The surveyors of the Large Scale Metrology section of the European Organization for Nuclear Research (CERN) use hydrostatic levelling systems (HLS) to perform precise vertical alignment measurements. The HLS achieves micrometer accuracy, which allows it to be used for the fundamental physics experiments such as the Large Hadron Collider (LHC). An HLS measures the deformations that lead to the misalignment of any particle accelerator linked to the ground, but it also measures other phenomena with very particular characteristics. Among these measured phenomena Earth tides form the main part of the signal. Their effect on HLS measurements is periodic and produces a long baseline tilt that does not lead to a relative misalignment of the magnets that constitute the accelerator.The objectives of this doctoral research are to be able to predict the effects which do not disturb the relative alignment of a particle accelerator and to remove these signals from the HLS measurements. Indeed, the horizontal and vertical positioning tolerances to be respected in the realm of particle accelerators are becoming tighter and tighter. For example, the Compact Linear Collider (CLIC), currently the object of a feasibility study, requires a 3σ alignment accuracy of 10 μm in a sliding window of 200 m, in both the transverse and vertical directions. The HLS is a candidate for measuring the vertical alignment but the amplitude of the tidal effect is approximately +/-20 μm over200 m, making it necessary to take into account this long baseline phenomenon for the instrument to meet the CLIC requirements.This doctoral thesis is inspired by previous work on the long baseline tiltmeters and describes the effects measured by HLS in order to classify the measured phenomena according to whether they could produce a misalignment of a particle accelerator or not. Finally, the tools and models to predict those effects that are well understood are used to anticipate the different signals measured by HLS installed at CERN.