Abstract
By considering a deformable geodetic network, deforming in a linear-in-time mode, according to a coordinate-invariant model, it becomes possible to get an insight into the rank deficiency of the stacking procedure, which is the standard method for estimating initial station coordinates and constant velocities, from coordinate time series. Comparing any two out of the infinitely many least squares estimates of stacking unknowns (initial station coordinates, velocity components and transformation parameters for the reference system in each data epoch), it is proven that the two solutions differ only by a linear-in-time trend in the transformation parameters. These pass over to the initial coordinates (the constant term) and to the velocity estimates (the time coefficient part). While the difference in initial coordinates is equivalent to a change of the reference system at the initial epoch, the differences in velocity components do not comply with those predicted by the same change of reference system for all epochs. Consequently, the different velocity component estimates, obtained by introducing different sets of minimal constraints, correspond to physically different station velocities, which are therefore non-estimable quantities. The theoretical findings are numerically verified for a global, a regional and a local network, by obtaining solutions based on four different types of minimal constraints, three usual algebraic ones (inner or partial inner) and the lately introduced kinematic constraints. Finally, by resorting to the basic ideas of Felix Tisserand, it is explained why the station velocities are non-estimable quantities in a very natural way. The problem of the optimal choice of minimal constraints and, hence, of the corresponding spatio-temporal reference system is shortly discussed.
Similar content being viewed by others
References
Ampatzidis D, König R, Glaser S, Heinkelmann R, Schuh H (2016) An alternative methodology for the assessment of the temporal evolution of global terrestrial reference frames. Ann Geophys (under review)
Altamimi Z, Sillard P, Boucher C (2002) ITRF2000: a new release of the international terrestrial reference frame for earth science applications. J Geophys Res 107(B10):2214
Altamimi Z, Collilieux X, Legrand J, Garayt B, Boucher C (2007) ITRF2005: a new release of the International Terrestrial Reference Frame based on time series of station positions and Earth orientation parameters. J Geophys Res 112(B09):401. doi:10.1029/2007JB004949
Altamimi Z, Collilieux X, Métivier L (2011) ITRF2008: an improved solution of the international terrestrial reference frame. J Geod 85:457–473
Altamimi Z, Dermanis A (2012) The choice of reference system in ITRF formulation. In: Sneeuw N, Novàk P, Crespi M, Sansò F (eds) VII Hotine-Marussi symposium on mathematical geodesy. IAG Symposia, vol 137. Springer, Berlin, pp 329–334
Altamimi Z, Dermanis A (2013) Theoretical foundations of ITRF determination. The algebraic and the kinematic approach. In: On measurements of lands and constructions. Volume in honor of Prof. D. Vlachos. School of Rural & Surveying Engineering, Aristotle University of Thessaloniki, pp 331–359
Baarda W (1973) S-Transformations and criterion matrices. Netherlands Geodetic Commission, Publ. in Geodesy, New Series, Volume 5, no. 1, Delft
Baarda W (1995) Linking up spatial models in geodesy. Extended S-Transformations. Netherlands Geodetic Commission, Publ. in Geodesy, New Series, no. 41, Delft
Bjerhammar A (1951) Rectangular reciprocal matrices with special emphasis to geodetic calculations. Bull Géodésique 1951:188–220
Blaha G (1971) Inner adjustment constraints with emphasis on range observations, Department of Geodetic Science, Report 148. The Ohio State University, Columbus
