Agricultural and Forest Meteorology 9899 (1999) 305324

Distribution of soil moisture and groundwater levels

at patch and catchment scales
S. Beldringa,*, L. Gottschalkb, J. Seibertc, L.M. Tallaksenb
a

Department of Hydrology, Norwegian Water Resources and Energy Administration,

P.O. Box 5091 Majorstua, N-0301 Oslo, Norway
b
Department of Geophysics, University of Oslo, P.O. Box 1022 Blindern, N-0315 Oslo, Norway
c
Department of Earth Sciences, Uppsala University, Norbyvagen 18B, S-752 36 Uppsala, Sweden

Abstract
This study is a contribution to the northern hemisphere climate processes land surface experiment (NOPEX). Its purpose is
to investigate the spatial variability of groundwater levels and soil moisture content at different scales in a landscape
dominated by boreal forest and till soils, which is characteristic of the Nordic countries. The analysis of data from the NOPEX
area are based on a review of previous studies on the spatial distribution of these state variables and their signicance for
runoff formation. Soil moisture content in the unsaturated zone and depth to the groundwater table show characteristic patterns
which are related to the landscape elements (patches) of the drainage basins. Similar behaviour is observed in different parts of
the NOPEX region. The variability of average values between areas decreases to a minimum for catchments with size larger
than 1 km2. It can therefore be concluded that the main part of the spatial variability of soil moisture content and depth to the
groundwater level in the till soils of the NOPEX area is found within small drainage basins. Based on a physical description of
the soil, distribution functions of soil moisture content conditioned on the depth to the groundwater table have been developed,
both for the patch scale and the catchment scale. # 1999 Elsevier Science B.V. All rights reserved.
Keywords: Till; Groundwater; Unsaturated zone; Spatial distribution

1. Introduction
This work is a contribution to the northern hemisphere climate processes land surface experiment
(NOPEX). The purpose of NOPEX is to study
landsurfaceatmosphere interactions at local and
regional scales in a Northern European landscape
dominated by boreal forest and crystalline rocks
and affected by a number of glaciations. Since the
*
Corresponding author. Fax: 47-22-95-92-01.
E-mail address: stein.beldring@nve.no (S. Beldring).

dynamics and water balance of land surface hydrology

have profound inuence on these interactions, one of
the objectives of NOPEX is to investigate the relationship between hydrological state variables, in particular
soil moisture and groundwater storage, and landscape
characteristics at different spatial scales (Halldin et al.,
1999).
The hydrological response of a drainage basin is
determined by small- and mesoscale variations in its
geology, topography, vegetation and other landscape
characteristics. The independent input variable precipitation is transformed in the hydrological system

0168-1923/99/$ see front matter # 1999 Elsevier Science B.V. All rights reserved.
PII: S 0 1 6 8 - 1 9 2 3 ( 9 9 ) 0 0 1 0 3 - 3

306

S. Beldring et al. / Agricultural and Forest Meteorology 9899 (1999) 305324

into the dependent output variables evapotranspiration

and runoff and changes in the soil moisture and ground
water storage of the system. In order to develop
physically based models of hydrological processes
the large variability in system properties and their
inuence on the water balance components must be
considered. This requires a subdivision of the landscape into hydrological units or patches with characteristic time and space scales with regard to runoff
formation, i.e. different water pathways and travel
times (O'Loughlin, 1981). A patch is any area of
the landscape that has broadly similar hydrological
responses in terms of the quantities of interest, in this
instance, runoff production and evapotranspiration
(Beven, 1995). The spatial variation of these patches
or source areas of runoff and their impact on moisture
conditions and runoff in catchments at different scales
is a major problem in hydrological research.
The main surface deposits in the Nordic countries
are till soils. Till is a non-sorted sediment deposited by
glacier ice and composed by a variety of clasts of a
wide range of sizes (Haldorsen and Kruger, 1990). A
characteristic feature of the boreal landscape dominated by till soils is a thin soil cover with a saturated
hydraulic conductivity which decreases with depth
(Lind and Lundin, 1990). As a consequence, the
groundwater table largely follows topography, with
a depth of a few metres below ground surface in
elevated areas. In the lower parts of certain hillslopes
and in some low areas, the groundwater table is at or
above the ground surface (Lundin, 1982; Rodhe,
1987). The depth to the groundwater table controls
the process of runoff formation in these areas, both
during stormow and recession. When the depth is
small, the large hydraulic conductivity of the surface
near soil layers compared to the deeper layers leads to
larger streamow contribution from hillslopes (Grip
and Rodhe, 1994). Several studies on runoff generation dynamics in humid temperate environments have
emphasized the domination of groundwater in runoff
generated by storms or snowmelt (Sklash and Farvolden, 1979; Rodhe, 1981, 1987; Espeby, 1990; Sklash,
1990; Bonell, 1993). (Nyberg, 1995, 1996) found that
the runoff from a small headwater catchment increased drastically when the depth to the groundwater
table diminished.
Soil moisture conditions are closely related to
groundwater conditions. Between precipitation

events, the soil moisture content will have a maximum

value at a state of hydrostatic equilibrium which
is determined by the physical characteristics of
the soil and the depth to the groundwater table.
The soil moisture decit which develops due to evapotranspiration will have to be lled up before the
groundwater table rises. In those parts of a catchment
which are located close to the stream, parts of the
storm runoff may be generated as saturation excess
overland ow due to a combined effect of small soil
moisture decit prior to the storm and a rise in the
groundwater table during the storm (Anderson and
Burt, 1990).
In summary, the various delivery mechanisms of
catchment runoff in a landscape dominated by till soils
depend on soil moisture conditions and the depth to
the groundwater table. They are the main hydrological
variables that can be used to distinguish between
different hydrological response units, i.e. patches in
the landscape mosaic which produce `similar' hydrological response on snowmelt or precipitation. Flugel
(1995) denes hydrological response units as distributed, heterogeneously structured entities having a
common climate, land use and underlying pedological, topological and geological associations controlling their hydrological dynamics. The variation of
hydrological process dynamics within one type of
hydrological response unit is small compared to the
dynamics in a different type of unit. Topography has a
signicant effect on subsurface ow and moisture
conditions in a catchment and it can therefore be used
to distinguish between hydrological response units
(Beven and Kirkby, 1979; Beven and Wood, 1983;
O'Loughlin, 1986; Bonell, 1993; Thompson and
Moore, 1996). In a study which showed that spatial
response units in a landscape dominated by till soils
can be identied with the help of geomorphology,
Krasovskaia (1985) used a classication of catchment
topography based on Hack and Goodlett (1960) to
differentiate between three types of units in a catchment: (1) nose, the driest part, including the ridge crest
and the nearby slopes where the contours are convex
outward; (2) hollow, the central part of the basin along
the stream with favourable moisture conditions, an
area in which the contours are concave outward; (3)
slope, the zone between nose and hollow with transitional moisture conditions where the contours are
straight or nearly so.

