Topographic controls on seepage distribution in 3D mountain systems

Seepage areas, i.e., areas where the water table intersects the land surface, are strong indicators of groundwater-surface water interaction and have a critical role on ecosystems and on water quality. Numerous studies have been carried out aiming at characterizing seepage areas and the factors controlling their occurrence. Still, most of literature focused on theoretical or synthetic systems. Then seepage areas in complex environments, such as mountain systems, are to be further studied. In this context, we propose to test the pertinence of well-known and widely used frameworks, either analytical or numerical, against 3D complex systems aiming at proposing corrections to better represent the complexity inherent to mountain systems.

The methodology follows the development of 3D homogeneous and uniformly recharge numerical models at steady state using MODFLOW. The system complexity specifically lies in its 90m-resolution topography based on a real mountain catchment, the Quebrada de Tarapacá (North Chile). Regional scale catchment area (~900km²) allows incorporating various sub-catchments, hence studying a panel of geomorphological settings differing in slope variation and characteristic length. We perform a sensitivity study of the seepage area to recharge rate which is varied over 6 orders of magnitude as a proxy for variable climatic conditions. Results are analyzed in relation to the ratio between hydraulic conductivity and recharge (K/R).

Consistently with previous studies, the K/R ratio highly influences seepage distribution showing the contraction of the river network and groundwater flow redistribution. At high recharge rates (low K/R), seepage area tends to 100% of the catchment area, a fully saturated catchment. On the other hand, at low recharge rates (high K/R), seepage tends to be null, without totally disconnecting the water table from the surface. At intermediate recharge rates (10^-1 < K/R < 10), the seepage area linearly decreases while K/R increases. Numerical results differ from estimation of theoretical solutions or 1D numerical models as those tend to overestimate seepage area except when fully saturated. Even though K/R exerts the principal control on seepage distribution, the geomorphological characteristics illustrated by the characteristic length influences seepage organization. Defining the characteristic length can be challenging in mountain context due to the high variability of geomorphologic features. Various definitions of the characteristic length were then tested: (i) from drainage density; (ii) from the relation between slope and drainage area and (iii) from the equivalent hillslope method. The first two methods tend to underestimate the characteristic length, and hence, catchments appear to be only recharge-controlled following Haitjema and Mitchell-Bruker criterion (2005), counter-intuitively even at high recharge rates. Contrarily, the equivalent hillslope method shows promising results, as the balance between topographic and recharge control catchment is respected, following the same criteria.

Therefore, we show a significant overestimation of seepage area from theoretical models in comparison to 3D fully distributed models due to their inability to incorporate details of the topography such as very high hilltops. Consequently, 3D model development in mountain system is crucial as other analytical or 1D numerical models cannot illustrate the geomorphological details influencing seepage areas distribution.