Professional Documents
Culture Documents
Prediction of Active Zone Depth in Oklahoma Using Soil Matric Suction
Prediction of Active Zone Depth in Oklahoma Using Soil Matric Suction
Prediction of Active Zone Depth in Oklahoma Using Soil Matric Suction
ABSTRACT
Soil matric suction is an important parameter in unsaturated soils, and it controls the shrink-swell properties of expansive soils.
Matric suction variations cause volume change in expansive soils, which further causes damages to pavements and foundations.
This study made use of field measurements of matric suction obtained from the Oklahoma Mesonet to estimate active zone depth in
Oklahoma. The long-term matric suction measurement indicated a seasonal pattern of matric suction variation. The matric suction
at shallower depth of soil is more sensitive to the climatic conditions. An empirical equation was used to calculate the active zone
depth. Most parts of Oklahoma have an active zone depth ranging from 1 to 3 m, depending on the soil properties and climatic
conditions. The diffusion coefficient is an important parameter controlling the active zone depth. Field measurements of matric
suction from the Oklahoma Mesonet and the Department of Energy’s Atmospheric Radiation Measurement (ARM) network were
used to verify the active zone depth obtained from this study.
evaluation of field measurement of soil matric suction and Oklahoma Climatological Survey (OCS) at the University of
diffusion coefficient; and (3) evaluating how the climatic Oklahoma remotely collects the temperature change data every
conditions can affect the active zone depth. To calculate the 30 minutes, and the Oklahoma Mesonet provides daily averaged
active zone depth, an empirical equation proposed by McKeen normalized temperature change, Tref, which can be used to
and Johnson (1990) is used in this study. In this methodology, the calculate soil matric suction (Illston et al. 2008). The equation
active zone depth is a function of seasonal soil matric suction used by the Oklahoma Mesonet to determine soil matric suction
variation, climate frequency, and field diffusion coefficient. To is
determine the spatial distribution of the active zone depth,
ArcGIS software was used to create choropleth maps. The
choropleth map offers an easy way to display how a u c exp (a Tref ) (1)
measurement varies across a geographic area, or it shows the
level of variability within a region.
where u is soil matric suction (kPa); a and c are calibration
2. DATA AND METHODOLOGY constants, which are set as 1.788C1 and 0.717 kPa, respectively,
and they were calibrated by a series of laboratory tests conducted
2.1 The Soil Moisture Network in Oklahoma by the Mesonet researchers (Reece 1996); Tref is measured
temperature difference (C). There were 74 Mesonet stations
The State of Oklahoma is a region with complex climatic
selected for this study. The selection was made to use one station
conditions and geological features, consisting of mountains,
in each of the 77 counties in Oklahoma. However, three counties
streams, and lakes in different parts of the State (Johnson 2008).
do not have the sensors for suction measurement, since the
The climate of Oklahoma ranges from humid subtropical in the
Mesonet gradually installed the sensors at some stations during
east to semi-arid in the west (The Oklahoma Climatological
the 1990s due to the sensor reliability and affordability (Basara
Survey 2011). The terrain of Oklahoma varies from nearly flat in
and Crawford 2000). Any Mesonet sites that do not have soil
the west to rolling in the east, and it has a general slope upward
moisture sensors are due to issues beyond control such as
from east to west across the State (Arndt 2003). Based on the soil
bedrock at that depth or nearby irrigation causing invalid
survey data from the Natural Resources Conservation Service
readings (Illston et al. 2008). As a result, 74 stations were
(NRCS), western Oklahoma contains brown to light-brown
selected in 74 counties. Figure 1 shows the geologic features in
loamy soils with clay; central Oklahoma has the dark loamy soils
Oklahoma (Heran et al. 2003) and the spatial distribution of
with clayey to loamy subsoils; and soils in eastern Oklahoma are
selected Mesonet stations.