S. Beldring et al. / Agricultural and Forest Meteorology 9899 (1999) 305324

The scaling problem (aggregation/disaggregation

problem) is one of the key issues for the NOPEX
project. A critical question is whether it is possible to
generalize information about hydrological behaviour
from small catchments to large catchments (regions)
or not. Large scale hydrological models with calibrated effective parameters are frequently used,
although non-linearities and structural heterogeneity
in natural catchments make it unlikely that the equations of hydrological theories developed at small space
and time scales can be applied to large scale problems
(Beven, 1995). An alternative approach is the use of
spatial distribution functions which assume that the
frequency of occurrence of state variables or parameters are more important than the actual pattern of
the parts within the catchment. A critical problem in
the application of these models is to determine at what
scale the description of actual patterns of heterogeneity in catchment properties can be replaced by their
distribution functions. An attempt to dene this scale
is the representative elementary area (REA) concept
introduced by (Wood et al., 1988, 1990). It implies that
(1) for a range of spatial scales the pattern of landscape
elements is no longer signicant for the formation of
runoff, only their relative proportion as given by a
spatial distribution function matters; (2) the similarity
in catchment behaviour within a region due to similar
climate, soil, topography and other factors allows the
same parameterization of distribution functions to be
used for different catchments within this range of
spatial scales.
Several studies of the REA concept have focused on
runoff, since it represents the integrated effect of
hydrological processes within a catchment. (Wood
et al., 1988, 1990) studied the effects of spatially
variable topography, precipitation and soil characteristics using synthetic data and a hydrological model.
They found that the variability in runoff between
catchments of similar size decreased with increasing
area and stabilized at a minimum at above 1 km2.
Their conclusion was that the runoff generation population appears to be stationary and that the size of the
REA is determined by topography, while the spatial
variability of rainfall and soil characteristics only have
a secondary role. In a similar study, Bloschl et al.
(1995) concluded that the size of the REA is governed
by storm duration and correlation length of precipitation. The main difference between the studies is that

307

(Wood et al., 1988, 1990) compared catchments of

similar size irrespective of their relative position,
whereas Bloschl et al. (1995) considered the runoff
from nested catchments of variable size around a xed
point. Woods et al. (1995) used distributed measurements of runoff in river basins to study the effect of
scale. They found that for areas smaller than 1 km2,
the variance of specic discharge between catchments
of similar size decreased faster with increasing area
than would be expected for a random sample. At larger
scales the variance decreased in a manner consistent
with sampling from a stationary random eld. These
results were taken as an evidence of organization in
catchment behaviour, supporting the idea of a REA. In
an investigation of the scaling behaviour of soil
moisture, Wood (1995) concluded that the size of
the REA may be as large as 510 km2.
The purpose of this work is to investigate the spatial
variation of soil moisture in the unsaturated zone and
the depth to the groundwater table at different scales in
the boreal forest landscape in the NOPEX region. In
order to achieve a realistic description of land-surface
hydrological processes, it is necessary to take into
account small- and mesoscale variations in land surface properties important for runoff formation. This
can be achieved by a regional scale hydrological
model which is based on integrating the contributions
from several sub-catchments or small scale elements.
An investigation of the spatial distribution of the state
variables which determine the hydrological response
of the landscape is an initial step in the development of
such a model.
2. Study area
The NOPEX region is an area of 80  100 km2 in
the southern part of the boreal forest zone, north of
Stockholm, the capital of Sweden. The landscape is
dominated topographically and morphologically by
the very at sub-Cambrian peneplane. Altitude differences are small, with the main part of the area conned
between 30 and 70 m above the sea level, and extreme
values at 1 and 131 m. The geology is characterized by
granite, sedimentary gneiss and leptite. Lakes and
bogs of different sizes are numerous. Tills are the
dominating soil type in the research area, however,
ne-grained clay soils together with areas of sandy

308

S. Beldring et al. / Agricultural and Forest Meteorology 9899 (1999) 305324

Table 1
Characteristics of the experimental catchments (Lundin and
Halldin, 1994; Sulebak, 1997)
Basin

Area
(km2)

Buddby
0.5
Dansarhallarna 0.9
stfora
O
0.45

Fig. 1. The NOPEX region with the drainage basin of River

Fyrisan (1982 km2). The triangles show the location of the
experimental sites in areas with till. The three experimental
stfora
catchments Buddby (BUD), Dansarhallarna (DAN) and O
(OST) are marked with filled triangles. The esker site (sand and
gravel) is represented by a cross and the agricultural site (clay) by a
square. The climate station Uppsala is shown by a filled circle.

and silty material are common in the southern parts.

Areas with till soils and bogs occupy more and more
space towards the northern part. The area is covered by
coniferous forest (spruce and pine) with a small fraction of deciduous trees. Towards the south, the forest is
gradually substituted by agricultural elds (Lundin
and Halldin, 1994).
The corrected mean annual precipitation at the
station Uppsala in the southern part of the NOPEX
area (Fig. 1) is 636 mm. The precipitation increases
towards the north, with corrected mean annual values
above 700 mm. Maximum precipitation occurs in
August and minimum in February and March. Precipitation in the form of snow constitutes 2030% of
the total annual precipitation. Mean annual temperature for Uppsala is 68C with maximum monthly
mean 178C in July and minimum 58C in February
(Seibert, 1994). The river ow regime of the region is
characterized by spring snowmelt and autumn rain
high ows and winter and summer low ows
(Gottschalk et al., 1979). The ow regime is, however,
unstable and the seasonal pattern can differ from this
average pattern during individual years (Krasovskaia,
1995).
The data used in this study were collected at eight
experimental sites in different parts of the NOPEX
region in the period 19941996. Three of these sites

Dominating land use

Elevation
(m asl)

Forest (27 % swampy forest) 3345

Forest (10 % bog)
4353
Forest (30 % bog)
75103

are small experimental catchments where detailed

observations of soil moisture conditions and groundwater levels were performed along hillslopes. Tables 1
and 2 summarize some of the characteristics of these
stfora catchment is not
three catchments. The O
included in Table 2 since a digital elevation model
was not available. At the other ve sites, soil moisture
and groundwater data were collected at a limited
number of locations. The bedrock of the experimental
sites consists of massive or folded granitoids of precambrian age which are generally unweathered. The
regional hydraulic conductivity of the bedrock is of
order 107 m/s, which is considerably lower than in
the soils above, where the bulk volume of groundwater
ow takes place (Geological Survey of Sweden,
1982a, b, 1983a, b).
The three experimental catchments and three of the
other sites are located in areas covered mainly by
sandy till with low or medium boulder frequency. The
vegetation is dominated by coniferous forest, mostly
spruce. The soil thickness in the investigated catchments varies from several metres in the valley bottoms
to nearly zero on the top of ridges where exposed
bedrock is often found. The two remaining sites are
located on an esker (glaciouvial deposit of sand and
gravel) and in an agricultural area with postglacial
clay. These two experimental sites differ from the
other sites with respect to soil type, soil depth, topography and vegetation (Geological Survey of Sweden,
Table 2
Distribution of topographic units in the experimental catchments
Buddby and Dansarhallarna based on a digital elevation model with
horizontal resolution 5 m (Sulebak, 1997)
Basin

Nose (%)

Slope (%)

Hollow (%)

