Integration of airborne gravimetry data filtering into residual least-squares collocation: example from the 1 cm geoid experiment

Low-pass filters are commonly used for the processing of airborne gravity observations. In this paper, for the first time, we include the resulting correlations consistently in the functional and stochastic model of residual least-squares collocation. We demonstrate the necessity of removing high-frequency noise from airborne gravity observations, and derive corresponding parameters for a Gaussian low-pass filter. Thereby, we intend an optimal combination of terrestrial and airborne gravity observations in the mountainous area of Colorado. We validate the combination in the frame of our participation in ‘the 1 cm geoid experiment’. This regional geoid modeling inter-comparison exercise allows the calculation of a reference solution, which is defined as the mean value of 13 independent height anomaly results in this area. Our result performs among the best and with 7.5 mm shows the lowest standard deviation to the reference. From internal validation we furthermore conclude that the input from airborne and terrestrial gravity observations is consistent in large parts of the target area, but not necessarily in the highly mountainous areas. Therefore, the relative weighting between these two data sets turns out to be a main driver for the final result, and is an important factor in explaining the remaining differences between various height anomaly results in this experiment.