Estimates of Soil Moisture from the Oklahoma Mesonet
Oklahoma Climatological Survey
Norman, OK 73019
October 2002
Version 2.1
Table of Contents
������������������������������������� I.
������� Summary of important notes to
users
������������������������������������� II.
������ Definition of quantities provided
������������������������������������� III.
����� Ancillary data sets which may be
useful in analysis of data
������������������������������������� IV.
���� Background on the estimates of soil
moisture from the Oklahoma Mesonet
������������������������������������� V.
����� Processing applied to the data in
this delivery
������������������������������������� VI.����� Contacts for further information
������������������������������������� VII.
��� References
I.
Summary of important notes to users
This data delivery was
the product of a combined effort by the soil moisture research
team and
the Mesonet team at the Oklahoma Climatological Survey (OCS). The
Oklahoma Mesonet is operated
and maintained through a partnership of the University of
Oklahoma and Oklahoma
State University.
Additional processing
to compute the volumetric water content was provided by Alan Rodbock and
Lifeng Luo of the Department
of Environmental Sciences of Rutgers, the Stat University of New Jersey.
The research team
developed procedures to estimate soil water potential, soil water
content, and
fractional water index using raw measurements provided by the Mesonet
sensors.
The sensor used for
soil moisture measurements in the Oklahoma Mesonet is identical to
that used in the two
other Southern Great Plains Networks (the ARM/SWATS and the
ARS/SHAWMS network). The
quantity measured by the sensor is DeltaT (a change in
temperature over time
after a heat pulse is introduced). These DeltaT values are
normalized into a
value termed DeltaT reference. The
calibrated and quality assured
values of DeltaT reference are provided in this
dataset.
Data from the sensors
have been carefully calibrated to provide estimates of soil
water
potential as
a function of DeltaT reference. An empirical relationship to estimate soil
water potential from
DeltaT reference values is included in section V. C.
An empirical
relationship to estimate volumetric water content also
is provided by the
Oklahoma Mesonet
research team (section V. D.). However, users should understand
that volumetric water
is a �derived� quantity. Because
the coefficients used to convert
soil water potential to volumetric water content are
subject to change as the
knowledge of soil properties at Mesonet sites (soil
texture, bulk density, porosity,
etc.) increases, estimates of volumetric water content ARE
NOT provided
in this
dataset. However, the empirical relationship to
estimate volumetric water content
(section V. D.) are provided and the latest
coefficients can be obtained from the
Mesonet Manager (405-325-2541). As
new coefficients are developed, they will be
provided to the
research community.
NOTE:� These coefficients were used by Alan Robock and Lifeng Luo to compute
The volumetric water
content in this dataset.
Overall, the soil
water content values obtained using sensor data and the empirical
coefficients compare
well with validation data, which consists of soil water content data
derived from
gravimetric and neutron probe samples at a large number of sites at depths
of 5cm, 25cm, 60cm,
and 75cm. The sensor values tend to overestimate soil water
content slightly under
very dry soil conditions. This overestimate results from
inaccuracies in our
ability to convert soil water potential measurements to soil water
content values for
some sites.
The relationship
between soil water potential and soil water content for a particular soil is
referred to as a soil
water retention curve. This relationship is soil-specific and is highly
dependent on soil
properties such as pore size distribution, organic matter, and the
presence of
macropores. For some soils, this relationship sometimes exhibits hysteresis.
Soil water potential
values drier (i.e., more negative) than -6.0 bars should be considered
�suspect�. Although a
datum that falls below -6.0 bars is considered useful, soil
potential values in
that range have been unreproducable in the laboratory.
Because of the large
number of site and sensor depth combinations involved in the
Mesonet, it was not
economically feasible to perform laboratory measurements of the soil
water retention curve
for all sites and depths. The soil water retention relationships used
to compute soil water
content have been derived using a combination of published
techniques and
empirical corrections to these techniques. This approach utilized
information on the
size distribution of soil particles available for all sites and depths (12-
15 size categories for
each site/depth and location), measured bulk density at some sites,
and an empirical
correction derived from examination of laboratory-measured retention
curves for the
ARM/SWATS sites. Our validation data indicate that the curves derived
by this method are
reasonable for most sites and depths.
Many models require
both soil water potential and soil water content. (Soil
water
content is a key
quantity for monitoring mass balance; soil water potential is the quantity
which controls the
movement of water in the soil). Any model which numerically
simulates water flow
with a physically-based equation at some point must convert
between these two
quantities. Many models have �hard-wired� expressions imbedded in
them which convert
between potential/content based on texture-dependent coefficients.