Buddby
Dansarhallarna

31.7
38.5

18.8
11.5

49.5
50.0

S. Beldring et al. / Agricultural and Forest Meteorology 9899 (1999) 305324

309

specic soil conditions have been established by

different investigators, e.g. Topp et al. (1980) and
Roth et al. (1990). However, the TDR-method suffers
from uncertainties due to variations in dielectric properties for soils with changing content of organic matter
(Gottschalk et al., 1995). Tallaksen and Erichsen
(1995) investigated the inuence of organic matter
on calibration curves for the till soils of the experimental basins in this study by comparison with water
content determined from soil samples by the gravimetric method. No systematic variations were found.
The soil moisture data used in this study have been
calculated from TDR-measurements with calibration
curves given by Tallaksen and Erichsen (1995) which
agrees with the results of Topp et al. (1980).
The effect on the TDR-measurements of boulders
and stones in the till has not been investigated, but
since they are bedrock fragments just like the mineral
grains, they represent parts of the soil volume with
volumetric water content equal to zero. Unrepresentative sampling will occur if boulders or stones prevent
the TDR-electrodes from being inserted at the correct
location according to the data collection program.
However, the frequency of boulders and stones within
the surface layers of soil in the experimental catchments was so low that this problem was considered
negligible.
The TDR-measurements were performed using
15 cm long electrodes installed perpendicular to the
soil surface. This means that the soil moisture was
determined in a layer below the soil surface with the
same depth. Two different instruments were used:
Tektronix 1502 C MTDR with two electrodes and
Trime TDR-system with three electrodes.
Two replicate measurements of soil moisture were
made at 5 m intervals along the transect lines in the
experimental catchments. At 25 locations along the
transects intensive soil moisture measurements were

1982c, 1989, 1991, 1993). Fig. 1 shows soil type and

location of the experimental sites within the NOPEX
region.
3. Data collection
In each of the three experimental catchments,
ground water and soil moisture data were collected
along transect lines extending from the nose prole in
the upper parts of a slope to the hollow prole at the
bottom of the slope, with the exception of transect No.
2 in the Dansarhallarna catchment where only depths
to the groundwater table were measured. The number
of transects were two in the Buddby and Dansarhal stfora catchment.
larna catchments and one in the O
Table 3 gives a description of the data collection in
each experimental catchment. In order to have a soil
depth sufcient to perform the measurements, the
transect lines did not cover the entire hillslope from
the stream to the ridge, nevertheless all three topographic units; nose, slope and hollow were represented
in each transect.
Groundwater table depths in the three experimental
catchments were observed in piezometers with diameter 2.5 cm located at irregular intervals ranging
between 1 m and 30 m along the transect lines. The
piezometers were open at the base and slotted at
intervals of 5 cm along the lower 6080 cm. They
were installed by manual drilling to the maximum
depth possible in the compact till soil, varying
between 0.6 and 1.6 m with average depth 1 m.
Soil moisture was measured using the principle of
time-domain-reectometry (TDR), which is based on
a unique relation between the volumetric water content and the dielectric constant for mineral soils.
Calibration curves for soil moisture measurements
with the TDR-method which are not restricted to
Table 3
Description of data collection in the three experimental catchments
Basin (transect No.)

Length of transect (m)

No. of piezometers

No. of soil moisture squares in different

topographic units

Buddby No. 1
Buddby No. 2
Dansarhallarna No. 1
Dansarhallarna No. 2
stfora
O

130
80
110
90
40

13
4
13
6
5

nose
nose
nose
0
nose

(1), slope (1), hollow (1)

(2), slope (2), hollow (2)
(1), slope (1), hollow (1)
(1), hollow (1)

310

S. Beldring et al. / Agricultural and Forest Meteorology 9899 (1999) 305324

Fig. 2. Mean volumetric water content from the transect lines in the experimental catchments and from two other experimental sites in the
stfora are located in areas with till, Tarnsjo on an esker (sand and gravel) and Marsta in an
NOPEX region. Buddby, Dansarhallarna and O
agricultural area (clay).

made in squares which dene the patch scale in this

paper. Mostly the squares were 8 m  8 m with 25
electrode pairs on a 2 m  2 m grid. The exception
was transect line No. 2 in the Buddby catchment
which used ve squares 10 m  10 m and nine electrode pairs at a 5 m spacing. The squares are located in
areas with different topographic characteristics, either
nose, slope or hollow. This means that each square
represents a hydrological response unit or patch in the
landscape. In general 13 piezometers were located
within the squares, the exception was transect line No.
2 in the Buddby catchment where piezometers were
stfora
located within 20 m from the squares and the O
catchment where a piezometer was located 10 m from
the square in the nose unit, in direction of the hollow
unit.
At the other ve experimental sites, soil moisture
data were measured in one or two squares with 25
equally spaced points in a grid net of size 2 m  2 m,
while groundwater levels were observed in at most
two piezometers within a distance of 20 m from the

squares. With the exception of the agricultural site,

these piezometers were installed by the Geological
Survey of Sweden. Their depth varies with soil type,
being more than 20 m at the esker site and around 5 m
at the other sites.
Fig. 2 shows mean values of volumetric water
content from the transect lines in the experimental
catchments and from two other experimental sites
during the period from April 1995 to October 1995.
Fig. 3 shows mean values of depth to the groundwater
table in topographic units along the transect lines
Buddby No. 1 and Dansarhallarna No. 1 during the
period from April 1995 to February 1996. The data
series from the nose units stop when the groundwater
table is below the bottom of the piezometers.
Three groups of three piezometers and eight pairs of
piezometers are located within squares with size less
than 10 m  10 m. Data from these piezometers show
that the depth to the groundwater table varies slowly in
space, with a maximum difference between simultaneous observations in a patch between 5 and 15 cm.

S. Beldring et al. / Agricultural and Forest Meteorology 9899 (1999) 305324

311

Fig. 3. Mean values of depth to the groundwater table in topographic units along the transect lines Buddby No. 1 and Dansarhallarna No. 1.

Fig. 4 shows groundwater table depths from squares

with three piezometers along transect lines Buddby
No. 1 and Dansarhallarna No. 1 in May and June 1996.
4. Soil moisture in the unsaturated zone
A characteristic feature of till soils is the drastic
decrease in porosity and saturated hydraulic conductivity with depth. This has the consequence that the
groundwater level follows the land surface at a shallow depth, generally between 0 and 2 m, which in turn
leads to a close relationship between groundwater
table depth and soil moisture content in the unsaturated zone (Lundin, 1982). This is due to the capillary
transport of water and means that the groundwater
level determines the possible range of uctuations of
the soil moisture content.
When there is no movement or uptake of water in a
soil prole it reaches a state of hydrostatic equilibrium
determined by the depth of the groundwater table. Soil

evaporation and extraction of water by plants cause a

decit relative to the water content at equilibrium to
develop in the root zone. After inltration of water,
this decit must be lled up before the groundwater
level can rise. During episodes of inltration the water
content in the unsaturated zone is above the equilibrium to allow vertical drainage, but it is quickly
redistributed to equilibrium with a rise in the groundwater table.
To describe the soil moisture content in the unsaturated zone we use the relationship between volumetric water content and soil moisture tension given
by Brooks and Corey (1966):
 
r
a

> a
(1)
nr
where  is the volumetric water content, r the residual
water content, the tension head of soil water, n the
porosity of the soil, a the air entry tension head and 
a parameter that depends on the pore size distribution
of the soil.

