Skip to main content

Part of the book series: International Association of Geodesy Symposia ((IAG SYMPOSIA,volume 140))

Abstract

In this study we propose the complementation of satellite-only gravity field models by additional a priori information to obtain a complete model. While the accepted gravity field models are restricted to a sub-domain of the frequency space, the complete models form a complete basis in the entire space, which can be represented in the frequency domain (spherical harmonics) as well as in the space domain (data grids). The additional information is obtained by the smoothness of the potential field. Using this a priori knowledge, a stochastic process on the sphere is established as a background model. The measurements of satellite-only models are assimilated to this background model by a subdivision into the commission, transition and omission sub-domain. Complete models can be used for a rigorous fusion of complementary data sets in a multi-mission approach and guarantee also, as stand-alone gravity-field models, full-rank variance/covariance matrices for all vector-valued, linearly independent functionals.

This is a preview of subscription content, log in via an institution to check access.

Access this chapter

Chapter
USD 29.95
Price excludes VAT (USA)
  • Available as PDF
  • Read on any device
  • Instant download
  • Own it forever
eBook
USD 129.00
Price excludes VAT (USA)
  • Available as EPUB and PDF
  • Read on any device
  • Instant download
  • Own it forever
Softcover Book
USD 169.99
Price excludes VAT (USA)
  • Compact, lightweight edition
  • Dispatched in 3 to 5 business days
  • Free shipping worldwide - see info
Hardcover Book
USD 169.99
Price excludes VAT (USA)
  • Durable hardcover edition
  • Dispatched in 3 to 5 business days
  • Free shipping worldwide - see info

Tax calculation will be finalised at checkout

Purchases are for personal use only

Institutional subscriptions

Notes

  1. 1.

    For practical reasons we work with \(\ell_{o_{\mathit{max}}} = 18, 000\). It is well known that the variance \(\mathrm{cov}(0,\sigma _{l}^{2})\) of the stochastic process using Kaula’s degree variances (20) is finite and given by

    $$\displaystyle\begin{array}{rcl} \mathrm{cov}(0,\sigma _{\ell}^{2})& =& 10^{-10}\sum _{ \ell=1}^{\infty }\frac{2\ell + 1} {\ell^{4}} \\ & =& 10^{-10}\left (2\sum _{\ell =1}^{\infty }\frac{1} {l^{3}} +\sum _{ \ell=1}^{\infty }\frac{1} {l^{4}}\right ) \\ & =& 10^{-10}\left (2\zeta (3) +\zeta (4)\right )\, {}\end{array}$$
    (21)

    where ζ(3) and ζ(4) denote the function values of Riemann’s zeta function. R. Apéry proved in 1977 that ζ(3) is irrational with a value of ζ(3) = 1. 20205690315959… (Hata 2000). Euler (1740) p. 133, §18 already derived \(\zeta (4) = \frac{\pi ^{4}} {90}\). These constants can be used to compute the relative approximation error for the finite summation up to 18,000 with 1 ⋅ 10−4 (2 ⋅ 10−4) starting the omission space at 181 (251).

References

  • Becker S, Freiwald G, Losch M, Schuh WD (2011) Rigorous fusion of gravity field, altimetry and stationary ocean models. J Geodyn (online first). doi:10.1016/j.jog.2011.07.006

    Google Scholar 

  • Euler L (1740) De summis serierum reciprocarum. Commentarii academiae scientiarum Petropolitanae 7:123–134. http://eulerarchive.maa.org/pages/E041.html

  • Hata M (2000) A new irrationality measure for ζ(3). Acta Arithmetica 92(1):47–57

    Google Scholar 

  • Jekeli C (1981) Alternative methods to smooth the Earth’s gravity field. Reports of the Department of Geodetic Science, Ohio State University (OSU), Ohio, No. 327

    Google Scholar 

  • Jekeli C (1996) Spherical harmonic analysis, aliasing, and filtering. J. Geod 70:214–223. doi:10.1007/BF00873702

    Article  Google Scholar 

  • Kaula W (1966) Theory of satellite geodesy. Blaisdell Publishing Company, Toronto

    Google Scholar 

  • Kusche J (2007) Approximate decorrelation and non-isotropic smoothing of time-variable GRACE-type gravity field models. J Geod 81:733–749. doi:10.1007/s00190-007-0143-3

    Article  Google Scholar 

  • Losch M, Sloyan B, Schöter J, Sneeuw, N (2002) Box inverse models, altimetry and the geoid: problems with the omission error. J Geophys Res 107(C7):3078. doi:10.1029/2001JC000855