often brown to light-brown silty soils with clayey subsoils
(Carter and Gregory 2008). More than 75% of Oklahoma
bedrock units are potential sources for expansive soils with 2.2 Previous Field Investigations into Active Zone
shrink-swell characteristics (Luza and Johnson 2005). The
Depth
volume change of expansive soils typically happens in regions
with dry periods followed by wet periods. The precipitation in Several field investigations were conducted in Oklahoma to
Oklahoma varies by season. Winter is the season with the lowest
evaluate soil suction and active zone depth. Nevels (1995)
precipitation and spring is the season with the highest measured in situ total and matric suction of a soil sample
precipitation (Johnson 2008). Due to Oklahoma’s soil and obtained in Sulphur, Oklahoma (the city is located in central
climatic characteristics, the problem caused by expansive soils is Oklahoma). The suction was measured by the filter paper method
one of the main geologic hazards in Oklahoma. As a result, it is on thin-walled tube samples taken in February 1994. By
critical to characterize Oklahoma’s soil moisture and suction
evaluating the initial and final suction profiles versus depth,
properties and how they change in response to climatic condi- Nevels (1995) indicated an estimated active zone depth at 3.17 m.
tions.
Collins and Miller (2013) conducted cone penetration test (CPT)
The State of Oklahoma has a world-class network of and standard penetration test (SPT) at two sites near Norman,
environmental monitoring stations the Oklahoma Mesonet, Oklahoma (the city is located in central Oklahoma) to predict
which consists of 120 automated stations collecting a variety of changes in soil moisture. Both CPT and SPT results showed
real-time hydrometeorological and soil parameters. Due to the noticeable moisture variations to a depth of roughly 1.5 m at each
need for soil moisture measurement in Oklahoma, the Oklahoma site with little change below this depth, so a 1.5 m depth of the
Mesonet installed the Campbell Scientific 229-L heat dissipation active zone was predicted for both sites. In addition to research
matric potential sensor at four depths (5 cm, 25 cm, 60 cm, and and field investigations conducted in Oklahoma, some
75 cm) at some stations to measure soil moisture conditions since neighboring states have done some similar investigations.
1996 (Illston et al. 2008). The Campbell Scientific 229-L heat Typical values of active zone depths in Texas range from 1 m to
dissipation matric potential sensor consists of a thermocouple and 5 m, although values up to 10 m have also been reported (Wray
a resistance heater housed in a hypodermic needle, and the 1987; McKeen and Johnson 1990). Durkee (2000) conducted the
hypodermic needle is embedded in a ceramic matrix (Campbell field and laboratory tests as well as numerical modeling to
Scientific, Inc 2009). This sensor records a temperature change evaluate the active zone depth for expansive soil test site located
after a 50-mA electric current is sent through the resistance at Colorado State University in Fort Collins, Colorado. His
heater. The temperature change after the electric current is research indicated that the active zone depth was approximately
directly related to the matric suction of soil near the sensor equal to the depth of soil moisture fluctuation, which ranged
(Basara and Crawford 2000). A central ingest system at the from 1.5 m to 2.1 m.
Yue, Veenstra: Prediction of Active Zone Depth in Oklahoma Using Soil Matric Suction 31
Fig. 1 Oklahoma geologic map and selected Mesonet stations with station ID
2.3 Determination of Active Zone Depth Based on Mitchell’s (1979) approach, McKeen and Johnson
(1990) proposed an equation to calculate active zone depth using
Mitchell (1979) analyzed seasonal soil suction variation in diffusion coefficient and matric suction data:
expansive soils. He pointed out that the climate, drainage, and
vegetation type controlled the moisture flow at the soil surface. 2U 0
Mitchell’s research indicated that the suction change is a periodic ln
ΔU max
function of time. The suction change at the soil surface is given Z (4)
by: nπ
α
u(0, t ) Ue U0 cos (2nt ) (2)
where Z is active zone depth (m); U0 is the amplitude of suction
where u(0, t) is the suction at the soil surface (pF or kPa); Ue is variation (pF or kPa); Umax is maximum suction change, which
the equilibrium suction (pF or kPa); U0 is the amplitude of suc- is the difference between the maximum and minimum extremes
tion variation (pF or kPa), which is equal to the difference be- or envelopes of the suction profile for a given depth; n is climate
tween the equilibrium matric suction and the minimum suction frequency number, which is the number of cycles of wetting and
during wet periods, or the difference between the equilibrium drying in a year (cycles/year). Umax is the maximum suction
matric suction and the maximum suction during dry periods; n is change below which the soil volume change is considered
climate frequency number, which is the number of cycles of wet- insignificant. Umax is usually assumed based on long-term
ting and drying in a year. Based on Eq. (2), Mitchell developed moisture conditions and previous studies. Richards and Chan
the suction u(y, t) at any depth y: (1971) analyzed moisture changes that occurred in pavements
and subgrades of sealed roads. They found that there is no soil
n 0.5 volume variation in pavements when the suction change was
n 0.5
u (y, t ) U e U 0 exp y cos 2nt y smaller than 0.1 pF. In addition, El-Garhy and Wray (2004) used
a computer program called SUCH (SUCtion-Heave) to evaluate a
relationship among the edge moisture variation distance, the
(3)
amplitude of suction change, the diffusion coefficient of the soil,
where u(y, t) is the suction at depth y (m) and time t (days) (pF or and the active zone depth. They also set Umax as 0.1 pF to
kPa); is diffusion coefficient (cm2/s). Equation (3) indicates determine the distribution of soil suction in the expansive soil
that the suction at any depth depends on the diffusion coefficient. mass under a covered area. As a result, the maximum suction
The diffusion coefficient is not constant in soil. As a result, the change was set as 0.1 pF in this study.
