 |
| Titel |
A segmentation-Lasso approach for the homogenization of GPS series |
| VerfasserIn |
Emilie Lebarbier, Karine Bertin, Xavier Collilieux, Cristian Meza |
| Konferenz |
EGU General Assembly 2014
|
| Medientyp |
Artikel
|
| Sprache |
Englisch
|
| Digitales Dokument |
PDF |
| Erschienen |
In: GRA - Volume 16 (2014) |
| Datensatznummer |
250089246
|
| Publikation (Nr.) |
 EGU/EGU2014-3442.pdf |
|
|
|
|
|
| Zusammenfassung |
| The GPS series are generally affected by two different types of biases : abrupt changes du to instrument changes or re-location and functional biases (periodic signals) due to environmental effects (changes in atmospheric pressure and hydrology). The detection and correction of such biases, called homogenization, is a fundamental step in all dataset pre-processing in geosciences study.
The segmentation methods are now well known to detect abrupt-changes in the distribution of the observations along time. We propose here a segmentation model in which a functional part is included. To estimate these two parts of the model, we propose an iterative procedure combining a Dynamic Programming algorithm (which is the well-established efficient strategy to locate the abrupt-changes) and a Lasso-type procedure to estimate the functional part. An application to real data is done. |
|
|
|
|
|
|