DIFFERENCES IN RECORDINGS OF COLLOCATED IDENTICAL SENSORS

It is widely assumed that calibrated, identical instruments operating under similar conditions record the same signal with only minor differences due to random noise; hence recordings of a single instrument are fully representative of a specific signal. However, experience from measurements of bridge oscillations of sub-cm level with collocated GPS receivers, at the limit of the instrument range of operation, revealed important differences in the instrument output, and this result led us to investigate this problem on the basis of systematic experiments. The latter were based on sets of different types of receivers focusing on short-duration (a few sec to a few hours) high-frequency (10Hz) measurements. The output of these experiments was to confirm the existence of differences between identical, collocated instruments, even an absence of correlation between their recordings, which was not, however, due to phase shifts. Spectral analysis revealed that the observed differences are due to colored noise for low frequencies, and white noise for high frequencies. The limit between colored and white noise seems to be a function of the duration of observations, and for this reason, long-term observations are practically contaminated by white noise only and hence permit mm-level accuracy. This suggests that certain instruments, especially when functioning close to the upper limit of their range, may be affected by an instrument-dependent, (non-systematic) colored noise. The above result is very important because GPS is used as a standard for the measurement of time and of frequency, while differences between identical collocated sensors have been found also in other types of instruments recording dynamic measurements.