Blewitt G, Kreemer C, Hammond WC, Goldfarb JM (2013) Terrestrial reference frame NA12 for crustal deformation studies in North America. J Geodyn 72:11–24
Bruyninx C (2004) The EUREF Permanent Network: a multi-disciplinary network serving surveyors as well as scientists. GeoInformatics 7:32–35
Chatzinikos M (2013) Study of the earth’s crust displacements in the area of Greece analyzing GNSS data. PhD Thesis, School of Rural and Surveying Engineering, Aristotle University of Thessaloniki, Greece
Chatzinikos M, Dermanis A (2015) Visualization of horizontal station motions as a diagnostic tool for coordinate time series. Volume in honour Prof. S. Yannopoulos. Publication of the School of Rural & Surveying Engineering, Aristotle University of Thessaloniki, Thessaloniki (in print)
Chatzinikos M, Dermanis A (2016) A study on the impact of reference frame implementation strategy on GNSS time series for regional network analysis. In: Proceedings of the IAG Commission I Symposium. Reference frames for applications in geosciences, October 13–17 2014, Kirchberg, Luxemburg, IAG Symposia, vol. 146. Springer, Berlin (in print)
Dow J, Neilan RE, Rizos C (2009) The international GNSS service in a changing landcape of global navigation satellite systems. J Geod 83(3–4):191–198
Dermanis A (1987) Adjustment of observations and estimation theory, vol 2. Ziti, Thesalloniki (in Greek)
Dermanis A (2001) Global reference frames: Connecting observation to theory and geodesy to geophysics. In: IAG Scientific Assembly, 2–8 Sept. 2001, Budapest, Hungary. http://der.topo.auth.gr
Dermanis A (2003) The rank deficiency in estimation theory and the definition of reference frames. In: Sansò F (ed) V Hotine-Marussi symposium on mathematical geodesy. IAG Symposia, vol 127. Springer, Berlin, pp 145–156
Dermanis A (2015) Global Reference Systems: theory and open questions. Invited paper at the Academia dei Lincei Session, VIII Hotine-Marussi Symposium, Rome, Italy, 17–21 June 2013. In: Sneeuw N, Novàk P, Crespi M, Sansò F (eds) VIII Hotine-Marussi symposium on mathematical geodesy, IAG Symposia, vol 142, pp 9–16
Dermanis A (2016) Problems in parameter estimation with nonlinear models. In: Arvanitis A, Basbas S, Lafazani P, Papadopoulou M, Paraschakis I, Rossikopulos D (eds) Cartographies of Mind, Soul and Knowledge. Special issue for Emeritus Professor Myron Myridis. Publication of the School of Rural and Surveying Engineering, Aristotle University of Thessaloniki, pp 117–133. https://www.researchgate.net
Grafarend E, Schaffrin B (1974) Unbiased free net adjustment. Surv Rev 22:200–218
Grafarend E, Schaffrin B (1976) Equivalence of estimable quantities and invariants in geodetic networks. Zeitschrift für Vemessungswesen 101(11):485–491
Koch K-R (1999) Parameter estimation and hypothesis testing in linear models, 2nd edn. Springer, Berlin
Kotsakis C (2013) Generalized inner constraints for geodetic network densification problems. J Geod 87:661–673
Meissl P (1965) Über die innere Genauigkeit dreidimensionaler Punkthaufen. Zeitschrift für Vemessungswesen 90(4):109–118
Meissl P (1969) Zusammenfassung und Ausbau der inneren Fehlertheorie eines Punkthaufens. Deutsche Geodätische Kommission Reihe A Nr 61:8–21
Munk WH, MacDonald GJF (1960) The rotation of the earth. A geophysical discussion. Cambridge University Press, Cambridge
Pearlman MR, Degnan JJ, Bosworth JM (2002) The international laser ranging service. Adv Space Res 30(2):135–143
Pope A (1972) Some pitfalls to be avoided in the iterative adjustment of nonlinear problems. In: Proc. 38th Ann. ASPRS Meeting. Amer. Soc. of Photogrammetry, Falls Church, pp 449–477
Rao CR (1973) Linear statistical inference and its applications, 2nd edn. Wiley, New York
Rao CR, Toutenburg H, Shalabh Ch, Heumann (2008) Linear Models and Generalizations. Least Squares and Alternatives. Third Extended Edition. Springer-Verlag, Berlin, Heidelberg