312

S. Beldring et al. / Agricultural and Forest Meteorology 9899 (1999) 305324

Fig. 4. Groundwater table depths from squares with three piezometers along transect lines Buddby No. 1 and Dansarhallarna No. 1 in May and
June 1996.

A large range of grain sizes gives till soils a

heterogeneous structure accentuating the spatial variation of soil properties (Nyberg, 1995). The porosity
shows a systematic variation with depth, with the
highest values near the ground surface (Johansson,
1986; Nyberg et al., 1993). Freezing and thawing, root
and soil fauna activities and chemical processes
change the texture and structure of the soil (Lundin,
1982). These processes are most active in the upper
layers. The air entry tension and the pore size distribution do not show a systematic variation with depth
(Udns, 1991; Stahli et al., 1995).
The residual water content is dened as the water
content at which the water lms in the soil become
discontinuous (Dingman, 1994). Soil moisture retention curves for till soil samples from the NOPEX
Central Tower Site were used to determine the residual
water content as a parameter in Brooks and Corey's
relation. Stahli et al. (1995) found that r was of the
order 103 which is much smaller than the volumetric
water content and the porosity. By neglecting r and
assuming a state of vertical hydrostatic equilibrium,

the volumetric water content in the unsaturated zone is

given by:

 e nz
 e nz



hz >

hz
hz 

(2)

where the z-axis is assumed positive downwards with

zero at the ground surface and h the depth to the
groundwater table. The porosity, nz, depends on the
depth below the ground surface.
The equilibrium water content at the soil surface
(z 0) is equal to:
 
a
h> a
(3)
e0 n0
h
e0 n0

h

where n0 is the porosity near the ground surface.

In the analysis that follows soil water content at the
surface given by Eq. (3) will be compared to measured
soil water content from the upper 15 cm of the soil.

S. Beldring et al. / Agricultural and Forest Meteorology 9899 (1999) 305324

Strictly, the measured water content could be considered to apply to a soil depth of 7.5 cm, however it is
assumed to be valid at the ground surface.
The equilibrium state assumes that the soil moisture
tension at a given depth depends on the distance to the
groundwater table. In the root zone the water content
is generally below the equilibrium value due to evapotranspiration. Since sufcient data on root geometry
or root pressures are not known, this soil moisture
decit must be evaluated by treating the root system as
a lumped system which penetrates the upper layer of
the soil uniformly (Guymon, 1994). The moisture
content at the surface is then given by:
 
a
h> a
(4)
 0 n0
r0 h
 0 n0

h

The factor r0 is the ratio between the actual soil

moisture tension at the surface and the soil moisture
tension at equilibrium. If the groundwater table
is below the depth given by the air entry tension
head, r0 is generally >1. During episodes of infiltration, the soil moisture content may rise above
the equilibrium value with r0 < 1, followed by a rise
of the groundwater table and a new equilibrium
state in the unsaturated zone. When the depth to the
groundwater table is less than or equal to the air entry
tension head, the soil is saturated at the surface and
r0 1.
The depth to the groundwater table and the parameters of Brooks and Corey's equation; porosity, air
entry tension and pore size distribution index varies in
space and may be considered as sources of variability
of the volumetric soil moisture at the soil surface. In
addition, so will microscale topography and the factor
r0. Although the depth to the groundwater table is
nearly constant at the patch scale (Fig. 4), the volumetric soil moisture in the upper 15 cm varies considerably over short distances due to variations in soil
characteristics, root concentration and microscale
topography. This implies that the variability of surface
soil moisture conditions at the patch scale depends on
the heterogeneous nature of soil characteristics.
Above a certain threshold scale, the variations of soil
characteristics are smoothed out and the spatial variability of soil moisture content is determined by the
depth to the groundwater table.

313

The lack of knowledge regarding the factors controlling the soil moisture content can be expressed by
treating them as random variables (Romanowicz et al.,
1995). In principle, it would be correct to consider a
joint distribution function of all relevant factors which
determine the volumetric water content at the soil
surface. However, in order to make the problem
mathematically tractable, some simplications must
be made and the variability of porosity at the patch
scale is assumed to account for the variability in the
other soil parameters as well as the variability in root
concentration, microscale topography and the factor
r0. The reason for selecting porosity as the stochastic
variable which is explaining the variability in soil
characteristics is that it is the parameter in Brooks
and Corey's equation which resembles most what we
want to model; the volumetric water content. In
addition, the study by Tallaksen and Erichsen
(1995) shows a variability in saturated water content
at small scale which cannot be neglected. This
approach implies that air entry tension head and pore
size distribution index are considered as effective
parameters, while r0 is considered constant at the
patch scale.
At the catchment scale, we assume that the depth to
the groundwater table is the main factor determining
variations in soil moisture at the surface. The frequency distribution of volumetric soil moisture near
the ground surface at this scale can now be determined. However, we must take into account the following result from the experimental catchments: at a
given instant of time, r0 increases with increasing
depth to the groundwater table (Table 4). This means
that the soil moisture decit relative to the equilibrium
value is largest in the upper parts of slopes where the
groundwater table is deep, a result which is supported
by Lundin (1982). This is the recharge area where a
vertically downward movement of groundwater leaves
less water available for lling the decit than in the
discharge areas at the bottom of slopes where an
upward directed groundwater ow supplies water.
Since only limited information about this relationship
is available, we assume that r0 at a given instant of
time has a linear variation over a range of values
determined by the observed depths to the groundwater
table:
r0 Ah B

(5)

314

S. Beldring et al. / Agricultural and Forest Meteorology 9899 (1999) 305324

Table 4
Mean depth to the groundwater table within squares along transect line No. 1 in the Buddby catchment, vs. r0, the ratio between the actual soil
moisture tension at the soil surface and the soil moisture tension at equilibrium, as given by Eq. (11)
Date

