Geometric estimates of variations in space geodetic data
Abstract Well-known statistical parameters have some disadvantages when analyzing space geodetic data. Geometric parameters are proposed here for estimating the variation properties of samples for various discrete datasets. The proposed parameters are logically related to each other and are based on the simplest well-known statistical parameters; they do not depend on the type of distribution of the sample under study. “Variation asymmetry” shows the shift of the arithmetic mean relative to the center of the variation interval in the units of the studied sample. “Density of variation” characterizes the level of average variability in sample units. This parameter has several times greater discriminatory sensitivity to extremely different types of variations than linear and standard deviations. The relative parameter “proportion of maximum density” shows the closeness of variation to a uniform distribution in the ranked sample and complements the indicator of variation density. An algorithm for separating different structural levels of the useful signal from emissions (noise) is proposed here based on the calculation of geometric characteristics. The iterations of dividing the sample into structurally homogeneous segments can be stopped at the level of the proportion of maximum density ≥0.9 when analyzing real GPS coordinates.