suction profile can be determined by the diffusion coefficient. To calculate the active zone depth, a diffusion coefficient
Bai et al. (2007) tested the water diffusion coefficients of must be determined first. The soil diffusion approach is relatively
horizontal soil columns obtained in a wetland in China. Their new in geotechnical practice, so only a little research on soil
tests revealed that the water diffusion coefficients of wetland diffusivity has been conducted in clay material (Chemkhi and
soils had a significant positive correlation with the soil water Zagrouba 2005). Since the diffusion coefficient is one of the
volumetric contents, and they increased exponentially with critical factors that influence the matric suction variation, a good
increases in the soil water volumetric contents. engineering judgement on the diffusion coefficient is necessary
32 Journal of GeoEngineering, Vol. 13, No. 1, March 2018
y 1
t (5)
2 n
where t is time lag in matric suction between two depths; y is Fig. 2 Statewide annual total precipitation
depth measured from the soil surface. This relationship had been
confirmed by the field measurements of the matric suction profile
depth than at or near the soil surface, usually from a few tens of
(Mitchell 1979). If the matric suction at the soil surface and
centimeters down to five meters, since the deeper depths are less
certain depths is measured, and if the time lag in matric suction
affected by the surface conditions and the more stable
between two depths is measured, the diffusion coefficient can be
performance of suction (Wray 1987; Nichol et al. 2003; Nguyen
determined from Eq. (6):
et al. 2010).
y2 A curve fitting method was used to analyze the sinusoidal
(6) pattern of matric suction variations. Curve fitting constructs a
t 4n
2
curve and/or a mathematical function to best fit a series of data
where is diffusion coefficient; y is depth measured from the points. Origin software (OriginLab Corporation, Northampton,
soil surface; t is time lag in matric suction between two depths; n MA) was used to complete the curve fitting process. Origin is a
is climate frequency number. Mitchell (1979) used this equation scientific graphing and data analysis software. It has a nonlinear
to determine the diffusion coefficient for an expansive soil in an curve fitting tool that includes more than 200 built-in fitting
open field in Adelaide, Australia. A time lag of three months was functions. The function for sinusoid fitting is
observed from the soil surface to 1.2 m depth during a 12 months
x xc
period. Mitchell estimated = 5.9 104 cm2/s by using Eq. (6). y y0 A sin
(7)
Aubeny and Long (2007) compared different methods of field
estimates of diffusion coefficient including Eq. (6). They
where y0 is offset; A is amplitude; xc is phase shift; is period.
indicated that the diffusion coefficient estimated by Eq. (6)
By comparing Eqs. (2) and (7), y0 is the equilibrium suction (Ue);
highly depends on the matric suction waveform and the surface
A is the amplitude of suction variation (U0); 365/2 is the
suction variation, which can affect the diffusion coefficient up to
climate frequency number (n), which is the number of wetting
nearly two orders of magnitude greater than the laboratory range.
and drying cycles per year.
