We study the time series of vertical ground displacements from continuous global navigation
satellite system (GNSS) stations located in the European Alps. Our goal is to improve
the accuracy and precision of vertical ground velocities and spatial gradients across
an actively deforming orogen, investigating the spatial and temporal features of the
displacements caused by non-tectonic geophysical processes. We apply a multivariate
statistics-based blind source separation algorithm to both GNSS displacement time
series and ground displacements modeled from atmospheric and hydrological loading,
as obtained from global reanalysis models. This allows us to show that the retrieved
geodetic vertical deformation signals are influenced by environment-related processes
and to identify their spatial patterns. Atmospheric loading is the most important
process, reaching amplitudes larger than 2 cm, but hydrological loading is also important,
with amplitudes of about 1 cm, causing the peculiar spatial features of GNSS ground
displacements: while the displacements caused by atmospheric and hydrological loading
are apparently spatially uniform, our statistical analysis shows the presence of N-S
and E-W displacement gradients.We filter out signals associated with non-tectonic
deformation from the GNSS time series to study their impact on both the estimated
noise and linear rates in the vertical direction. Taking into account the long time
span of the time series considered in this work, while the impact of filtering on
rates appears rather limited, the uncertainties estimated from filtered time series
assuming a power law plus white noise model are significantly reduced, with an important
increase in white noise contributions to the total noise budget. Finally, we present
the filtered velocity field and show how vertical ground velocity spatial gradients
are positively correlated with topographic features of the Alps.