Groundwater
table depth (m)

r0

Groundwater.
table depth (m)

r0

Groundwater.
table depth (m)

r0

25/04/95
03/05/95
05/05/95
17/05/95
24/05/95
31/05/95
07/06/95
06/07/95

0.055
0.025
0.030
0.300
0.045
0.080
0.080
0.240

1.000
1.000
1.000
1.000
1.000
1.000
1.000
0.610

0.055
0.020
0.040
0.040
0.090
0.195
0.240
0.390

1.000
1.000
1.000
1.000
1.000
1.933
1.977
1.191

0.340
0.140
0.250
0.200
0.375
0.580
0.725
1.215

5.691
4.098
3.849
6.253
4.859
7.376
7.633
4.458

where
A

rmax rmin
;
hmax hmin

B rmin hmin

rmax rmin
hmax hmin

The soil physical parameters of the surface layer have

been determined from available data and by reference
to other investigations in areas with till soils. The
porosity of samples from the upper horizons of till
soils is generally in the range 0.50.7 (Lundin, 1982;
Johansson, 1986; Norden, 1991; Udns, 1991;
Nyberg et al., 1993; Stahli et al., 1995). This is in
agreement with maximum values of water content in
samples from the experimental basins (Tallaksen and
Erichsen, 1995). The air entry tension head in till soils
is generally in the range 0.10.2 m (Grip and Rodhe,
1994). Soil moisture characteristic curves determined
from till soil samples by Udns (1991) and Stahli et al.
(1995) gave values of a below 0.1 m, while Johansson (1986) found good agreement between empirical
and theoretical pF-curves using a 0.1 m. The value
of a 0.09 m has been chosen in this study. When
the depth to the ground water table is <0.09 m, the
mean value of observed volumetric soil moisture near
the surface is equal to 0.6 (Fig. 5), which is used as a
mean value of porosity near the soil surface at the
patch scale. Some values in Fig. 5 are above 0.6,
however a certain range of values of saturated water
content must be expected. Results from Udns (1991)
and Stahli et al. (1995) give values of the pore size
distribution index, , in the range 0.110.3. With the
selected values of porosity and the air entry tension
head, we find that  0.3 gives a good agreement
between empirical and theoretical distribution functions of soil moisture at patch and catchment scales.

These values of the soil physical parameters are used

at both scales. At the patch scale, the porosity of the
soil near the surface varies in space, while the mean
porosity, the air entry tension head and the pore size
distribution index are assumed constant. This means
that the spatial variation of soil characteristics is
accounted for by the porosity. At the catchment scale
the soil physical parameters of the surface layer are
assumed constant in space.
The simplications regarding the sources of variability of soil moisture at patch and catchment scales
are appropriate since the purpose of this paper is not to
give a detailed description of soil characteristics, root

Fig. 5. Mean volumetric soil moisture vs. mean depth to the

groundwater table for each set of observations in the squares of the
experimental catchments.

S. Beldring et al. / Agricultural and Forest Meteorology 9899 (1999) 305324

concentration and microscale topography at the patch

scale, but rather to describe the distribution of the state
variables which determine the hydrological response
of catchments at scales ranging from 10 to 104 m. A
detailed description at the patch scale would require a
different data collection program which in turn would
prevent collection of data from a large area. We have
assumed that the physical parameters of the upper
layer of the soil are constant in space at the catchment
scale. This assumption is justied by Nyberg (1995)
who considers it possible to establish catchment representative values for porosity and water retention and
Lundin (1982) who shows that the porosity values do
not show large differences between subareas of a
catchment.
5. Patch scale variability
If the inuence of variability of the air entry tension
head, a, and the pore size distribution index, , is
neglected at the patch scale, the vertical equilibrium
model states that the variations in volumetric water
content in the upper soil layers is caused by variations
in the stochastic variable N0, the porosity near the
ground surface. Since N0 has a lower bound of 0 and an
upper bound of 1 its variation can be described by the
b-distribution which takes values in the same interval
(Yevjevich, 1972). The b-distribution ts easily to
empirical data since it can take a variety of shapes
without restrictions regarding, e.g. symmetry. It has
previously been used for approximating frequency
distributions of recorded soil properties (Haskett
et al., 1995). The b-frequency distribution is given
by (Kendall, 1986):
fN0 n0

1
np1 1n0 q1
Bp; q 0

p > 0;

q>0

0  n0  1;
(6)

where B(p,q) is the b-function and p and q are parameters.

The mean and variance of the b-distribution are
given by:
p
(7)
EN0 N0
pq
pq
(8)
VarN0 2N0
p q 1p q2

315

When the depth to the groundwater table in a patch is

larger than the air entry tension head, the frequency
distribution of the stochastic variable 0, the volumetric soil moisture content near the ground surface,
can be found by transforming the distribution of N0.
The frequency distribution of a monotonic function
Y(X) of a univariate continuous random variable X is
given by (Haan, 1977):
 
dx
(9)
fY yfX x 
dy
Combining Eqs. (4), (6) and (9) we find:
"   #p1
 
r0 h
1
r0 h 
0
f0 0
Bp; q
a
a
"
#
  q1
r0 h
 10
h > a;
a

 0 < n0
(10)

The mean and variance of this distribution can be

expressed in terms of those of the b-distribution of
N 0:
 
a
(11)
E0 0 N0
r0 h
 2
a
2
2
(12)
Var0 0 N0
r0 h
For each set of data from a patch, r0 can be calculated
from Eq. (11) using the sample mean of volumetric
soil moisture and depth to the groundwater table
within the square.
Method of moments estimators of the parameters p
and q are given by:
"
 
  #
m0
r0 h
r0 h
2
2
^p 2 m0 m0
s0
(13)
s0
a
a
"
 
  #
a =r0 h m0
r0 h
r0 h
2
2
^q
m0 m0
s0
2
s0
a
a
(14)
s20

is the sample mean and variance

where m0 and
of 0.
When the depth to the groundwater table in a patch
is less than the air entry tension head, the vertical
equilibrium model states that the entire soil prole is

316

S. Beldring et al. / Agricultural and Forest Meteorology 9899 (1999) 305324

Fig. 6. Cumulative probability distributions of soil moisture within squares of size 8 m  8 m along transect line No. 1 in the Buddby
catchment on 6 July 1995 based on observed data (points) and the theoretical expression in Eq. (10) (lines).

saturated. This means that the frequency distribution

of 0 is equal to the frequency distribution of N0.
The theoretical distribution function of volumetric
soil moisture near the surface at the patch scale is
given by Eq. (10). For each set of simultaneously
observed data in a square, the parameters of this
distribution were estimated using Eqs. (13) and
(14). Fig. 6 shows cumulative distribution functions
of soil moisture at the patch scale in Buddby on 6 July
1995. The observed soil moisture varies considerably
within the patches. This is in agreement with results
from Myrab (1986) and Nyberg (1996) who observed
large variations in soil moisture content over short
distances in catchments with till soils in southeastern
Norway and southwestern Sweden. Fig. 6 also illustrates the variability of soil moisture conditions along
a hillslope.

groundwater table, at the catchment scale at a given

instant of time. The g-distribution takes values in the
range from 0 to 1 just like the variable we want to
model and it has the exibility in shape which is
necessary to t empirical data. It has been widely
used in hydrology (Haan, 1977). The g-frequency
distribution is given by (Kendall, 1986):
fH h

 1 h
h e

h  0;

 > 0;

>0
(15)

where ( ) is the gamma function and  and the

parameters.
The mean and variance of the g-distribution are
given by:
EH H

(16)

2

6. Catchment scale variability

VarH 2H

We select the two parameter g-distribution distribution to describe the stochastic variable H, depth to the

Method of moments estimators of the parameters 

and are given by:

(17)

S. Beldring et al. / Agricultural and Forest Meteorology 9899 (1999) 305324

mH
s2H

(18)

m2H
s2H

(19)

where mH and s2H is the sample mean and variance of

H.
The frequency distribution of volumetric soil moisture near the ground surface consists of two parts; one
continuous part for the case when the depth to the
groundwater table is larger than the air entry tension
head and one discrete probability for saturation near
the soil surface. The continuous part is found by
combining Eqs. (4), (5), (9) and (15):




1=
 a n0
1 CB 1
CB
exp

f0 0
11= C
2A
2A

0

h>

a;

0 < n0

(20)

where
"

 1= #1=2
n0
;
C B 4A a
0
2

A and B are defined in Eq. (5).

