Mass distribution on Earth is continuously changing due to various physical processes beneath the Earth's surface or on the surface. Some of the primary sources for these mass displacements are tidal forces, atmospheric and oceanic loading, and seasonal changes in continental water distribution. The development of relative cryogenic gravimeters, the Superconducting Gravimeters (SGs), has made it possible to characterize and monitor such mass variations at orders of magnitudes as small as a few nm/s 2 (1 nm/s² ~ 10 -10 g where g is the mean gravity at the Earth's surface). Our study focuses on the hydrodynamics of the 900 m thick unsaturated zone of the low-noise underground research laboratory (Laboratoire Souterrain à Bas Bruit, LSBB) located in Rustrel (France) using a unique configuration of two SGs vertically arranged 520 m depth apart. The installation of an SG (iGrav31) at the site surface several years after installing the first (iOSG24) inside a tunnel has provided several new insights into the understanding of the hydrological processes occurring in the LSBB. By comparing differential and residual gravity time-series together with global hydrological loading models, we find that most water-storage changes occur in the unsaturated zone between both SGs. The misfit between the observed gravity time-series and the gravity effect corresponding to local hydrological contribution calculated from global hydrological models can be explained by large lateral fluxes and rapid runoff occurring in the LSBB site. Finally, we implement a rectangular prism method to compute forward gravity responses to water storage changes for a homogeneous water-layer following the site topography using a 5-m digital elevation model.In particular, we analyse the sensitivity of the differential record from both SGs to the extent and depth of the water storage changes by computing the corresponding 2D admittances. This gravity difference is sensitive to an extension up to about 2500 m laterally before tending towards an asymptotic value corresponding to the Bouguer plate approximation. We show that the zone of water-storage changes that best fits observed differential gravity signal is located at depths larger than 500 m (below iOSG24). This fitting is improving when the integration radius increases with depth. This is the first time that hydrological processes are investigated when the baseline configuration of two SGs is vertical.