ORIGINAL ARTICLE
Proposing a concept of least-squares-based outlier-exposing potential of Gauss-Markov models: Examples in geodesy
More details
Hide details
1
Faculty of Geodesy and Cartography, Warsaw University of Technology, Pl. Politechniki 1, 00-661, Warsaw, Poland
A - Research concept and design; B - Collection and/or assembly of data; C - Data analysis and interpretation; D - Writing the article; E - Critical revision of the article; F - Final approval of article
Submission date: 2024-07-20
Final revision date: 2024-11-01
Acceptance date: 2024-11-05
Publication date: 2024-11-29
Corresponding author
Prószyński Witold
Faculty of Geodesy and Cartography, Warsaw University of Technology, Pl. Politechniki 1, 00-661, Warsaw, Poland
Reports on Geodesy and Geoinformatics 2024;118:101-108
W.P. theory and analyses, figure design; Z.P. computations, preparation of figures.
KEYWORDS
TOPICS
ABSTRACT
Outlier detection and identification are still important issues in the quality control of geodetic networks based on least squares estimation (LSE). In addition to existing network reliability measures, the paper proposes the LSE-based concept (together with the associated measures) of the Outlier-Exposing Potential (OEP) for Gauss-Markov models. The greater the model's redundancy, the more the configuration of its responses to gross errors exposes the location of these errors, and hence, the greater the model's OEP. The potential is given in the basic version and the extended version. The former considers only the effect of the model's redundancy, while the latter also considers the masking effect due to random observation errors at a specified magnitude of gross error. For models that have regions of unidentifiable errors, the corresponding OEP components have zero values. The reflection of OEP in the values of Minimal Identifiable Bias (MIB) is shown. It is proposed that OEP derived based on least squares adjustment be treated as a property of the model itself. The theory is illustrated on several 1D and 2D networks. The research is limited to models with uncorrelated observations and the case of a single gross error. These limitations enabled the formulation of clear properties of general character, not complicated by observation correlations and multiple-outlier combinations.
ACKNOWLEDGEMENTS
The authors would like to express their thanks to Dr. Krzysztof Nowel from the University of Warmia and Mazury in Olsztyn for sharing his original MDB_MIB computation program.
REFERENCES (18)
1.
Baarda W. (1968). A testing procedure for use in geodetic networks. Technical report. Netherlands Geodetic Commission. Publications on Geodesy, New Series. 2 (5).
2.
Chatterjee Samprit, Hadi Ali S. (2009). Sensitivity analysis in linear regression. John Wiley & Sons.
3.
Ding Xiaoli, Coleman Richard. (1996). Multiple outlier detection by evaluating redundancy contributions of observations. Journal of geodesy. 70: 489-498. doi:10.1007/BF00863621.
4.
Durdag Utkan Mustafa, Hekimoglu Serif, Erdogan Bahattin. (2022). What is the relation between smearing effect of least squares estimation and its influence function?. Survey Review. 54 (385): 320-331. doi:10.1080/00396265.2021.1939590.
5.
Förstner Wolfgang. (1983). Reliability and discernability of extended Gauss-Markov models. Deut. Geodact. Komm. Seminar on Math. Models of Geodetic Photogrammetric Point Determination with Regard to Outliers and Systematic Errors, Munchen, Germany. 79-104.
6.
Huber Peter J. (1972). Robust statistics: A review. The Annals of Mathematical Statistics. 43 (4): 1041-1067. doi:10.1214/aoms/1177692459.
7.
Imparato D, Teunissen PJG, Tiberius CCJM. (2018). Minimal detectable and identifiable biases for quality control. Survey review. 51 (367): 289-299. doi:10.1080/00396265.2018.1437947.
8.
Lehmann Rüdiger. (2013). On the formulation of the alternative hypothesis for geodetic outlier detection. Journal of geodesy. 87: 373-386. doi:10.1007/s00190-012-0607-y.
9.
Lehmann Rüdiger, Lösler Michael, Neitzel Frank. (2020). Mean shift versus variance inflation approach for outlier detection – A comparative study. Mathematics. 8 (6): 991-991. doi:10.3390/math8060991.
10.
Maronna Ricardo A, Martin R Douglas, Yohai Víctor J. (2006). Robust Statistics: theory and methods. John Wiley & Sons.
11.
Prószyński Witold. (1994). Criteria for internal reliability of linear least squares models. Bulletin géodésique. 68: 162-167. doi:10.1007/BF00808289.
12.
Prószyński Witold. (2008). The vector space of imperceptible observation errors: a supplement to the theory of network reliability. Geodesy and Cartography. 57 (1): 3-19.
13.
Prószyński Witold. (2012). Internal robustness of linear models to disturbances in data – uncorrelated and correlated observations (in Polish). Warsaw University of Technology.
14.
Prószyński Witold. (2015). Revisiting Baarda’s concept of minimal detectable bias with regard to outlier identifiability. Journal of geodesy. 89: 993-1003. doi:10.1007/s00190-015-0828-y.
15.
Rofatto Vinicius Francisco, Matsuoka MT, Klein I, Veronez MR, Bonimani ML, Lehmann Rüdiger. (2020). A half-century of Baarda’s concept of reliability: a review, new perspectives, and applications. Survey review. 52 (372): 261-277. doi:10.1080/00396265.2018.1548118.
16.
Teunissen PJG. (2006). Testing theory: an introduction. VSSD.
17.
Teunissen Peter JG, Imparato Davide, Tiberius Christian CJM. (2017). Does RAIM with correct exclusion produce unbiased positions?. Sensors. 17 (7): 1508-1508. doi:10.3390/s17071508.
18.
Teunissen Peter JG. (2018). Distributional theory for the DIA method. Journal of geodesy. 92 (1): 59-80. doi:10.1007/s00190-017-1045-7.