 |
| Titel |
A multi-temporal image correlation method to characterize landslide displacement fields |
| VerfasserIn |
J. Travelletti, C. Delacourt, G. Koval, J.-P. Malet, J. Schmittbuhl, D. B. van Dam |
| Konferenz |
EGU General Assembly 2009
|
| Medientyp |
Artikel
|
| Sprache |
Englisch
|
| Digitales Dokument |
PDF |
| Erschienen |
In: GRA - Volume 11 (2009) |
| Datensatznummer |
250026136
|
|
|
|
|
|
| Zusammenfassung |
| A monitoring system using a high resolution optical camera to characterize the displacement
field of the Super Sauze mudslide through time. The mudslide exhibits a strong
activity, with velocity values ranging from 0.01Â m.day-1 to 0.4 m.day-1 in periods of
acceleration.
The monitoring system is operational since 2007, and consists of a high resolution optical
camera installed on a fixed concrete pillar located on a stable crest in front of the mudslide.
The camera is controlled by a datalogger, and registers 3008 x 2000 pixels photographs at
11:00 AM, 12:00 PM, 13:00 PM and 14:00 PM each 4 days.
The objective of this work is to discuss the possibility of deriving the displacement field
from the multi-temporal images by using two algorithms of image correlation (MicMac;
CORRELI2D), and to present the first analysis of the landslide dynamics.
The quality of the image correlation is controlled by (i) changes of illumination angles
and intensities through time leading to specific shadow patterns, (ii) surface texture changes
and (iii) possible slight movements of the camera. These possible errors can be corrected by
the image correlation procedure.
The correlation results are first interpreted in terms of pixel displacements, pixel velocity
and direction of movements in the camera image plane. The computed pattern of
displacement is in good accordance in terms of direction and amplitude with the benchmarks
displacement observed in the field. The velocity pattern of the mudslide is very well
differentiated from the stable parts where the displacement amplitudes tend to be zero and the
directions are randomly distributed. The main challenge is to convert the pixel displacements
into metric displacements for which the critical point is to characterize accurately the image
geometry. |
|
|
|
|
|
|