Schlüter W, Behrend D (2007) The international VLBI service for geodesy and astrometry (IVS). Current capabilities and future prospects. J Geod 81(6–8):379–387
Schuh H, König R, Ampatzidis D, Flechtner F, Glaser S, Heinkelmann R, Nilsson T (2015) GGOS-SIM: simulation of the global geodetic observing system. In: Proceedings of the IAG Commission 1 Symposium 2014: reference frames for applications in geosciences (REFAG2014). IAG Symposia, vol 146, Springer, Berlin (in print)
Sternberg S (1964) Lectures on differential geometry. Prentice-Hall Inc., Englewood Cliffs
Tisserand F (1890) Traité de Mechanique Céleste, Tome II. Théorie de la Figure des Corps Célestes et de Mouvement de Rotation. Gautier-Villars, Paris
Willis P, Fagard H, Ferrage P, Lemoine FG, Noll CE, Noomen R, Otten M, Ries JC, Rothacher M, Soudarin L, Tavernier G, Valette JJ (2010) The international DORIS service, toward maturity. In: Willis P (ed) DORIS: scientific applications in geodesy and geodynamics. Advances in Space Research, vol 45(12), pp 1408–1420. doi:10.1016/j.asr.2009.11.018
Wu X, Abbondanza\(,\) C, Altamimi Z, Chin TM, Collilieux X, Gross RS, Heflin MB, Jiang Y, Parker JW (2015) KALREF—a Kalman filter and time series approach to the international terrestrial reference frame realization. J Geophys Res Solid Earth 120:3775–3802. doi:10.1002/2014JB011622
Author information
Authors and Affiliations
Corresponding author
Appendices
Proof of Proposition 2
Let us first note that the observables are always estimable quantities (Dermanis 2003) and, therefore, the least squares estimates \({{ \hat{\mathbf{x}}}}_{i\vert S_{k} } (t_{k} )={{ \hat{\mathbf{x}}}}_{i0\vert S_{0} } +(t_{k} -t_{0} ){{ \hat{\mathbf{v}}}}_{i\vert S_{0} } +\mathbf{{ E}}_{i} {{ \hat{\mathbf{p}}}}_{k} \) of the observables \({{ {\hat{\mathbf{x}}}'}}_{i} (t_{k} )={{ {\hat{\mathbf{x}}}'}}_{i0\vert S_{0} } +(t_{k} -t_{0} ){{ {\hat{\mathbf{v}}}'}}_{i} +\mathbf{{ E}}_{i} {{ {\hat{\mathbf{p}}}'}}_{k} \) will be unique. To simplify the notation, we remove here the hats as a symbol for estimates and the network point index i. All initial coordinates and velocities refer to the same point and they are estimates, same as the transformation parameters \(\mathbf{{ p}}_{k} \). We retain the index \(S_{0} \) to denote the reference system to which initial coordinates and velocities refer.
Therefore, for any two sets of least squares solutions to the stacking problem \({\mathbf{x}}_{0\vert S_{0} } \), \({\mathbf{v}}_{S_{0} } \), \(\mathbf{{ p}}_{k} \) and \(\mathbf{{ {x}'}}_{0} \), \(\mathbf{{ {v}'}}\), \(\mathbf{{ {p}'}}_{k} \) it holds that
At the same time, we also have two different estimates of \({\mathbf{x}}(t_{k} )={\mathbf{x}}_{0} +(t_{k} -t_{0} ){\mathbf{v}}\), namely
For \(t_{k} =t_{0} \), Eq. (A1a) gives \({\mathbf{x}}_{S_{k} } (t_{0} )={\mathbf{x}}_{0\vert S_{0} } +\mathbf{{ Ep}}_{0} =\mathbf{{ {x}'}}_{0} +\mathbf{{ E{p}'}}_{0} \) and thus
This means that \(\mathbf{{ {x}'}}_{0} \) describes the same network shape as \({\mathbf{x}}_{0\vert S_{0} } \), but in a different reference system \(\tilde{{S}}_{0} \), where \(\mathbf{{ p}}_{0} -\mathbf{{ {p}'}}_{0} \) are the transformation parameters from \(S_{0} \) to \(\tilde{{S}}_{0} \). Setting \(\mathbf{{ p}}_{0} -\mathbf{{ {p}'}}_{0} =\Delta \mathbf{{ p}}_{0} \), we may rewrite (A4) as:
which is just the first of the results that we wanted to prove.
Note that this allows a direct interpretation of the different estimate \(\mathbf{{ {x}'}}_{0} \) as a transformation \({\mathbf{x}}_{0\vert \tilde{{S}}_{0} } \) of the first estimate \({\mathbf{x}}_{0\vert S_{0} } \) from an original reference system \(S_{0} \) to a new one \(\tilde{{S}}_{0} \), with transformation parameters \(\Delta \mathbf{{ p}}_{0} \) (for \(S_{0} \rightarrow \tilde{{S}}_{0} )\).
