Application of the Allan Variance to Time Series Analysis in Astrometry and Geodesy: A Review

TL;DR

This review discusses the application and modifications of the Allan variance for analyzing weighted, multi-dimensional, and unevenly sampled time series in astrometry and geodesy over the past 50 years.

Contribution

It summarizes the development and use of AVAR and its variants, highlighting their adaptations for complex geodetic and astrometric data analysis.

Findings

AVAR has been increasingly used in geodesy and astrometry.

Modified AVAR methods address data weighting and multi-dimensionality.

These methods improve noise assessment in complex time series.

Abstract

The Allan variance (AVAR) was introduced 50 years ago as a statistical tool for assessing of the frequency standards deviations. For the past decades, AVAR has increasingly being used in geodesy and astrometry to assess the noise characteristics in geodetic and astrometric time series. A specific feature of astrometric and geodetic measurements, as compared with the clock measurements, is that they are generally associated with uncertainties; thus, an appropriate weighting should be applied during data analysis. Besides, some physically connected scalar time series naturally form series of multi-dimensional vectors. For example, three station coordinates time series $X$, $Y$, and $Z$ can be combined to analyze 3D station position variations. The classical AVAR is not intended for processing unevenly weighted and/or multi-dimensional data. Therefore, AVAR modifications, namely weighted AVAR (WAVAR), multi-dimensional AVAR (MAVAR), and weighted multi-dimensional AVAR (WMAVAR), were introduced to overcome these deficiencies. In this paper, a brief review is given of the experience of using AVAR and its modifications in processing astro-geodetic time series.