    Article  Google Scholar 

  • Mayer-Gürr T, Kurtenbach E, Eicker A (2010) ITG-Grace2010 gravity field model. University Bonn, Institute of Geodesy and Geoinformation. http://www.igg.uni-bonn.de/apmg/index.phg?id=itg-grace2010

  • Meissl P (1971) A study of covariance functions related to the earth’s disturbing potential. Reports of the Department of Geodetic Science, Ohio State University (OSU), Ohio, No. 151

    Google Scholar 

  • Moritz H (1980) Advanced physical geodesy. Wichmann, Karlsruhe

    Google Scholar 

  • Pail R, Bruinsma S, Miggliaccio F, Förste C, Goiginger H, Schuh WD, Höck E, Reguzzoni M, Brockmann J, Abrikosov O, Veicherts M, Fecher T, Mayrhofer R, Krasbutter I, Sansó F, Tscherning C (2011) First GOCE gravity field models derived by three different approaches. J Geod 85:819–843. doi:10.1007/s00190-011-0467-x

    Article  Google Scholar 

  • Schuh WD, Becker S (2010) Potential field and smoothness conditions. In: Contadakis M, Kaltsikis C, Spatalas S, Tokmakidis K, Tziavos I (eds) The apple of knowledge - in honour of Prof. N. arabelos, pp 237–250. University of Thessaloniki, AUTH - Faculty of Rural and Surveying Engineering

    Google Scholar 

  • Swenson S, Wahr J (2006) Post-processing removal of correlated errors in GRACE data. Geophys Res Lett 33:L0840211. doi: 10.1029/2005GL025285

    Google Scholar 

  • Wahr J, Molenaar M, Bryan F (1998) Time variability of the Earth’s gravity field: hydrological and oceanic effects and their possible detection using GRACE. J Geophys Res 103:30205–30299. doi:10.1029/98JB02844

    Article  Google Scholar 

Download references

Acknowledgements

This work was funded within the DFG priority program SPP 1257 ‘Mass transport and mass distribution in the system Earth’. The authors acknowledge the European Space Agency for the provision of the GOCE data and the GOCO-group for providing the normal equations of the GOCE-TIM2.0 and the ITG-Grace2010s model.

Author information

Authors and Affiliations

Authors

Corresponding author

Correspondence to S. Müller .

Editor information

Editors and Affiliations

Appendix

Appendix

1.1 Expectation and Variance of a Stochastic Process in Amplitude/Phase Notation on the Sphere

The stochastic process \(\mathcal{U}(\vartheta,\lambda )\) on the sphere is defined by

$$\displaystyle{ \mathcal{U}(\vartheta,\lambda ) =\sum _{ \ell=0}^{\infty }\sum _{ m=0}^{\ell}A_{\ell m}\bar{P}_{\ell m}(\cos \vartheta )\cos \left (m\lambda + \mathcal{P}_{\ell m}\right ) }$$
(22)

where the phases constitute random variables. The distribution is defined by

$$\displaystyle\begin{array}{rcl} & & f_{\boldsymbol{\mathcal{P}}}(\mathbf{p}) =\prod \limits _{ \ell=0}^{\infty }\prod \limits _{ m=0}^{\ell}f_{ \mathcal{P}_{\ell m}}(p_{\ell m}) \\ & & \qquad \quad \mbox{ with}f_{\mathcal{P}_{\ell m}}(p_{\ell m}) = \left \{\begin{array}{crcccr} 0 & & &p_{\ell m}& \leq &-\pi \\ \frac{1} {2\pi } & \ \ -\pi &\leq &p_{\ell m}& \leq & \pi \\ 0 & \pi & \leq &p_{ \ell m}&. \end{array} \right.{}\end{array}$$
(23)

This means we have uniformly distributed phases, for each degree â„“ and order m, and phases for different degrees/orders are mutually independent.

