My last post talked about some of the problems with finding out exactly where an aircraft lands. Now I will start to look at how the system works in practice.
Pick an Airfield
I will skip over the detail of how to identify the point of landing (this was covered in an earlier blog), so we know when the touchdown happened. On nice aircraft with a recorded latitude and longitude, we can find out where the aircraft landed. Armed with this, we just do a closest airport lookup to find which airport the aircraft landed on. Nearly all cases are found by this method, but it’s always the exceptions which cause more interest.
No Recorded Position
The first problem arises where the aircraft does not record latitude and longitude, so all we know is the date and time of landing. In this case we ask the operator to provide a list of the flights flown, an “Achieved Flight Record” (AFR), which tells us where the aircraft flew from and to for each flight. We then check the takeoff date and time against the AFR records and select the record with the correct takeoff time. This record then tells us which airport the aircraft landed on.
Close, but Not Close Enough
The technique can also go wrong if the navigation system errors are so large that the closest reported airport is not the one they landed on. For example, some years ago we monitored data for a Boeing 757 that flew a weekly service from England to Nigeria and back. On the homeward leg into Gatwick the inertial navigation system drift was large enough to make Redhill Aerodrome* (about 5 miles North of Gatwick) the closest airport in the database. This is what Redhill looks like:
Photo courtesy UKGA
Although the aerodrome has three runways and lots of helicopter pads, grass is not a good surface to land 100 tonnes of Boeing 757!
There are two ways to deal with this, namely the fast and dirty fix and the correct algorithmic solution. The fast and dirty fix was to delete Redhill from the airport database so that Gatwick became the closest airport to the landing point and the problem went away. The correct algorithmic solution is to determine whether the airport in question has runways with suitable surface and length to land an aircraft of a given type. One day we’ll get around to doing this, but in practice the problem is rarely encountered and this problem is fading out as older aircraft go out of service.
*This is where I used to work when I was a member of the Bristow Helicopters design department. We worked in a Second World War hut and during a storm I noticed the walls were moving relative to the ceiling. Eventually the end wall blew so far that the ceiling lost contact and dropped, sagging across the entire width of the building. The hut was subsequently demolished which is why I can’t point to my office in the photograph.