GPS calculations
Monday 26 October 2009 by Thomas Lebourg

traduction [English ] [français ]

GPS data processing was performed with Gamit software (release 10.32) developped by MIT (Massachussetts Institute of Technology). One of the future purposes is to succeed in automatic processing. CLAP, which is supposed permanent and very close to the landslide, is chosen as the reference station and is highly constrained. On the contrary, CLP1 and CPL2 are let free. As the two main stations (CLP1 and CPL2) are two mono frequency GPS, the data processing can only be performed with the L1 frequency. Therefore a mono frequency processing is carried out in phase on a network. Moreover, regarding the little size of the latter, the ionospheric corrections are considered as constant on the site (CLAP is very close to CLP1 and CPL2). Gamit is a scientific software including numerous options. Aiming at quasi real time automatic processing, different calculation modes have been tested and, considering the results, calculation parameters assessed as optimum have been set up. After the referencing of daily calculations thanks to Globk module, we obtain for the stations CLP1, CPL2 and RABU the position time series below (respectively from the top to the bottom, the North, East and Vertical components).

 Times series obtained with Gamit: from left to right, CLP1, CPL2 and RABU. From the top to the bottom, the North, East and Vertical components

The results from Gamit are presented above. The CLP1 and CPL2 displacements and their decrease between 2003.5 and 2007.5 are obvious. Concerning RABU, time series indicate a slight long-term displacement with regard to CLAP, but above all an important annual sinusoidal component on the plane components. The main results are mentionned below:

GPS mean velocities in cm/year between 2003.52466 and 2007.26712
East North Vertical length/CLAP Total(**)

CLP1 -15,7 -17,8 -12,9 -20,7 -27,0
CPL2 -13,4 -23,3 - 17,9 -27,8 -32,3
RABU (*) -0,18 -0,15 -0,20 -0,19 -0,3
* after the removal of the annual sinusoidal trend
** total velocity = 3D velocity

In order to analyze the CLP1 and CPL2 motions with more accuracy, we take an interest in their velocities (graph below). The decrease is well visible, concerning either CLP1 or CPL2. Nevertheless many velocity shift events exist for every station. Most of them are common to the two stations. Is this an artefact generated by the analysis or is this related to a real physical phenomenon? The comparison of the results with the prism ones will enable to clarify partly this question.

Velocity variations (meters/year) estimated at 30 days, of the distance CLAP/CLP1 (in blue color) and CLAP/CPL2 (in red color). The velocity is negative because the length decreases.

Some quality statistical indicators
Number of days calculated: between 88 and 93 % according to the stations (nearly 95% with the removal of the 2003.5-2004.0 period). A table summarizes the repeatibility values calculated: (thus this estimator conveys the short-term diversion of the results)
rms (mm) East North Vertical length/CLAP
CLP1 3,7 3,4 14,4 3,3
CPL2 3,6 3,8 15,6 4,2
RABU 5,6 4,9 15,7 7,4