We are interested in the expectation

$$\displaystyle\begin{array}{rcl} & & E\left \{\mathcal{U}(\vartheta,\lambda )\right \} =\int \limits _{ -\infty }^{\infty }u(\vartheta,\lambda )f_{\boldsymbol{ \mathcal{P}}}(\mathbf{p})d\mathbf{P} {}\\ & & =\!\int \limits _{ -\infty }^{\infty }\sum _{ \ell=0}^{\infty }\sum _{ m=0}^{\ell}\!\!A_{\ell m}\bar{P}_{\ell m}(\cos \vartheta )\cos \left (m\lambda + \mathcal{P}_{\ell m}\right )f_{\boldsymbol{\mathcal{P}}}(\mathbf{p})\;d\mathbf{P}\ . {}\\ \end{array}$$

Due to the independence of the phases the integral can be rewritten as

$$\displaystyle{ \int \limits _{-\infty }^{\infty }\sum _{ \ell=0}^{\infty }\sum _{ m=0}^{\ell}\!\!A_{\ell m}\bar{P}_{\ell m}(\cos \vartheta )\cos \left (m\lambda \,+\,\mathcal{P}_{\ell m}\right )f_{\mathcal{P}_{\ell m}}(p_{\ell m})\mathit{dp}_{\ell m}. }$$

If we now interchange integration and summation and introduce the individual distribution (23) we get

$$\displaystyle\begin{array}{rcl} & & E\left \{\mathcal{U}(\vartheta,\lambda )\right \} =\sum _{ \ell=0}^{\infty }\sum _{ m=0}^{\ell}A_{\ell m}\bar{P}_{\ell m}(\cos \vartheta ) {}\\ & & \qquad \qquad \qquad \qquad \times \ \frac{1} {2\pi }\int _{-\pi }^{\pi }\cos \left (m\lambda + p_{\ell m}\right )\mathit{dp}_{\ell m} {}\\ & & =\sum _{ \ell=0}^{\infty }\sum _{ m=0}^{\ell}A_{\ell m}\bar{P}_{\ell m}(\cos \vartheta )\;\frac{1} {2\pi }\big(-\sin m\lambda +\sin m\lambda \big)\ . {}\\ \end{array}$$

Finally we see that

$$\displaystyle{ E\left \{\mathcal{U}(\vartheta,\lambda )\right \} = 0\ . }$$
(24)

The variance of this process is given by

$$\displaystyle{ \mbox{ Cov}\left \{\mathcal{U}(\vartheta,\lambda ),\mathcal{U}(\vartheta ',\lambda ')\right \} =\int _{ -\infty }^{\infty }u(\vartheta,\lambda )\;u(\vartheta ',\lambda ')\;f_{\boldsymbol{ \mathcal{P}}}(\mathbf{p})d\mathbf{P}\ . }$$
(25)

Because of the independence of the random variables this can be written as

$$\displaystyle\begin{array}{rcl} & & \mbox{ Cov}\left \{\mathcal{U}(\vartheta,\lambda ),\mathcal{U}(\vartheta ',\lambda ')\right \} \\ & & \qquad =\sum _{ \ell=0}^{\infty }\sum _{ m=0}^{\ell}A_{\ell m}^{2}\bar{P}_{\ell m}(\cos \vartheta )\bar{P}_{\ell m}(\cos \vartheta ') \\ & & \qquad \times \ \int _{-\infty }^{\infty }\cos \left (m\lambda + p_{\ell m}\right )\cos \left (m\lambda ' + p_{\ell m}\right ) \\ & & \qquad \times f_{\mathcal{P}_{\ell m}}(p_{\ell m})\;dp_{\ell m}. {}\end{array}$$
(26)

If we extend the first cosine term in the integral to \(\left (m(\lambda -\lambda ') + m\lambda ' + p_{\ell m}\right )\), use the relation

$$\displaystyle{ cos(x + y)\cos (x - y) = \frac{1} {2}(\cos 2x +\cos 2y) }$$

and substitute

$$\displaystyle\begin{array}{rcl} x& =& \frac{1} {2}m(\lambda -\lambda ') + m\lambda ' + p_{\ell m} {}\\ y& =& \frac{1} {2}m(\lambda -\lambda ') {}\\ \end{array}$$

the integral can be solved and yields

$$\displaystyle\begin{array}{rcl} & & \frac{1} {2}\sin \left (m(\lambda -\lambda ') + 2m\lambda ' + 2p_{\ell m}\right )\Big\vert _{p_{\ell m}=-\pi }^{p_{\ell m}=\pi } + {}\\ & & \qquad \qquad \qquad \qquad \qquad \qquad + \frac{1} {2}\cos \left (m(\lambda -\lambda ')\right )p_{\ell m}\Big\vert _{p_{\ell m}=-\pi }^{p_{\ell m}=\pi }\ \ . {}\\ \end{array}$$