The corresponding transformation of velocity components \(\mathbf{{ v}}_{\tilde{{S}}_{0} } =(1+s_{0} )\mathbf{{ R}}({\varvec{ \uptheta }}_{0} ){\mathbf{v}}_{S_{0} } \approx (1+s_{0} )\left( {\mathbf{{ I}}-[\mathbf{{ \uptheta }}_{0} \times ]} \right) {\mathbf{v}}_{S_{0} } \) becomes in first-order approximation \({\mathbf{v}}_{\tilde{{S}}_{0} } \approx {\mathbf{v}}_{S_{0} } -[{\varvec{ \uptheta }}_{0} \times ]{\mathbf{v}}_{S_{0} } +s_{0} {\mathbf{v}}_{S_{0} } \) and since \({\varvec{ \uptheta }}_{0} \), \(s_{0} \) as well as \({\mathbf{v}}_{S_{0} } \) are very small, the last two second terms may be neglected so that finally
This allows us to compute an intermediate estimate
Subtracting (A7) from (A3), we find that the two estimates differ by
It is also possible to relate \(\mathbf{{ {x}'}}(t_{k} )\) and \({\mathbf{x}}_{\tilde{{S}}_{0} } (t_{k} )\) in a different way. From the first part of (A1a) and (A2), it follows that
while from the second part of (A1a), (A1b) and (A3), it follows that
Subtracting (A10) from (A9) gives
On the other hand, subtracting (A2) from the second part of (A7) and taking (A5) into account gives
Replacing \({\mathbf{x}}_{S_{0} } (t_{k} )={\mathbf{x}}_{\tilde{{S}}_{0} } (t_{k} )-\mathbf{{ E}}\Delta \mathbf{{ p}}_{0} \) from (A12) into (A11) gives
Since the left hand sides of (A8) and (A13) are equal, their right hand sides will also be equal giving
which must hold for all epochs \(t_{k} \). The only term depending on \(t_{k} \) in the right hand side of (A14) is \(\mathbf{{ p}}_{k} -\mathbf{{ {p}'}}_{k} \), which must thus be of the linear form \(\mathbf{{ p}}_{k} -\mathbf{{ {p}'}}_{k} =\mathbf{{ a}}+(t_{k} -t_{0} ){{ \dot{\mathbf{p}}}}\). This leads to
a relation which holds for all \(t_{k} \) only if \(\mathbf{{ a}}-\Delta \mathbf{{ p}}_{0} =\mathbf{{ 0}}\) and \(\mathbf{{ {v}'}}-{\mathbf{v}}_{S_{0} } =\mathbf{{ E}}{\dot{\mathbf{p}}}\). It follows that \({{ \mathbf{v}'}}={\mathbf{v}}_{S_{0} } +\mathbf{{ E}}\dot{\mathbf{p}}\), \(\mathbf{{ a}}=\Delta \mathbf{{ p}}_{0} \), and \(\mathbf{{ p}}_{k} -\mathbf{{ {p}'}}_{k} =\Delta \mathbf{{ p}}_{0} +(t_{k} -t_{0} ){{ \dot{\mathbf{p}}}}\). Therefore
and our proof has been completed.
Best fitting of initial transformation parameters between two sets of initial coordinates obtained using different minimal constraints in the stacking problem
The normal equations for the solution of the least squares problem (30) are
or explicitly
where
It is relatively easy to verify that the solution to the normal equations is given by
Best fitting of transformation parameter derivatives between two sets of velocities obtained using different minimal constraints in the stacking problem
By complete analogy, the solution to the least squares problem (31) is given by
where
Best fitting of initial transformation parameters and their derivatives between two sets of transformation parameter time series obtained using different minimal constraints in the stacking problem
The normal equations for the least squares solution of (32) are
where
with solution
Rights and permissions
About this article
Cite this article
Chatzinikos, M., Dermanis, A. A coordinate-invariant model for deforming geodetic networks: understanding rank deficiencies, non-estimability of parameters, and the effect of the choice of minimal constraints. J Geod 91, 375–396 (2017). https://doi.org/10.1007/s00190-016-0970-1
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00190-016-0970-1