In order for this to be from an error in measurement you need to be a few meters off (18 meters if that’s the only problem). There are standard GPS techniques and surveying techniques which can be used to get very precise values. They state in the paper and elsewhere that they are confident to around 30 cm. Differential GPS can have accuracy down to about 10-15 cm, and careful averaging of standard GPS can get you in the range of 20 cm, so this isn’t at all implausible but it is still a definite potential source of error.
A more plausible issue is that since parts of the detectors are underground they didn’t actively use GPS for those parts. But even then, a multiple meter error seems unlikely, and 18 meters is a lot. It is possible that there’s a combination of errors all going in the same direction, say a meter error in the distance, a small error in the clock calibration, etc. And all of that add up even as each error remains small enough that it is difficult to detect. But they’ve been looking at things really closely so one would then think that at least one of the errors would turn up.
In order for this to be from an error in measurement you need to be a few meters off (18 meters if that’s the only problem). There are standard GPS techniques and surveying techniques which can be used to get very precise values. They state in the paper and elsewhere that they are confident to around 30 cm. Differential GPS can have accuracy down to about 10-15 cm, and careful averaging of standard GPS can get you in the range of 20 cm, so this isn’t at all implausible but it is still a definite potential source of error.
A more plausible issue is that since parts of the detectors are underground they didn’t actively use GPS for those parts. But even then, a multiple meter error seems unlikely, and 18 meters is a lot. It is possible that there’s a combination of errors all going in the same direction, say a meter error in the distance, a small error in the clock calibration, etc. And all of that add up even as each error remains small enough that it is difficult to detect. But they’ve been looking at things really closely so one would then think that at least one of the errors would turn up.