The first term vanishes because of the skew symmetry of the sine and only the cosine term is relevant. Substituting this result into (26) yields

$$\displaystyle\begin{array}{rcl} & & \quad \mbox{ Cov}\left \{\mathcal{U}(\vartheta,\lambda ),\mathcal{U}(\vartheta ',\lambda ')\right \} {}\\ & & =\sum _{ \ell=0}^{\infty }\sum _{ m=0}^{\ell}\frac{1} {2}A_{\ell m}^{2}\bar{P}_{\ell m}(\cos \vartheta )\bar{P}_{\ell m}(\cos \vartheta ')\cos \left (m\lambda - m\lambda '\right )\ \ . {}\\ \end{array}$$

Applying the addition theorem

$$\displaystyle{ \cos \left (m\lambda - m\lambda '\right ) =\cos m\lambda \cos m\lambda ' +\sin m\lambda \sin m\lambda ' }$$

and recalling the definition of Laplace’s surface spherical harmonics (9) the right hand side can be reformulated as

$$\displaystyle\begin{array}{rcl} & & \sum _{\ell=0}^{\infty }\sum _{ m=0}^{\ell}\frac{1} {2}A_{\ell m}^{2}\Big(\bar{C}_{\ell m}(\vartheta,\lambda )\bar{C}_{\ell m}(\vartheta ',\lambda ') {}\\ & & \quad +\bar{ S}_{\ell m}(\vartheta,\lambda )\bar{S}_{\ell m}(\vartheta ',\lambda ')\Big)\ \ . {}\\ \end{array}$$

Introducing now the amplitudes defined in (12) we get

$$\displaystyle\begin{array}{rcl} & & \sum _{\ell=0}^{\infty }\sum _{ m=0}^{\ell} \frac{1} {2\ell + 1}\sigma _{\ell}^{2}\Big(\bar{C}_{\ell m}(\vartheta,\lambda )\bar{C}_{\ell m}(\vartheta ',\lambda ') {}\\ & & \quad +\bar{ S}_{\ell m}(\vartheta,\lambda )\bar{S}_{\ell m}(\vartheta ',\lambda ')\Big)\ \ . {}\\ \end{array}$$

The decomposition formula or addition theorem for spherical harmonics (cf. e.g. Moritz 1980, p. 23 (3–30))

$$\displaystyle\begin{array}{rcl} & & \quad P_{\ell}\big((\vartheta,\lambda );\;(\vartheta ',\lambda ')\big) = {}\\ & & \frac{1} {2\ell + 1}\sum _{m=0}^{\ell}\left (\bar{C}_{\ell m}(\vartheta,\lambda )\bar{C}_{\ell m}(\vartheta ',\lambda ') +\bar{ S}_{\ell m}(\vartheta,\lambda )\bar{S}_{\ell m}(\vartheta ',\lambda ')\right ) {}\\ \end{array}$$

allows the for further simplification

$$\displaystyle{ \mbox{ Cov}\left \{\mathcal{U}(\vartheta,\lambda ),\mathcal{U}(\vartheta ',\lambda ')\right \} =\sum _{ \ell=0}^{\infty }\sigma _{ \ell}^{2}P_{\ell}\big((\vartheta,\lambda );\;(\vartheta ',\lambda ')\big) }$$

where the function value of the Legendre polynomial \(P_{\ell}\big((\vartheta,\lambda );\;(\vartheta ',\lambda ')\big)\) depends only on the spherical distance cosψ between \((\vartheta,\lambda )\) and \((\vartheta ',\lambda ')\)

$$\displaystyle{ \cos \psi =\cos \vartheta \cos \vartheta ' +\sin \vartheta \sin \vartheta '\cos (\lambda -\lambda ')\ . }$$
(27)

Finally this results in

$$\displaystyle\begin{array}{rcl} \mbox{ Cov}\left \{\mathcal{U}(\vartheta,\lambda ),\mathcal{U}(\vartheta ',\lambda ')\right \}& =& \sum _{\ell=0}^{\infty }\sigma _{ \ell}^{2}\;P_{\ell}(\cos \psi ) = \mathrm{cov}(\psi,\sigma _{\ell}^{2}). {}\\ \end{array}$$

Rights and permissions

Reprints and permissions

Copyright information

© 2015 Springer International Publishing Switzerland

About this paper

Cite this paper

Schuh, WD., Müller, S., Brockmann, J.M. (2015). Completion of Band-Limited Data Sets on the Sphere. In: Kutterer, H., Seitz, F., Alkhatib, H., Schmidt, M. (eds) The 1st International Workshop on the Quality of Geodetic Observation and Monitoring Systems (QuGOMS'11). International Association of Geodesy Symposia, vol 140. Springer, Cham. https://doi.org/10.1007/978-3-319-10828-5_25

Download citation

Publish with us

Policies and ethics