Users with these types
of models should be aware of several key points:
o Because
it is a derived quantity, soil water content values
from the Mesonet�s
measurements should not
be ingested into a model and then converted to
soil
water potential using relatively
crude expressions based only on soil texture.
o Application
of a relatively crude (e.g., based only on texture) expression for
converting to soil
water content will likely yield estimates of soil water content
with greater
uncertainty than those we have provided.
o If
at all possible, models should ingest the Mesonet�s estimates of both soil
water
potential and soil
water content.
This data delivery
includes data from 100 sites across the state of Oklahoma.
Approximately 60 sites
comprised the initial set of stations at which soil moisture sensors
were installed in
1996/97. As part of a project known as OASIS, soil moisture sensors
were installed at
approximately 40 more sites at various times during 1999.
All soil moisture data
prior to 1 October 2002 are considered �experimental� and are not
available unless
special permission is obtained from the Director of the Oklahoma
Climatological Survey.
Users who plan to
compare Mesonet point-based measurements with areally integrated
values from models or
remote sensing may find some of the ancillary data sets useful
(described below).
These data sets include tables which describe the distribution of soil
and vegetation types
in variable-sized footprints surrounding the Mesonet sites.
Recent research focusing
on the use of the Mesonet soil moisture sensors as soil
temperature sensors
demonstrated that:
o The
Mesonet soil moisture "start temperature" variable is not an accurate
measure
of soil temperature.
o The Mesonet cannot at this time make
a statement regarding the relationship
between �start temperature� and soil temperature.
II.
Definition of quantities provided
The variables provided in this data delivery
include quantities measured at 4 depths (5, 25, 60,
and 75 cm below the surface). For each depth,
one variable (TR) is reported. This value can be
used with algorithms to produce other variables
(MP, WC, and FWI).
The variable provided for each depth is:
TR: The DeltaT reference temperature. This is
the value of DeltaT (which is the quantity
directly observed by these sensors), but
normalized for individual sensor response. After
normalization, the quantity is referred to as
the DeltaT reference temperature.
Estimates of soil moisture that can be derived
from TR include:
MP: Matric, or soil
water potential (bars)
WC: Volumetric soil
water content (m3 water/m3 soil)
FWI: Fractional water
index (unitless)
Times shown are UTC. The naming convention for
the variables is such that the first two letters
refer to the quantity (TR) and the last two
numbers refer to the depth of the measurement for that
quantity.
The ASCII data provided are:
TR05, TR25, TR60, TR75
III.
Ancillary data tables which may be useful in analysis of data
Two sets of ancillary data may be of interest to
users:
1. A table of spatial
coordinates for Mesonet sites (GEOMESO.TBL).
2. Tables that
summarize the soil and vegetation characteristics surrounding the Mesonet
sites.
These tables are intended to be helpful to users
who desire to compare point-based Mesonet
measurements to areally integrated estimates
generated with models or remote sensing data.
They provide information related to the
distribution of different soils and vegetation types
surrounding the Mesonet stations. The tables are
available at 4km X 4km and at 10km X 10km
footprints centered on Mesonet sites. These
support data are available for all Mesonet and all
ARM/SWAT sites within Oklahoma. The tables
related to soils information exist for all 4 depths
of the soil moisture measurements. The input data
used to generate these tables are the
NRCS/MIADS 200 m resolution raster data set for
soils and land use/land cover. The tables
will be updated as better data sources become
available (for example, when NRCS/SURGO data
become available for more Oklahoma counties). To
obtain these tables, please contact Karen
Humes (khumes@uidaho.edu; 208-885-6506).
IV.
Background on the estimates of soil moisture from the Oklahoma Mesonet
The Oklahoma Mesonetwork (Mesonet) began in 1991
as a state-wide mesoscale environmental
monitoring network. (Brock et al. 1995). In
1996, with NSF EPSCoR support, soil moisture
sensors were added to 60 of the approximately
115 Mesonet sites. At these 60 sites, sensors
were installed at 3 depths: 5cm, 25cm, and 60cm
below the surface. For most sites, a fourth
sensor was installed at 75 cm below the surface.
The soil profile at a few sites would not
accommodate the deepest sensor. The data are
sampled every 30 minutes and transmitted to a
central ingest computer at the Oklahoma
Climatological Survey (OCS) at the University of
Oklahoma. In partnership with Oklahoma State
University, OCS is the Oklahoma agency which
maintains and operates the Mesonet.