2.4 Determination of Seasonal Variation of Matric
3. RESULTS AND DISCUSSION
Suction
3.1 Seasonal Variation of Matric Suction
It is difficult to use one single sinusoidal equation to
represent long-term variations of matric suction. Some research Of all 74 stations selected in this study, 54 stations have
simplified the dataset by separating the long-term data into short- sensors installed at 60 cm depth. The matric suction data from
term data and paying attention to specific short-term data. Visser these 54 stations were input individually into Origin software for
et al. (2006) used 20 years (1980 ~ 2000) of daily phreatic sur- curve fitting. After the fitting process, by comparing the fitting
face depth and daily meteorological data to forecast water table curve and actual measured points for each station, stations with
depth and soil moisture profiles. In their research, to test model an R-square larger than 0.5 were selected for further calculation
performance, they selected three sets of 5-year periods, which of the active zone depth. There is no general rule of what values
can be considered as representing relative dry, average, and of R-square are high, adequate, or low. In some pipeline
relative wet climatic conditions. Similar to Visser et al.’s (2006) constructions in geotechnical engineering, an R-square of
method, based on the duration and availability of matric suction pipeline repair rate and ground velocity larger than 0.7 was
data, this study selected three 2-year periods to represent considered as acceptable (O’Rourke and Bonneau 2007). Figure
relatively dry, average, and relatively wet conditions from 2000 3 shows typical two-year measured matric suction at 5 cm, 25 cm,
to 2014. Figure 2 shows the statewide annual total precipitation and 60 cm depths in Central Oklahoma. When doing the curve
from 2000 to 2014. According to the Oklahoma Climatological fitting, the unit of matric suction was converted from kPa to pF,
Survey, the statewide average annual precipitation from 1895 to since the matric suction in pF is the logarithmic form. A
2016 was 85.7 cm. Three sets of 2-year periods are 2001-2002 logarithmic form is more useful when the rate of change in the
(average), 2007-2008 (relatively wet), and 2011-2012 (relatively data increases or decreases quickly.
dry). The plots of matric suction values of most of Mesonet
Soil matric suction measured at 60 cm was analyzed in this stations indicate seasonal trends during a given period. Matric
study since the Oklahoma Mesonet decommissioned sensors at suction is low from November to March and starts to increase
75 cm depth in January 2011. However, it should be noted that from April to July. Matric suction reaches the maximum in
geotechnical engineers are more interested in suction at a deeper August and September and starts to decline in October.
Yue, Veenstra: Prediction of Active Zone Depth in Oklahoma Using Soil Matric Suction 33
(a) 5 cm depth, R-square for fitting model: 0.58 (b) 25 cm depth, R-square for fitting model: 0.69
3.2 Determination of Diffusion Coefficient surface due to the less effect of near-surface climatic conditions.
The time lag is the time of maximum matric suction at 60 cm
Since the diffusion coefficient is a function of water content, depth minus the time of maximum matric suction at 25 cm depth.
it will vary when water content varies. A wide range of diffusion Figure 4 shows a set of matric suction measured at 25 cm and
coefficients was reported from previous literature. Aubeny and 60 cm depths at two stations. Since a two-year period of suction
Long (2007) tested diffusion coefficient of soil samples obtained was evaluated, the time lag was the average of the time
in several Texas sites (Fort Worth, Austin, and Waco) in the difference at the maximum suction between 25 cm and 60 cm
laboratory. The tests yielded a range of diffusion coefficients depths during these two years. From Fig. 4, for BUTL station,
from 2.7 cm2/day to 8.2 cm2/day. However, by estimating there was a 20-day time lag at the maximum suction between the
diffusion coefficient using Eq. (3), McKeen and Johnson (1990) two depths in June 2011 and 34-day time lag from April to June
reported the diffusion coefficient for Dallas/Fort Worth as in 2012. As a result, the time lag for the BUTL station was
151 cm2/day, which indicated that the field values of diffusion 27 days during 2011-2012. For the LANE station, there was a
coefficient could be much greater than the laboratory values to three-day time lag in September 2007 and nine-day time lag in
nearly two orders of magnitude. August 2008, so the time lag for the LANE station was six days
This study used Eq. (6) to determine the diffusion during 2007-2008. Based on the analysis of all selected Mesonet
coefficient. Time lags were estimated by the field measurements stations, the time lag between 25 cm and 60 cm depths ranged
of matric suction at 25 cm and 60 cm depths, since matric suction from one day to more than 60 days.