The probability of saturation at the soil surface is
equal to the probability that the depth to the groundwater table is less than or equal to the air entry tension
head. This probability is given by the incomplete
g-function:
Z  a
1
et t 1 dt
(21)
f0 n0 P ;  a
0
Data from the hillslopes describe the different moisture conditions in the experimental catchments and can
therefore be considered representative for conditions
at the catchment scale. The distribution function of
volumetric soil moisture at this scale is given by Eqs.
(20) and (21). For a given situation, the parameters of
this distribution can be found from Eqs. (18) and (19)
using the sample mean and variance of depth to the
groundwater table along a slope. A, B and C in Eqs. (5)
and (20) are found by inserting minimum and maximum values of r0 from the squares along the transects
as given by Eq. (11). Figs. 79 show cumulative
distribution functions of soil moisture along hillslopes
stfora catchments
in the Buddby, Dansarhallarna and O
on two occasions with different moisture conditions.

317

Fig. 7. Cumulative probability distributions of soil moisture along

transect line No. 1 in the Buddby catchment based on observed data
(points) and the theoretical expression in Eqs. (20) and (21) (lines).

These and similar data on soil moisture variations

along the slopes confirm that the groundwater level
controls the soil moisture conditions in the unsaturated
zone in the experimental catchments. The step in the
distribution function gives the proportion of the catchment area where the depth to the groundwater table is
less than the air entry tension head. These are the
discharge areas with a shallow groundwater table and

Fig. 8. Cumulative probability distributions of soil moisture along

transect line No. 1 in the Dansarhallarna catchment based on
observed data (points) and the theoretical expression in Eqs. (20)
and (21) (lines).

318

S. Beldring et al. / Agricultural and Forest Meteorology 9899 (1999) 305324

Fig. 9. Cumulative probability distributions of soil moisture along

stfora catchment based on observed data
the transect line in the O
(points) and the theoretical expression in Eqs. (20) and (21) (lines).

saturated conditions near the soil surface where saturation excess overland flow will occur during precipitation or snowmelt (Anderson and Burt, 1990). This
catchment proportion varies with time as a result of
varying precipitation, evapotranspiration and downslope flow through the soil. The total area which can
be expected to contribute actively to runoff during
storms is larger, since the high proportion of groundwater must be caused by water which has infiltrated in
parts of the recharge area (Rodhe, 1987).
7. Regional scale variability
Soil moisture and groundwater data from catchments in till soils in different parts of the NOPEX
region show more or less identical uctuations (Figs. 2
and 3), indicating similar hydrological behaviour. This
similarity suggests that above a certain threshold scale
a catchment might contain a sufcient sample of the
geology, topography, vegetation and other landscape
characteristics of a region to decrease the variability of
catchment average uxes to a minimum. This is a
prerequisite of the representative elementary area
concept, which assumes that the distribution of characteristics within a catchment may be important in
determining these uxes, but the pattern of those
characteristics is no longer important (Wood et al.,
1988, 1990). The REA can be interpreted as a scale at

which the local patterns in runoff production are

sufciently well integrated to produce a similarity
in response, before non-stationarity in catchment
characteristics or hydrological processes starts to
increase the variances at larger scales. Such nonstationarity may be due to changes in geology, physiography or land-use, changes in the scale of rainfall
variability or changes in the nature of hydrological
processes (Beven, 1995). The existence of a REA is
equivalent to a minimum in the power spectrum of
catchment properties. This refers to a process consisting of a small scale component superimposed on a
much larger component. The REA relates to a preferred element size for distributed catchment modelling of large regions or drainage basins, since this
requires deterministic representation of the large scale
variability as different values in different model elements and parameterization of small scale processes
within the elements. Provided a minimum in the power
spectrum of the processes which governs catchment
response exists, the same parameterization can be used
for a range of spatial scales for catchments with
similar distributions of catchment characteristics
and precipitation (Bloschl et al., 1995).
The state variables which are at the focus of this
work determine how the climatic input to a catchment
is transformed to uxes in the form of evapotranspiration and runoff. According to Beven (1995) it is the
distribution of hydrological responses in the landscape
which must be achieved to provide realistic predictions of discharge and evapotranspiration uxes within
heterogeneous terrain. A criterion for the existence of
a scale at which the average hydrological response
varies only slowly with increasing catchment area is
that the variability of the responses between different
areas falls to acceptably low values for a sufciently
large area, for larger areas this variability may rise
again (Wood et al., 1988, 1990). This effect must also
be manifest in the mechanisms which control the
responses of the catchments, i.e. the state variables
soil moisture content and groundwater level.
Mean values of volumetric water content and depth
to the groundwater table over areas with increasing
size have been calculated using all available observations from experimental sites in forest areas with till
soils at specic instants of time. Figs. 10 and 11 show
that the large variability between areas of similar size
in till soils decreases as the averaging area increases

S. Beldring et al. / Agricultural and Forest Meteorology 9899 (1999) 305324

319

Fig. 10. Mean volumetric water content in the upper 15 cm of the soil versus size of the averaging area. Data from experimental sites in areas
with till on 8 May and 19 June 1996.

from the patch scale (100 m2) to the size of the small
experimental catchments (1 km2). By comparison,
averaging data from experimental sites in different
landscape and soil types does not yield the same
decrease in variability, as shown for soil moisture
data in Fig. 12. Groundwater data from different
landscape and soil types vary even more, as the depth

to the groundwater table in eskers is of the order of

1020 m.
8. Conclusions
This study has focused on the spatial variation of
soil moisture and groundwater levels in small catch-

320

S. Beldring et al. / Agricultural and Forest Meteorology 9899 (1999) 305324

Fig. 11. Mean depth to the groundwater table versus size of the averaging area. Data from experimental sites in areas with till on 8 May and 19
June 1996.

ments in areas dominated by boreal forest and till

soils. These state variables are the main factors controlling the response of a drainage basin on climatic
input in this landscape type. The spatial variability of
surface soil moisture conditions within hydrological
response units at the patch scale depends on the

heterogeneous nature of soil characteristics. Above

a certain threshold scale, the variations of soil characteristics are smoothed out and the spatial variability
of soil moisture content is determined by the depth to
the groundwater table. Statistical distribution functions of soil moisture content conditioned on the depth

S. Beldring et al. / Agricultural and Forest Meteorology 9899 (1999) 305324

321

Fig. 12. Mean volumetric water content in the upper 15 cm of the soil versus size of the averaging area. Data from experimental sites in areas
with till, on an esker (sand and gravel) and in an agricultural area (clay) on 8 May and 19 June 1996.

to the ground water table are well suited to the

description of the observed patterns of soil moisture
conditions at patch and catchment scales. The area
which is capable of saturation excess runoff generation can be identied from the distribution function of
soil moisture at the catchment scale.

