Filtering Digital Profile Observations

The elimination of noise in digital terrain profile data requires the analysis of these data in the frequency domain. Data processing in the frequency domain involves inferences regarding the periodogram interpretation. A periodogram is an estimate of the spectral density function of a given set of data. This estimate describes the frequency properties of these data. Knowing the frequency characteristics of random errors, it is possible to discriminate them from the "true" elevations of a terrain profile. The presented method of filtering depends on the fact that random errors have a nearly constant periodogram over all the frequency range. On the other hand, the periodogram of smooth data is concentrated at low frequencies and decreases rapidly when moving towards the high frequency region. The periodogram of actual terrain profiles represents a combination of both properties; however, it is always possible to distinguish between the zone that represents noise and the region representing true values. It was found, throw this investigation, that it is possible to modify the periodogram of a given set of data such that the zone representing noise is eliminated. The reconstruction of data from the modified periodogram is then performed to obtain a new set of data which is believed to represent more accurately the true terrain profile.