Geostatistical Methods for Predicting
Soil Moisture Continuously in a Subalpine Basin
Katherine E. Williams and Sharolyn J. Anderson
Abstract
This study presents spatial statistical methods for examining
the distribution of soil moisture in a sub-alpine environ-
ment. The high local variability of soil moisture is not well
characterized by spatial interpolation from dispersed data
points. Interpolation using only field samples from Loch
Vale, Rocky Mountain National Park, Colorado produced
coarse estimates that followed mean soil moisture trends,
but failed to capture local mid-slope variation. A properly
specified regression model was identified by using dispersed
field samples and ancillary data derived from Ikonos-2 and
lidar data. This model predicted soil moisture patterns at a
much finer spatial resolution. An intensive field campaign
provided independent soil moisture measurements that
were used to assess the model’s accuracy. The modeled soil
moisture estimates captured local variability associated with
topographic terrain differences along mid-slope areas.
Introduction
Soil moisture is an important environmental variable with
implications for hydrologic and climate change research in
sub-alpine environments such as Loch Vale watershed in
Rocky Mountain National Park, Colorado. As temperatures in-
crease worldwide, mountain environments are recognized as
regions that will be significantly affected. These systems hang
in a delicate balance between short growing seasons, harsh
environmental extremes, and dynamic processes related to
steep verticality (Funnell and Parish, 2001). Soil moisture is
an important part of mountain ecosystems and is related to
vegetation distributions, habitat boundaries, erosion poten-
tial, and change over time. Soil facilitates water transport and
supports vegetation structures that are sensitive to change
(Molotch
et al.
, 2009). Soil moisture regulates plant commu-
nity distribution and vigor, nutrient turnover in the soil, and
response to precipitation events. In high elevation ecosys-
tems, the maintenance of soil moisture is critical because the
soils are often shallow with poorly developed organic hori-
zons (Reuth
et al.
, 2003; Molotch
et al.
, 2009). Water tends to
evaporate more readily in these soils due to limited organic
material on the surface, leading to a disruption of moisture
regulated soil processes. As air temperatures increase, the
evaporation rate will also increase. Studies of the relation-
ship between vegetation, seasonal snow accumulation and
melt, ground and air temperatures, and topography and soil
moisture reveal complex, interdependent relationships (Kim
and Kim, 2007; Molotch
et al.
, 2009; Emanuel
et al.
, 2010;
Kampf
et al.
, In review). Understanding the distribution and
dynamics of soil moisture will help researchers and managers
identify areas susceptible to effects of climate change, such as
drought stress and increased erosion potential.
The intent of this Soil Moisture Spatial Survey (
SMSS
)
Study was to examine the relationships between soil moisture
and other physical variables such as land cover and micro-
topographical features such as bedrock outcrops and seeps.
However, soil moisture is difficult to predict or interpolate
based on field measurements alone because soil moisture ex-
hibits extreme local, spatial, and temporal variability (Kampf
and Burges, 2007). Generating a detailed predicted surface us-
ing standard interpolation methods results in smoothing that
mutes the real local variability of soil moisture in the field.
Other studies have shown that this local variability is often
associated with topographic features and suggested further
study of these relationships (Kampf and Markus, 2009). The
SMSS
Study was an effort to use both field samples and geosta-
tistical methods to provide a soil moisture surface based on
robust statistical predictions. A continuous surface of pre-
dicted soil moisture throughout a study area is desirable for
examining relationships with other environmental variables
as well as for other modeling purposes and identifying target
areas for local studies.
The
SMSS
used a combination of field sampling, regres-
sion, and interpolation techniques. This research extends
the Hillslope Study conducted in Loch Vale (Kampf and
Markus, 2009; Kampf
et al.
, In review). The
SMSS
Study used
a dispersed method for field measurements in contrast to
the Hillslope Study where soil moisture measurements were
taken at regular intervals along linear transects. The Hillslope
Study’s data were used as an independent validation of the
prediction surface we generated. The data products of the
SMSS
Study allow researchers to analyze soil moisture across
the larger study area where the two Hillslope transects are
undergoing extensive research.
Site Description and Background on the Hillslope Study
The
SMSS
Study was located in the Loch Vale watershed in
Colorado (Figure 1). High elevation and position at the Conti-
nental Divide create extreme seasonal weather conditions to
which the alpine/subalpine ecosystem is delicately attuned.
High levels of atmospheric nitrogen are deposited on this
location from Colorado’s Front Range, and annual mean tem-
Katherine E. Williams is with Placeways LLC, 1790 30th
Street, Suite 314, Boulder, CO 80301, and formerly with the
Department of Geography, University of Denver, 2050 E. Iliff
Ave., Denver, CO 80208 (
).
Sharolyn J. Anderson is with the University of South Aus-
tralia, Mawson Lakes Campus, Mawson Lakes, SA 5095, and
formerly with the Department of Geography, University of
Denver, 2050 E. Iliff Ave., Denver, CO 80208.
Photogrammetric Engineering & Remote Sensing
Vol. 80, No. 4, April 2014, pp. 333–341.
0099-1112/14/8004–333
© 2014 American Society for Photogrammetry
and Remote Sensing
doi: 10.14358/PERS.80.4.333
PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING
April 2014
333
