On June 8, 2017, Ryan, Sandra, and I completed installation
of a weather station and four soil moisture monitoring platforms in the PUSH
neighborhood of Buffalo for the Vacant to Vibrant Project.
With all of the monitoring platforms now in
place for the project, we need to have quantitative methods with which to
analyze the data.
The goal of the
analysis methods is to measure the storage and processing capacity of the green
infrastructure installed in the experimental parcels of the three cities in the
project.
As documented in a previous
post (
here), experimental and control plots display patterns of variability in soil
moisture at various time scales.
Figure 1 illustrates the types of variability patterns
observed in soil moisture at four of the sites in Gary’s Aetna
neighborhood. Basically, there are
three types of variability in Figure 1.
Rainfall events have an abrupt increase in soil moisture followed by a
rapid decline (see selection A in Figure 1).
Between rainfall events there is a slower decline in soil moisture (see
selection B in Figure 1) with a pattern of regular short-term variability (see
selection C in Figure 1). This blog post
focuses on finding the drivers of short-term variability.
|
Figure 1. Variation
in soil moisture at 3 cm depth in four study sites in the Aetna neighborhood of
Gary. Parcels are as follows: 1200
Oklahoma (Gary E1), 1035 Oklahoma (Gary E2), 1252 Dakota (Gary E3), and 910
Idaho (Gary C1). The emphasized patterns
include: A—changes associated with
rainfall events, B—steady decline during rain free intervals, and C—short-term
variability in the decline time series between rain events.
|
The regular pattern of short-term variability appears to be
diurnal. Figure 2 shows an expanded view
of the decline during the May 28 to June 6 interval in Figure 1, selection
C. The diurnal patterns of soil moisture
variability and temperature are synchronous.
An obvious hypothesis is that these short-term changes in soil moisture
are driven by temperature changes.
|
Figure 2. Comparison
of patterns of change in soil moisture and temperature over the interval of May
26 to June 6 at 3 cm in the Gary E1 site.
|
If temperature changes are driving short-term variability of
soil moisture, there should be both theoretical and experimental evidence to
support the claim. From a theoretical
perspective, there is ample evidence indicating that soil moisture measurements
are temperature sensitive. The soil
moisture probes that we use (Onset Part No. S-SMx-M005) use time domain
reflectometry (TDR) method to estimate volumetric water content. It is
well known that temperature dependence of TDR includes both instrumental and
physical effects (Or and Wraith, 1999), and there are methods of correcting for
instrumental sensitivity to temperature change (e.g. Chanzy et al., 2012). Because the physical effects of temperature on
soil moisture reflects the underlying influence of temperature on the bound to
free ratio of soil moisture, soil composition and soil moisture content
interact with temperature effects to limit the generality of correction methods
(Seyfried and Grant, 2007).
Nevertheless, diurnal variability of soil moisture with temperature
changes is an expected pattern.
From an empirical perspective, diurnal variability of soil
moisture should reflect diurnal variation in soil temperature. Using the extrema identification method
discussed in the previous post, I estimate the soil moisture increments for the
entire time series for Gary E1 in Figure 1.
Figure 3 shows the resulting distribution
|
Figure 3.
Histogram of soil moisture (SM, m3/m3) for the
Gary E1 time series in Figure 1. The red
vertical line is the mean increment, 0.012, and the blue vertical lines are the
quartiles.
|
The distribution in Figure 3 is skewed with the mean near
the third quartile. Using the mean as a
demarcation of diurnal soil moisture increments from rainfall event increments,
Figure 4 shows a very strong linear dependence of soil moisture increment on
temperature change. Figure 5 provides
evidence that the subsequent decline rate also depends on temperature change.
|
Figure 4.
Association of soil moisture increment and temperature change for soil
moisture increments less than 0.012. In
blue, is the regression line (r=0.95, p<0.0001). The shaded area represents the standard error
bounds. |
|
Figure 5.
Association between temperature change and the instantaneous slope of
the decline in soil moisture for increments of soil moisture less than
0.012. In blue, is the regression line
(r=0.77, p<0.0001). The shaded area
represents the standard error bounds.
|
As noted by Or and Wraith (1999), separation temperature-dependent
variation of soil moisture from changes associated with evapotranspiration and
soil infiltration is necessary to understand the contributions of
evapotranspiration and infiltration in individual experimental and control parcels
of the V to V project. Although there
are proposed methods for removal of temperature-dependent transients, their
lack general reliability across types of soils and range of soil moisture
conditions is a limitation (Seyfried and Grant, 2007). Furthermore, as shown in Figure 6, the
diurnal pattern of variability varies with depth, reflecting both attenuation
and phase shift of temperature variation.
Even with temperature correction, diurnal fluctuation of soil moisture
associated with temperature dependent release of bound soil moisture will
remain in the time series. Given the
regularity of the diurnal variation of soil moisture, a smoothing technique may
offer a satisfactory method to estimate rates of decline during longer event
free intervals.
|
Figure 6.
Comparison of soil moisture variability in Gary E1 for three soil depths
(3 cm, bottom black line; 10 cm, middle red line; and 20 cm, top blue line).
|
References:
Chanzy, André, Gaudu, Jean-Claude, and Marloie,
Olivier. 2012. Correcting the Temperature Influence on Soil
Capacitance Sensors Using Diurnal Temperature and Water Content Cycles. Sensors, 12, 9773-9790.
Or, Dani, and Wraith, Jon M.
1999. Temperature effects on soil
bulk dielectric permittivity measured by time domain reflectometry: A physical
model. Water Res. Research, 35(2):
371-383.
Seyfried, Mark S. and
Grant, Laura E. 2007. Temperature Effects on Soil Dielectric
Properties Measured at 50 MHz. Vadose
Zone J. 6:759–765.