. When the aircraft was turned on to the approach over the outer marker, the system moved from its initial fix to maximum accuracy within seconds.

Source: Flight International