Fourth Study Conference on BALTEX Scala Cinema Gudhjem

- 98 -
Introducing Lateral Subsurface Flow in Permafrost Conditions in a
Distributed Land Surface Scheme
Petra Koudelova, Toshio Koike
Dept. of Civil Engineering, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo, 113-8656, Japan
e-mail: petra@hydra.t.u-tokyo.ac.jp
1. Introduction
Specific features are associated with perennially frozen soil
(permafrost). When soil freezes, its hydraulic conductivity
dramatically decreases, making frozen soil an impervious
table for water flow, and thus most hydrologic processes are
confined to the active layer above the frozen table. The
water phase transitions involve consumption or release of a
considerable amount of fusion heat and they affect thermal
characteristics of soil due to the different thermal
characteristics of water and ice. Because of its effects on
thermal and hydrological regimes, frozen soil is an
important factor in land surface processes. Therefore, it is
essential to incorporate soil freezing/thawing process in land
surface schemes (LSSs). Li and Koike (2003) give a short
review of recently introduced frozen soil parameterizations
(FSPs) incorporated in LSSs. As a part of 1-D LSSs, these
parameterizations treat vertical hydrological processes
associated with frozen soil.
However, frozen soil also influences lateral transport of
water leading to areal redistribution of surface wetness and
surface fluxes. In hilly regions, lateral flow is a dominant
process affecting spatial distribution of surface wetness and
thus, it is important to introduce lateral flow effects in LSSs.
Koike and Koudelova (2003) showed the impact of surface
lateral flow on water budget and surface fluxes over a
mountain catchment in the Tibetan Plateau.
Beside the overland flow, subsurface lateral flow can
significantly contribute to the spatial soil moisture
distribution when hilly terrain is combined with permafrost.
Since frozen soil serves as an impervious bed, liquid water
accumulates above it. Consequently, overlying soil becomes
saturated, which results in the saturated flow along a slope.
Accordingly, the top of the slope is dryer than the bottom
part, which leads to different thermal properties of soil along
the slope. Due to the lower liquid soil moisture content, the
upper part of the slope thaws faster than the bottom.
In this study, we introduce a quasi-3D land surface scheme,
which accounts for both vertical and horizontal hydrologic
processes associated with permafrost conditions. The
attribute “quasi-3D” expresses an explicit linkage between
the vertical and horizontal parts of the model. The model is
developed by implementing the land surface scheme SiB2
(Sellers et al., 1996) incorporating a FSP (Li and Koike,
2003) in a Distributed Hydrologic Model (DHM). We carry
out a numerical experiment, in which the model is applied to
a single slope using CEOP Tibet forcing data.
2. Model description
In the coupled model, the SiB2 is embedded into the
framework of the DHM and solves all the vertical processes
for each grid cell individually using the meteorological
forcing data. The SiB2 generates the saturated zone above
the frozen table and surface runoff, which are then treated by
the saturated zone and the surface flow components of the
DHM. The updated values of the saturated zone and the
surface water storage are used as an initial state in the next
time step in the SiB2. Most of the parts of the original SiB2
are kept unchanged but several modifications were
necessary to involve the FSP and water input due to the
lateral flow. Implementation of the SiB2 in the framework
of the DHM and coupling with the surface flow and the
river flow components of the DHM is introduced in Koike
and Koudelova (2003). Here, we focus on generation of
the saturated zone above the frozen table and the
consequent saturated flow.
The FSP, which is described in details in Li and Koike
(2003), predicts frozen/thawed depth and phase changes of
soil water content over time. The three-layer soil model in
SiB2 is maintained in the FSP, but the governing
equations of water balance and surface heat balance are
modified to involve the soil freezing/thawing process. The
resolution of the three SiB2 soil layers is, however, too
coarse for the prediction of saturated zone above the
frozen table. Therefore, we introduce a multi-layer soil
model in this study. The calculation of liquid water and ice
contents is kept unchanged. The solution of vertical
unsaturated flow is expanded for the multi-layer structure.
The unsaturated flow proceeds only within the active layer.
The depth of the saturated zone is determined after the
unsaturated flow calculation. Firstly, the layers above the
frozen table are checked for the saturation, starting from
the bottom one. If there is a continuous saturated zone
comprising at least one layer just above the frozen table, it
represents the saturated zone. If the bottom layer is not
saturated, the concept of Ishidaira et al. (1998) is adopted.
It is assumed that the bottom layer is wetter than the upper
ones. If this assumption is valid, the thickness of saturated
zone is determined from the equation:
( D − D′
) l , −1
Dθ = D′
θ + θ , (1)
l , b
s , b
b
where D is the thickness of the bottom layer, D’ is the
thickness of the saturated zone, θl,b is the average
volumetric liquid content in the bottom layer b, θl,b-1 is the
average volumetric liquid content in the layer b-1, and θs,b is the saturated water content in the bottom layer. If the
condition θl,b > θl,b-1 is not fulfilled or if the active layer
consists of only the uppermost layer, the value of θl,b-1 in
the Eq. (1) is replaced with 0.9θl,b. 2-D subsurface saturated flow scheme is based on the
groundwater component of the DHM, which employs a
non-steady Boussinesq equation. In the case of frozen soil,
the impervious bed moves up or down over time according
to the freezing/thawing process. Moreover, the thickness
of saturated zone may vary greatly along a slope and over
time. Accordingly, the depth of the impervious bed and
the saturated zone are calculated every time step in SiB2.
Interactions between the saturated zone and overlying soil
are neglected in 2-D subsurface flow routing. The
moisture available for routing is determined by subtracting
the residual liquid content from the saturated value. The
obtained value represents the aquifer storage coefficient
and is calculated in each time step. The growth/drop of
water head due to the saturated flow is converted into the
increase/decrease of liquid soil moisture content in the
affected layers. The updated values of soil moisture are
used as initial conditions for the next time step in SiB2.