Résumé
;Estimating evapotranspiration (ET) is a primary challenge in modern hydrology. Hydrogravimetry is an integrative approach that provides highly precise continuous measurement of gravity acceleration. However, large-scale effects (e.g. tides, polar motion, atmospheric loading) limit the fine time-scale interpretation of this data and processing leads to residual signal noise. To circumvent this limitation, we exploited the difference between two superconducting gravimeters located 512 m apart on the same vertical. The difference calculation makes it possible to remove shared large-scale effects. Daily variation of this gravity difference is significantly correlated with daily evapotranspiration as estimated using the water balance model SimpKcET (p-value = 4.10<sup>-10</sup>). However, this approach is effective only during rain-free periods. In the future, comparison with direct ET measurements (e.g. eddy-covariance, scintillometer) may confirm and strengthen our interpretation. Improved hydrogravimetric data processing will allow to extend this approach to other experimental sites equipped with a single superconducting gravimeter.