at 25 cm depth had a better sinusoidal pattern than that at the soil
Figure 5 provides the spatial distribution of the diffusion tation is average, evapotranspiration is still high due to vegetation
coefficient under different moisture conditions in Oklahoma. coverage and increased surface temperatures during summer. As
Generally, no matter under dry, average, or wet conditions, a result, the forces between removing and replenishing soil
eastern Oklahoma has higher diffusion coefficient than western moisture are muted. Under the dry condition, the moisture
Oklahoma, which means the higher the water content, the higher replenishment to the soil is limited and the soil moisture is
the diffusion coefficient. However, since only stations with an R- dominated by the water removal force, so the annual sinusoidal
square larger than 0.5 were selected for diffusion coefficient and patterns of soil moisture reveal an obvious phase shift (Illston et
active zone depth determination, the spatial distribution of al. 2004). The diffusion coefficient ranged between 100 cm 2/day
selected Mesonet stations were different in three moisture to 500 cm2/day for most parts of Oklahoma during dry and wet
conditions, which may lead some uncertainties in the magnitude periods. However, the diffusion coefficient was smaller than
of the diffusion coefficient. The numbers of selected stations (R- 150 cm2/day for the whole State of Oklahoma during the average
square 0.5 from curve fitting) under dry, average, and wet period.
conditions are 38, 27, and 18, respectively. The number of
selected stations under the wet condition is much less than that 3.3 Determination of Active Zone Depth
under dry and average conditions because the sinusoidal pattern
observed under the wet condition is generated by the forces As presented in Fig. 6, the spatial distribution of the active
between removing and replenishing soil moisture. Evapotranspi- zone depth is similar to the spatial distribution of diffusion
ration and growing vegetation remove soil moisture and frequent coefficient. The active zone depth is deeper in eastern Oklahoma,
precipitation replenishes soil moisture, which make soil matric and shallower in western Oklahoma under all three moisture
suction varies frequently that cannot fit a sinusoidal pattern well. conditions, and most parts of Oklahoma had an active zone depth
Under the average condition, even though the statewide precipi- ranging from 1 to 3 m.
Bobrowsky, P.T. (2013). Encyclopedia of Natural Hazards. Springer. Nelson, J.D. and Miller, D.J. (1992). Expansive Soils: Problems and
Bond, D. (2005). Soil Water and Temperature System (SWATS) Practice in Foundation and Pavement Engineering, John Wiley
Handbook. The U.S. Department of Energy, Office of Science, & Sons.
and Office of Biological and Environmental Research. Nelson, J.D., Overton, D.D., and Durkee, D.B. (2001). “Depth of
Campbell Scientific, Inc. (2009). 229 Heat Dissipation Matric Water wetting and the active zone.” Proceedings of the Expansive Clay
Potential Sensor. Instruction Manual. Soils and Vegetative Influence on Shallow Foundations, ASCE,
Carter, B.J. and Gregory, M.S. (2008). Soil Map of Oklahoma. Earth Houston, 95109, doi.org/10.1061/40592(270)6.
Sciences and Mineral Resources of Oklahoma. Educational Nelson, J.D., Chao, K.C., Overton, D.D., and Nelson, E.J. (2014).
Publication 9, Oklahoma Geological Survey. Foundation Engineering for Expansive Soils. John Wiley &
Chemkhi, S. and Zagrouba, F. (2005). “Water diffusion coefficient in Sons.
clay material from drying data.” Desalination, 185(1-3), Nevels, J.B. (1995). “The use of soil suction in analysis of pavement
491498, doi.org/10.1016/j.desal.2005.04.052. cracking.” Soil Suction Applications in Geotechnical Engineer-
Collins, R.W. and Miller, G.A. (2013). Interpretation of in Situ Tests ing Practice, ASCE, Geotechnical Special Publication 48,
as Affected by Soil Suction. Oklahoma Transportation Center No. 1437.
OTCREOS11. 1-45-F. Nichol, C., Smith, L., and Beckie, R. (2003). “Long-term measure-
Durkee, D.B. (2000). Active Zone Depth and Edge Moisture Varia- ment of matric suction using thermal conductivity sensors.”
tion Distance in Expansive Soils. Ph.D. Dissertation, Department Canadian Geotechnical Journal, 40(3), 587597, doi.org/10.
of Civil Engineering, Colorado State University, USA. 1139/t03-012.