The soil moisture sensor operates on the
principle of heat dissipation. It is the same sensor
deployed in two other networks in the Southern
Great Plains region which include soil moisture
measurements: a) the ARM/SWATS network
(Schneider et al. 2002) and b) the USDA/ARS
SHAWMS network (Starks 1999) (located primarily
in the Little Washita Experimental
Watershed). The quantity measured by these
sensors is the temperature change over time within
a needle enclosed in a porous ceramic in contact
with the soil. These measurements of
temperature change (DeltaT) have been carefully
calibrated to soil water potential,
which is
typically expressed in units of pressure (or
force per unit area); potential represents the force
with which the water is �held� in the soil. This is an important quantity in
soil physics because
spatial
gradients in soil water potential provide the �driving force� for soil water
movement.
One of the advantages of a sensor which is
sensitive to soil water potential is that the calibration
equation used to convert the directly observed
quantity to soil water potential is typically not
dependent on site-specific factors such as soil
texture.
Though soil water potential is the variable
which controls the movement of water in the
subsurface soil layers and from the soil to
plants, many models that utilize mass balance as an
important constraint require volumetric
soil water content for a measure of the soil moisture.
The relationship between soil water potential
and water content (usually referred to as the �soil
water characteristic curve� or �retention
curve�) is highly dependent on many soil
characteristics, primarily pore size
distribution, macropore space and organic matter content.
Additionally, this relationship sometimes
exhibits hysteresis. Soil water retention curves can be
measured in the laboratory with a pressure plate
apparatus by desorbing an initially saturated
sample to a known soil water pressure, then
measuring the equilibrium soil water content, and
fitting curves to the measured points. These
pressure plate tests should be performed on intact,
"undisturbed" samples, which are
carefully extracted in the field.
Because of the large number of sites and depths
involved in the acquisition of Mesonet soil
moisture data, it was not feasible to perform
laboratory measurements of the soil water retention
curve for all sites and depths. The soil water
retention relationships used to compute soil water
content have been derived using a combination of
published techniques and empirical corrections
to these techniques. This approach utilized
information on soil particle size distribution available
for all sites and depths (12-15 size categories
for each site/depth location), measured bulk density
at some sites, and an empirical correction
derived from examination of laboratory-measured
retention curves for the ARM/SWATS sites. More
details on this approach are available in
Humes et al. (2002).
Soil water content validation data were acquired
at a large number of sites via both gravimetric
sampling of the soil profile and neutron probe
measurements. Details of these results are
provided in Humes et al. (2002). The validation
results indicate that the estimates of soil water
content derived by these techniques are
reasonable at most sites. There is some indication that
the sensor values tend to overestimate soil
water content slightly under very dry soil conditions.
We feel this is due to inaccuracies in the
Mesonet�s ability to convert soil water potential
measurements to soil water content values for
some sites.
Using NSF support obtained via an MRI (Major
Research Infrastructure) grant for a project
known as OASIS, soil moisture sensors were
installed at approximately 40 additional sites
during 1999.
All soil moisture data prior to 1 October 2002
are considered �experimental� and are not
available unless special permission is obtained
from the Director of the Oklahoma
Climatological Survey.
V.
Processing applied to the data in this delivery
A)
QA procedures
The Oklahoma Mesonet uses the Campbell
Scientific Model 229L sensor to provide indirect
measurements of the soil matric potential (MP)
volumetric soil water content (WC), and
fractional water indes (FWI). First the sensor
temperature (STxx) for each depth (xx) is
measured. Then the sensor is heated for a short
period, after which the sensor temperature is
measured again (FTxx). The difference in the two
temperatures (Tdiff) is used to derive the
reference delta temperature (TRxx). Because of
the way the sensors are wired to the datalogger,
a temperature bias exists in STxx and FTxx. A
reference thermistor is used to remove this bias.
The measurement returned from the reference
thermistor (TREF) is used only for QA purposes.
The Oklahoma Mesonet performs several routine QA
procedures on the variables STxx, FTxx,
TRxx, and TREF listed above. The procedures used
consist of range, freeze, and step tests. The
range test is used to ensure that ST, FT, TR,
and TREF do not go beyond normal pre-determined
limits. The freeze test is used to identify
sensors in possible frozen soils.
The step test checks for spikes in TRxx values
(large change in consecutive readings). The
change in TRxx values is calculated as the TRxx
value for the current time period minus the
TRxx value from the preceding time period. Only
the endpoints of the run of suspicious data are
flagged. Visual inspection is necessary to determine the
duration of the event.
