Uncovering the hillslope scale flow and transport dynamics in an
experimental hydrologic system
Minseok Kim1, Till H. M.
Volkmann1,2, Aaron Bugaj1, Yadi
Wang3, Antônio A. Meira Neto4,
Katarena Matos4, Ciaran J.
Harman5,6, Peter A. Troch1,4
1Biosphere 2, University of Arizona, Tucson, AZ, USA,2Applied Intelligence, Accenture, Kronberg im Taunus,
Germany, 3Department of Environmental Science,
University of Arizona, Tucson, AZ, USA, 4Department of
Hydrology and Atmospheric Sciences, University of Arizona, Tucson, AZ,
USA, 5Department of Environmental Health and
Engineering, Johns Hopkins University, Baltimore, MD, USA,6Department of Earth and Planetary Sciences, Johns
Hopkins University, Baltimore, MD, USA
Hillslope scale water flow and transport dynamics have been extensively
studied (Burt & McDonnell, 2015; Hewlett & Hibbert, 1963), but
observing those internal dynamics in high spatial and temporal
resolutions remains challenging. In this study, we uncover internal
water flow and transport dynamics in an artificial hillslope in the
Landscape Evolution Observatory (LEO), Biosphere 2, University of
Arizona, Tucson, USA, using the experimental dataset collected in
December 2016. Complete information about the hillslope and experiment
can be found elsewhere (Pangle et al., 2015; Till H. M. Volkmann et al.,
2018); Here, we only summarize some relevant information.
The first part of the animation describes the experimental system and
setup (time 00:12 – 04:14 in Animation S1). The LEO hillslope is 330
m3 (30 m long, 11 m wide, and 1 m deep) sloping soil
lysimeter. The hillslope is primarily made up of loamy-sand
textured basaltic tephra, and the most downslope 5.5
m3 is filled with gravel-textured basaltic tephra. A
custom irrigation system supplies reverse osmosis filtered water onto
the LEO surface. The downslope boundary is exposed to atmospheric
pressure, creating the seepage face boundary condition. The sensor
networks (including pressure transducers and volumetric water content
sensors) and the water isotope sampling locations and intervals (7 hrs
to 101 hrs) are illustrated in Animation S1 (time 02:09 – 03:01). The
isotope composition of subsurface water is obtained from laser-based
online measurements of vapor that is extracted via custom gas probes
through equilibrium calculation (T. H.M. Volkmann & Weiler, 2014). The
irrigation sequence of this experiment was designed to generate a
periodic steady state, which allows the application of the PERidoic
Tracer Hierarchy method (Harman & Kim, 2014) for the observation of the
time-variable transit time distributions and the StorAge Selection
functions. Deuterium-labeled water was irrigated during the first two
irrigation events.
The second part of the animation shows the dynamics of the perched water
table and soil water content (time 04:15 – 06:53). The extent of the
saturated zone was estimated using the pressure transducer data and
Delaunay triangulation (Delaunay, 1934). The experimental data
show the saturation from below mechanisms—wetting up from the bedrock
surface into the soil profile (McDonnell, 1997)—and the saturation
from downslope to upslope. The water table profile forms a wedge-like
shape, which is a characteristic of hillslope with a high hillslope
(Peclet) number (Berne et al., 2005; Brutsaert, 1994). The hillslope
Peclet number of the LEO hillslope during the experiment is high
(> 10) (Kim et al., 2020). Significant time delays
in the water table dynamics are observed at some upslope locations
(e.g., at 13 m upslope), which is mostly due to the delayed water supply
from the convergent upslope area. The water content data indicates that
the convergent upslope water content began to decrease around the timing
of the water table peak at 13 m upslope.
The third part of the animation shows the tracer dynamics (from time
06:43). The animated experimental data reveal two notable water
transport dynamics. First, the vertical tracer movement is faster at the
upslope. This faster movement at the upslope is, in a sense,
counter-intuitive because the upslope region is drier than the
downslope. This is due to the lateral flow in the saturated zone and the
tension saturated zone, that are thicker at the downslope. While water
velocity is higher at the downslope, the direction of velocity is not
vertical but rotated towards the downslope in those zones.Second, the animated data illustrate that old water is present only at
the downslope. This observation is a characteristic of hillslope with a
high hillslope number, in which old water is preferentially discharged
(Kim et al., 2020). Indeed, the observed SAS function in this
hillslope is concave (Kim et al., 2020), indicating that the hillslope
preferentially discharges old water that is stored at the downslope.