Preprints
https://doi.org/10.5194/se-2019-60
https://doi.org/10.5194/se-2019-60
05 Apr 2019
 | 05 Apr 2019
Status: this preprint has been withdrawn by the authors.

Regularization methods for the combination of heterogeneous observations using spherical radial basis functions

Qing Liu, Michael Schmidt, Roland Pail, and Martin Willberg

Abstract. Various types of heterogeneous observations can be combined within a parameter estimation process using spherical radial basis functions (SRBF) for regional gravity field refinement. However, this process is in most cases ill-posed, and thus, regularization is indispensable. We discuss two frequently used methods for choosing the regularization parameter which are the L-curve method and variance component estimation (VCE). Based on these two methods, we propose two new approaches for the regularization parameter determination, which combine the L-curve method and VCE.

The first approach, denoted as ‘VCE + L-curve method’, starts with the calculation of the relative weights between the observation techniques by means of VCE. Based on these weights the L-curve method is applied to determine the regularization parameter. In the second approach, called ‘L-curve method + VCE’, the L-curve method determines first the regularization parameter and it is set to be fixed during the calculation of the relative weights between the observation techniques from VCE.

These methods are investigated based on two different estimation concepts for combining various observation techniques. All the methods are applied and compared in six study cases using four types of observations in Europe. The results show that the ‘VCE + L-curve method’ delivers the best results in all the six cases, no matter using SRBFs with smoothing or non-smoothing features. The ‘L-curve method + VCE’ also gives rather good results, generally outperforming the cases just using the L-curve method or VCE. Therefore, we conclude that the newly proposed methods are decent and stable for regularization parameter determination when different data sets are combined and can be recommended regardless of the type of SRBFs used.

This preprint has been withdrawn.

Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this preprint. The responsibility to include appropriate place names lies with the authors.
Qing Liu, Michael Schmidt, Roland Pail, and Martin Willberg

Interactive discussion

Status: closed
Status: closed
AC: Author comment | RC: Referee comment | SC: Short comment | EC: Editor comment
Printer-friendly Version - Printer-friendly version Supplement - Supplement

Interactive discussion

Status: closed
Status: closed
AC: Author comment | RC: Referee comment | SC: Short comment | EC: Editor comment
Printer-friendly Version - Printer-friendly version Supplement - Supplement
Qing Liu, Michael Schmidt, Roland Pail, and Martin Willberg
Qing Liu, Michael Schmidt, Roland Pail, and Martin Willberg

Viewed

Total article views: 1,153 (including HTML, PDF, and XML)
HTML PDF XML Total BibTeX EndNote
733 336 84 1,153 88 91
  • HTML: 733
  • PDF: 336
  • XML: 84
  • Total: 1,153
  • BibTeX: 88
  • EndNote: 91
Views and downloads (calculated since 05 Apr 2019)
Cumulative views and downloads (calculated since 05 Apr 2019)

Viewed (geographical distribution)

Total article views: 899 (including HTML, PDF, and XML) Thereof 898 with geography defined and 1 with unknown origin.
Country # Views %
  • 1
1
 
 
 
 

Discussed

Latest update: 20 Nov 2024
Download

This preprint has been withdrawn.

Short summary
Regularization is indispensable in regional gravity field modelling. In this paper, we propose two new approaches for the regularization parameter determination, which combine the L-curve method and variance component estimation (VCE). The performance of each method is studied for combining heterogeneous observations using spherical radial basis functions. The results show that our newly proposed methods are decent and stable for regularization parameter determination.