Since the soil moisture content is conditioned on the

depth to the groundwater table, which in its turn is
closely related to topography, the variations of soil
moisture content in a catchment are also controlled by
topography. The observed data show that moisture
conditions vary along a hillslope, with high soil

322

S. Beldring et al. / Agricultural and Forest Meteorology 9899 (1999) 305324

moisture content and shallow groundwater table at the

bottom of hillslopes and low soil moisture content and
deep groundwater table at the top of hillslopes. The
spatial variation of soil moisture decit expressed by
the ratio between the actual soil moisture tension in
the root zone and the soil moisture tension at equilibrium conrms the signicance of topography for
moisture conditions. In many cases the measurements
of hydrological state variables are too widely spaced
and the natural variability too large for reliable estimation based on observations. When this occurs the
results of this study indicate that topography can be
used as a covariate in order to estimate the spatial
distribution of wetness states in a catchment.
The small variability in hydrological behaviour
between catchments located in till soils in different
parts of the NOPEX region is due to more or less
identical climatic input and catchment geology, soil
type, topography and vegetation. However, the variability may also be attributed to difculties with obtaining a representative sample of moisture conditions for
all the different geomorphological units in a catchment, which would require a more extensive data
collection program. Nevertheless, the results from this
study conrm that the dynamics of hydrological
response units with similar physiographic conditions
are identical in different parts of the NOPEX region.
This is in agreement with results from Krasovskaia
(1985) who showed that groundwater and soil moisture conditions differed signicantly between hydrological response units within a catchment, but not
between basins with similar physiographic conditions
or between the same type of units in different basins
with similar climatic input. The conclusion is that
moisture conditions in the till soils of the NOPEX
region are dominated by variations in topography,
vegetation and soil characteristics at scales <12 km2.
Although a separation of scales or equivalently a
minimum in variability between small and large scale
processes could not be detected, the similarity in
groundwater and soil moisture conditions suggests
that the variability of catchment responses in the form
of evapotranspiration and runoff decreases to a minimum at this spatial scale. This means that the same
parameterization can be used for modelling hydrological processes in catchments or grid elements at
scales of ca. 1 km2 in till soils in different parts of the
NOPEX region.

Acknowledgements
The data used in this study were collected by
students and staff at Uppsala University, Department
of Earth Sciences and University of Oslo, Department
of Geophysics. We are particularly indebted to Dr.
Allan Rodhe, Uppsala University for organization of
eld work and stimulating discussions. The eld work
has been made possible through grants from the
Nordic Council of Ministers. This work has been
carried out within the framework of NOPEX a
NOrthern hemisphere climate Processes land-surface
EXperiment. The data used in this investigation comes
from SINOP the System for Information in NOPEX
(Lundin et al., 1998; Rodhe et al., 1999; Tallaksen
et al., 1999).

References
Aanderson, M.G., Burt, T.P., 1990. Subsurface runoff. In:
Anderson, M.G., Burt, T.P. (Eds.), Process Studies in Hillslope
Hydrology, Wiley, Chichester, pp. 365400.
Beven, K., 1995. Linking parameters across scales: subgrid
parameterizations and scale dependent hydrological models.
In: Kalma, J.D., Sivapalan, M. (Eds.), Scale Issues in
Hydrological Modelling, Wiley, Chichester, pp. 263281.
Beven, K., Kirkby, M.J., 1979. A physically based, variable
contributing area model of basin hydrology. Hydrological
Sciences Bulletin 24, 4369.
Beven, K., Wood, E.F., 1983. Catchment geomorphology and the
dynamics of runoff contributing areas. J. Hydrology 65, 139
158.
Bloschl, G., Grayson, R.B., Sivapalan M., 1995. On the
Representative Elementary Area (REA) concept and its utility
for distributed rainfall-runoff modelling. In: Kalma, J.D.,
Sivapalan, M. (Eds.), Scale Issues in Hydrological Modelling,
Wiley, Chichester, pp. 7188.
Bonell, M., 1993. Progress in the understanding of runoff
generation dynamics in forests. J. Hydrology 150, 217275.
Brooks, R.H., Corey, A.T., 1966. Properties of porous media
affecting fluid flow. Proc. Am. Soc. Civ. Eng. J. Irrigation and
Drainage Division 92 (IR-2) 6188.
Dingman, S.L., 1994. Physical Hydrology, Prentice-Hall, New
Jersey, 576 pp.
Espeby, B., 1990. Tracing the origin of natural waters in a glacial
till slope during snowmelt. J. Hydrology 118, 107127.
Flugel, W.-A., 1995. Delineating hydrological response units by
geographical information system analysis for regional hydrological modelling using PRMS/MMS in the drainage basin of
river Brol, Germany. In: Kalma, J.D., Sivapalan, M. (Eds.),
Scale Issues in Hydrological Modelling, Wiley, Chichester, pp.
181194.

S. Beldring et al. / Agricultural and Forest Meteorology 9899 (1999) 305324

Geological Survey of Sweden, 1982a. Bedrock Map of Vastmanland County, Ah 2.
Geological Survey of Sweden, 1982b. Description and Appendices
to the Hydrogeological Map of Vastmanland County, 66 pp.
sthammar
Geological Survey of Sweden, 1982c. Quaternary Map O
SV, Ae 53.
Geological Survey of Sweden, 1983a. Bedrock Map of Uppsala
County, Ah 5.
Geological Survey of Sweden, 1983b. Description and Appendices
to the Hydrogeological Map of Uppsala County, 84 pp.
Geological Survey of Sweden, 1989. Quaternary Map Soderfors
SO, Ae 104.
Geological Survey of Sweden, 1991. Quaternary Map Enkoping
NO, Ae 110.
Geological Survey of Sweden, 1993. Quaternary Map Uppsala NV,
Ae 113.
Gottschalk, L., Jensen, J.L., Lundquist, D., Solantie, R., Tollan, A.,
1979. Hydrologic regions in the Nordic countries. Nordic
Hydrology 10, 273286.
Gottschalk, L., Tallaksen, L., Rodhe, A., 1995. Boreal hydrology:
Investigations during NOPEX CFE 1, Institute Report Series
No. 92. Department of Geophysics. University of Oslo,
Norway, 21 pp.
Grip, H., Rodhe, A., 1994. Vattnets vag fran regn till back, Hallgren
& Fallgren Studieforlag AB, Uppsala, 156 pp (in Swedish).
Guymon, G.L., 1994. Unsaturated zone hydrology, Prentice Hall,
Englewood Cliffs, New Jersey, 210 pp.
Haan, C.T., 1977. Statistical methods in hydrology, Iowa State
University Press, Ames, Iowa, 378 pp.
Hack, J.T., Goodlett, J.C., 1960, Geomorphology and forest
ecology of a mountain region in the Central Appalachians,
US Geological Survey Professional Paper No. 347, 66 pp.
Haldorsen, S., Kruger, J., 1990. Till genesis and hydrogeological
properties. Nordic Hydrology 21, 8194.
Halldin, S., Gottschalk, L., Gryning, S.E., Jochum, A., Lundin,
L.C., Van de Griend, A.A., 1999. Energy, water and carbon
exchange in a boreal forest NOPEX experiences. Agric. For.
Meteorol. 9899, 529.
Haskett, J.D., Pachepsky, Y.A., Acock, B., 1995. Use of the beta
distribution for parameterizing variability of soil properties at
the regional level for crop yield estimation. Agricultural
Systems 48, 7386.
Johansson, P.-O., 1986. Diurnal groundwater level fluctuations in
sandy till a model analysis. J. Hydrology 87, 125134.
Kendall, M., 1986. Kendall's advanced theory of statistics. In:
Stuart, A., Ord, J.K. (Eds.), Distribution Theory, 5th ed., vol. 1.
Griffin, London, 604 pp.
Krasovskaia, I., 1985. To the problem of joint soil moisture and
groundwater analysis in basin response studies. Catena 12,
241254.
Krasovskaia, I., 1995. Quantification of the stability of river flow
regimes. Hydrological Sciences J. 40, 587598.
Lind, B.B., Lundin, L., 1990. Saturated hydraulic conductivity of
Scandinavian tills. Nordic Hydrology 21, 107118.
Lundin, L., 1982. Soil moisture and groundwater in till soil and the
significance of soil type for runoff, Ph.D. thesis, UNGI Report
No. 56, Division of Hydrology, Department of Physical

