Time-lapse, Distributed Microgravity Observations as a Tool to Inform Hydrological Models

Friday, 19 December 2014
Sebastiano Piccolroaz1, Bruno Majone1, Francesco Palmieri2, Giorgio Cassiani3 and Alberto Bellin1, (1)University of Trento, Trento, Italy, (2)National Institute of Oceanography and Experimental Geophysics, Trieste, Italy, (3)University of Padua, Padua, Italy
Thanks to significant advances in technology and modeling capabilities, the use of geophysical monitoring techniques in support of hydrological modeling showed a rapid growth over the last decades. However, at the catchment scale, the payoff in terms of model reliability has not always justified the increased experimental efforts needed in order to collect useful additional information. The objective of this work is to evaluate the advantage gained by coupling traditional hydrological data with unconventional geophysical information in inverse modeling of a complex hydrological system. In particular, we explored how the use of time-lapsed, spatially distributed micro-gravity measurements may improve the conceptual modeling of a complex catchment, allowing for an unambiguous identification of the more appropriate model and a general reduction of uncertainty.

The study area is the Vermigliana catchment, a small alpine watershed located in the South-Eastern Alps, Italy. The morphology is relatively complex, being characterized by several minor valleys and ridges with elevation ranging from 950 m a.s.l. to 3558 m a.s.l. The catchment has been monitored by 13 micro-gravity stations (part of a network covering a larger area), where extensive micro-gravity measurements have been performed during 6 field campaign between 2009 and 2011. Sub-daily streamflow data are available at the Vermiglio stream gauging station (with a contributing area of 79 km2) for the same period. Once corrected to remove the effect of ocean tides and instrumental drift, micro-gravity measurements have been used as proxy of changes in the total water storage, including soil moisture, groundwater and snowpack. These data have been combined with streamflow measurements in a multi-objective optimization approach, coupling a semi-distributed hydrological model (GEOTRANSF) with a gravity model. The inclusion of microgravity data resulted in a better constraint of the inversion procedure and an improved capability to identify limitations of concurring conceptual models to a level that would be impossible relying only on streamflow data. This permitted a more reliable description of hydrological processes, with a significant (up to 90%) drop in the uncertainty associated to the evaluation of water storage variations within the system.