The very wet limit of sensitivity for the sensor
has been determined to �0.10 bars. Values of soil
water potential (MP) between 0.00 and �0.10 bars
are not accurate. Because the lower limit of
observed values of DeltaT reference is
approximately 1.4 �C, the equation for computing soil
water potential does not return values much more
moist than �0.10 bars.
Soil water potential values drier (i.e., more
negative) than �6.0 bars should be considered
�suspect�. Although data values are drier than
�6.0 bars are still considered useful, soil potential
values in that range have been unreproducable in
the laboratory.
Another possible source of sensor error has been
identified in Basara and Crawford (2000). This
error is identified by the rapid wetting of the
60 cm and 75 cm sensors after a precipitation event,
due to preferential flow through the (refilled)
trench dug to install the sensors. A potential
problem occurred with the installation of a
subsurface sensor. It typically takes time for a
refilled trench to �heal� (the term applied to
the development of good contact between the
disturbed soil in the trench and the undisturbed
soil in which the sensor is placed). In some soils,
it is nearly impossible to avoid the development
of permanent macropores near the interface
between the disturbed and undisturbed soils.
Soils whose properties are conducive to the
development of large macropores upon drying
(such as those with shrink/swell clays) are most
susceptible to preferential flow problems.
The most common condition when this type of
problem occurs is when a heavy precipitation
event follows an extended dry period (i.e., the
soil profile is dry throughout). The data pattern
characteristic of this problem is an immediate
wetting of the deeper (60 cm and 75 cm) sensors
after a heavy rain event, followed by a rapid
drying of these sensors, which returns the sensor to
its pre-wetted condition. When this problem occurs,
the shape of the time series of the data is an
upside-down spike (an immediate decrease
followed by a rapid increase in TR). The �normal�
response for a recently wetted-soil is a
decrease in TR, followed by a few days of level,
unchanging values, then a slow increase in TR as
the soil begins to dry.
It is difficult to codify this behavior in an
automated QA routine; thus, the data in this delivery
were not �filtered� to remove this preferential
flow problem.
The errant deep-layer values of soil moisture
have been detected following extended dry periods
in soils consisting of high silt and clay
fractions. However, in most cases and at most sites, these
measurement errors do not occur.
It should be noted that this problem is an
anomaly in the standard operation of soil moisture
sensors across Oklahoma. Of over three million
observations of soil moisture conditions
between 1996 and 1999, the number of
observations affected by this installation error account
for less than one percent.
B)
Conversion of DeltaT to DeltaT reference (TR)
This conversion requires sensor-specific
coefficients which are determined based on the
maximum and minimum observed DeltaT values for
each individual sensor. An initial value for
the maximum DeltaT is determined in the lab by
placing the sensor in a sealed container and
drying the sensor completely using desiccant
packs. An initial value for the minimum DeltaT is
determined in the lab by submerging the sensor
in water for several days.
After being deployed in the field for several
years, a sensor typically will become drier and/or
wetter than it ever had in the lab. When a
larger maximum DeltaT value or smaller minimum
DeltaT value is achieved in the field, the
sensor�s coefficients are updated. Therefore, the sensor-specific
coefficients change several times during a
sensor�s lifetime. To obtain the latest
coefficients, contact the Mesonet Manager at
(405) 325-2541.
A linear regression is used to
"normalize" the response of an individual sensor to that of an
idealized reference sensor having the following
maximum and minimum DeltaT (dT) values:
dTmaxref = 3.96 C
dTminref = 1.38 C.
The linear regression used to normalize an
individual sensor's response is of the form:
TR = m * dT + b
where TR is referred to as the DeltaT reference
temperature (normalized for individual sensor
response).
The regression coefficients m and b are
determined by substituting the maximum and minimum
dT values for an individual sensor and for the
reference sensor into the regression equation, and
solving for m and b. This results in the
equations:
m = ( dTmaxref �
dTminref ) / ( dTmaxsensor � dTminsensor )
���� or
m = ( 3.96 � 1.38 ) /
(dTmaxsensor � dTminsensor )
���� and
b = 3.96 � m * dTmaxsensor.
Each sensor, therefore, has its own unique
coefficients for normalizing its response.
Determining coefficients for each sensor results
in a table of sensor-specific coefficients used to
compute the DeltaT reference (TRxx) values
provided in this data set.