323

Geography, Uppsala University, Sweden, 216 pp (in Swedish,

English summary).
Lundin L.C., Halldin S., 1994. Experiment plan for NOPEX.
Department of Earth Sciences, Section of Hydrology. Uppsala
University, Sweden, 80 pp.
Lundin L.C., Halldin S., Nord, T., 1998. System of information in
NOPEX Retrieval, use and query of climate data. Agric. For.
Meteorol. 9899, 3151.
Myrab, S., 1986. Runoff studies in a small catchment. Nordic
Hydrology 17, 335346.
Norden, L.G., 1991. Soil water characteristics in glacial till at two
forest sites in Sweden. Scandinavian J. Forest Research 6, 289
304.
Nyberg, L., 1995. Water flow path interactions with soil hydraulic
properties in till soil at Gardsjon, Sweden. J. Hydrology 170,
255275.
Nyberg, L., 1996. Spatial variability of soil water content in the
covered catchment at Gardsjon, Sweden. Hydrological Processes 10(1), 89103.
Nyberg, L., Bishop, K., Rodhe, A., 1993. Importance of hydrology
in the reversal of acidification in till soils. Gardsjon, Sweden.
Applied Geochemistry Suppl. Issue 2, pp. 6166.
O'Loughlin, E.M., 1981. Saturation regions in catchments and their
relations to soil and topographic properties. J. Hydrology 53,
229246.
O'Loughlin, E.M., 1986. Predicition of surface saturation zones in
natural catchments by topographic analysis. Water Resources
Research 22(5), 794804.
Rodhe, A., 1981. Spring flood, meltwater or groundwater?. Nordic
Hydrology 12, 2130.
Rodhe, A., 1987. The origin of streamwater traced by oxygen-18,
Ph.D. thesis, Report Series A No. 41. Division of Hydrology.
Department of Physical Geography. Uppsala University,
Sweden, 260 pp.
Rodhe, A., Beldring, S., Erichsen, B., Gottschalk, L., Seibert, J.,
1999. Spatial variability in runoff, soil moisture and groundwater levels and its relation to terrain parameters. Agric. For.
Meteorol., this issue, Documented dataset on appended CD.
Romanowicz, R., Dooge, J.C.I., O'Kane, J.P., 1995. Spatial
variability of evaporation from the land surface random
initial conditions. In: Kundzewicz, Z.W. (Ed.), New Uncertainty Concepts in Hydrology. Cambridge University Press,
Cambridge, pp. 197205.
Roth, K., Schulin, R., Fluhler, H., Attinger, W., 1990. Calibration
of time domain reflectometry for water content measurements
using a composite dielectric approach. Water Resources
Research 26(10), 22672273.
Seibert, P., 1994. Hydrological characteristics of the NOPEX
research area, NOPEX Technical Report No. 3. Uppsala
University, Sweden, 51 pp.
Sklash, M.G., 1990. Environmental isotope studies of storm and
snowmelt runoff generation. In: Anderson, M.G., Burt, T.P.
(Eds.), Process Studies in Hillslope Hydrology, Wiley,
Chichester, pp. 401435.
Sklash, M.G., Farvolden, R.N., 1979. The role of groundwater in
storm runoff. J. Hydrology 43, 4565.
Stahli, M., Hessel, K., Eriksson, J., Lindahl, A., 1995. Physical and

324

S. Beldring et al. / Agricultural and Forest Meteorology 9899 (1999) 305324

chemical description of the soil at the NOPEX Central Tower

Site, NOPEX Technical Report No. 16. Swedish University of
Agricultural Sciences, Uppsala, Sweden, 19 pp.
Sulebak, J.R., 1997. Geomorphometric studies of different
topographic regions. Analysis and applications from Norway
and Sweden, Ph.D. thesis, Report No. 8. Department of
Geography. University of Oslo, Norway, 204 pp.
Tallaksen, L.M., Erichsen, B., 1995. Intercalibration of soil
moisture measurements using oscilloscope and TDR devices,
Technical Note. Department of Geography. University of Oslo,
Norway, 13 pp.
.Tallaksen, L., Gottschalk, L., Erichsen, B., Lystad, S., Olsen, F.,
Saelthun, N.R., 1999. Modelling hydrological response to
evapotranspiration losses from different types of land cover.
Agric. For. Meteorol. this issue, Documented dataset on
appended CD.
Thompson, J.C., Moore, R.D., 1996. Relations between topography
and water table depth in shallow forest soil. Hydrological
Processes 10, 15131525.
Topp, C.G., Davis, J.L., Annan, A.P., 1980. Electromagnetic
determination of soil water content: measurements in coaxial
transmission lines. Water Resources Research 16(3), 574582.

Udns, H.-C., 1991. En beskrivelse av grunnvannsmagasinets

egenskaper og markvannstrmningen i et lite skogsfelt, Thesis.
Department of Geophysics. University of Oslo, Norway, 97 pp
(in Norwegian).
Wood, E.F., 1995. Scaling behaviour of hydrological fluxes and
variables: empirical studies using a hydrological model and
remote sensing data. In: Kalma, J.D., Sivapalan M., (Eds.),
Scale Issues in Hydrological Modelling, Wiley, Chichester, pp.
89104.
Wood, E.F., Sivapalan, M., Beven, K., Band, L., 1988. Effects of
spatial variability and scale with implications to hydrological
modelling. J. Hydrology 102, 2947.
Wood, E.F., Sivapalan, M., Beven, K., 1990. Similarity and scale in
catchments storm response. Reviews of Geophysics 28(1), 1
18.
Woods, R., Sivapalan M., Duncan, M., 1995. Investigating
the Representative Elementary Area concept: an approach
based on field data. In: Kalma, J.D., Sivapalan, M. (Eds.), Scale
Issues in Hydrological Modelling, Wiley, Chichester, pp. 49
70.
Yevjevich, V., 1972. Probability and Statistics in Hydrology, Water
Resources Publications, Colorado, 302 pp.