C)
Conversion of DeltaT Reference (TR )Values to Soil Water Potential (MP)
The following generalized equation can be
applied to all DeltaT reference values (TR):
MP = - (c * exp ( a *
TR))/100
where
MP = matric potential
(bars)
TR = DeltaT reference
( C)
a = 1.788
c = 0.717
The limit of sensitivity for the sensor has been
determined as �0.10 bars. Values of soil water
potential (MP) between 0.00 and �0.10 bars are
not accurate. Because the lower limit of
observed values of DeltaT reference is
approximately 1.4 deg C, the equation for computing MP
does not return values much more moist than
�0.10 bars.
D)
Conversion of Soil Water Potentia (MP)l Values to Soil Water Content
WC = WC r + ( WC s -
WC r ) / ( 1 + ( a * -MP) ^ n ) ^ ( 1 � 1 / n )
where
WC = soil water
content on a volume basis (m3 water/m3 soil).
WC r = residual water
content (m3 water/m3 soil).
WC s = saturated water
content (m3 water/m3 soil).
a, n = empirical
constants
MP = matric
(soil-water) potential (bars).
The four coefficients, WC r , WC s , a, and n,
for each depth for each site are available in a table of
coefficients. Because the coefficients used to convert soil water
potential to volumetric
water
content are subject to change as the knowledge of soil properties at Mesonet
sites
(soil
texture, bulk density, porosity, etc.) increases, estimates of volumetric water
content
ARE NOT provided in this dataset. However, the empirical
relationship to estimate
volumetric
water content (section V. D.) are provided and the latest coefficients can be
obtained
from the Mesonet Manager. Users may obtain an electronic version of
these
coefficients by contacting Chris Fiebrich,
Manager of the Oklahoma Mesonet
(chris@mesonet.org; 405-325-6877).
E)
Conversion of DeltaT Reference (TR) Values to Fractional Water Index (FWI)
FWI = ΔTd � TR / ΔTd
� ΔTw
where
FWI = fractional water
index (unitless)
TR = DeltaT reference
(C)
ΔTd = 3.96 C
ΔTw = 1.38 C
VI.
Contacts for further information
This data delivery is the product of a combined
effort by the soil moisture research team and the
Mesonet team at the Oklahoma Climatological
Survey (OCS).
Additional processing to compute the volumetric
water content was provided by Alan Rodbock and
Lifeng Luo of the Department of Environmental
Sciences of Rutgers, the Stat University of New Jersey.
The research team consisted of Jeff Basara
(University of Oklahoma/OCS), Ron Elliott
(Oklahoma State University), Ken Fisher (private
consultant), Gary McManus (OCS), and Karen
Humes (University of Idaho). The data processing
team consisted of Jessica Thomale (OCS),
David Demko (OCS), Gary McManus (OCS), Jeff Basara
(OCS), Chris Fiebrich (OCS), Janet
Martinez (OCS), Brad Illston (OCS), and Mike
Wolfinbarger (OCS). For questions regarding
the processing of the data, particularly the
conversion of raw variables to soil water potential or
soil water content, please contact Jeff Basara
(jbasara@ou.edu; 405-325-1760) or other members
of the research team.
For any question about the QA procedures applied
to the operational data processing stream,
contact Janet Martinez (janet@mesonet.org;
405-325-1834). For questions relating to the format
and content of the variables delivered, please
contact David Demko of the Oklahoma
Climatological Survey
(ddemko@operations.ocs.ou.edu; 405-325-3231).
VII.
References:
Basara, J.B., and T.M. Crawford, 2000. Improved
installation procedures for deep layer soil
moisture measurements. Journal of Atmospheric
and Oceanic Technology, 17, 879-884.
Brock, F.V., K.C. Crawford, R.L. Elliott, G.W.
Cuperus, S.J. Stadler, H.L. Johnson, and M.D.
Eilts, 1995. The Oklahoma Mesonet: A technical
overview. Journal of Atmospheric and
Oceanic Technology, 12, 5-19.
Humes, K.S., J.B. Basara, R.L. Elliott, D.K.
Fisher, and K.C. Crawford, 2002. Validation of soil
moisture observations from the Oklahoma Mesonet.
To be submitted to the Journal of
Hydrometeorology.
Schneider, J.M., D.K. Fisher, R.L. Elliott, G.O.
Brown, and C.P. Bahrmann, 2002.
Spatiotemporal Variations in Soil Water: First
Results from the ARM SGP CART Network,
Submitted to Journal of Hydrometeorology. In
Review.
Starks, P. J., 1999: A general heat dissipation
sensor calibration equation and estimation of soil
water content. Soil Sci., 164,
655-661.