El-Garhy, B.M. and Wray, W.K. (2004). “Method for calculating the Nguyen, Q., Fredlund, D.G., Samarasekera, L., and Marjerison, B.L.
edge moisture variation distance.” Journal of Geotechnical and (2010). “Seasonal pattern of matric suctions in highway
Geoenvironmental Engineering, ASCE, 130(9), 945955, subgrades.” Canadian Geotechnical Journal, 47(3), 267280,
doi.org/10.1061/(ASCE)1090-0241(2004)130:9(945). doi.org/10.1139/T09-099.
Fredlund, D.G. and Vu, H.Q. (2003). “Numerical modelling of O’Rourke, T.D. and Bonneau, A.L. (2007). “Lifeline performance
swelling and shrinking soils around slabs-on-ground.” Proc., under extreme loading during earthquakes.” Earthquake
Post-Tensioning Institute Annual Technical Conference, Hun- Geotechnical Engineering: 4th International Conference on
tington Beach. Earthquake Geotechnical Engineering-Invited Lectures, 6,
Heran, W.D., Green, G., and Stoeser, D.B. (2003). A Digital Geologic 407432, doi.org/10.1007/978-1-4020-5893-6_17.
Map Database of Oklahoma, U. S. Geological Survey Open File Post-Tensioning Institute. (1980). Design and Construction of Post-
Report 03-247. Tensioned Slabs-on-Ground. Phoenix, Arizona.
Illston, B.G., Basara, J.B., and Crawford, K.C. (2004). “Seasonal to Reece, C.F. (1996). “Evaluation of a line heat dissipation sensor for
interannual variations of soil moisture measured in Oklahoma.” measuring soil matric potential.” Soil Science Society of America
International Journal of Climatology, 24(15), 18831896, Journal, 60(4), 10221028, doi.org/10.2136/sssaj1996.
doi.org/10.1002/joc.1077. 03615995006000040009x.
Illston, B.G., Basara, J.B., Fiebrich, C.A., Crawford, K.C., Hunt, E., Richards, B.G. and Chan, C.Y. (1971). “Theoretical analyses of sub-
Fisher, D.K., Ellictt, R., Fiebrich, C.A., Crawford, K.C., Humes, grade moisture under Australian environmental conditions and
K., and Hunt, E. (2008). “Mesoscale monitoring of soil moisture their practical implications.” Australian Road Research, 4(6),
across a statewide network.” Journal of Atmospheric and 3249.
Oceanic Technology, 25(2), 167182, doi.org/10.1175/ Rogers, J.D., Olshansky, R., and Rogers, R.B. (1993). “Damage to
2007JTECHA993.1. foundations from expansive soils.” Claims People, 3(4), 14.
Johnson, H.L. (2008). Climate of Oklahoma. Erath Sciences and Swenson, S., Famiglietti, J., Basara, J., & Wahr, J. (2008). “Estimat-
Mineral Resources of Oklahoma, Educational Publication 9, ing profile soil moisture and groundwater variations using
Oklahoma Geological Survey. GRACE and Oklahoma Mesonet soil moisture data.” Water
Luza, K.V. and Johnson, K.S. (2005). Geologic Hazards in Oklahoma. Resources Research, 44(1), W01413, doi.org/10.1029/
Oklahoma Geological Survey. 2007WR006057.
Lytton, R.L. (1994). “Prediction of movement in expansive clays.” The Oklahoma Climatological Survey. (2011). Climate of Oklahoma.
Vertical and Horizontal Deformations of Foundations and Retrieved January 20, 2015, from http://climate.ok.gov/index.
Embankments, ASCE, Geotechnical Special Publication 40, php/site/page/climate_of_oklahoma
18271845. The Oklahoma Mesonet. (2016). Data Descriptions. Retrieved Janu-
McKeen, R.G. and Johnson, L.D. (1990). “Climate-controlled soil ary 4, 2018, from https://www.mesonet.org/index.php/site/
design parameters for mat foundations.” Journal of Geotechnical about/data_descriptions
Engineering, ASCE, 116(7), 10731094, doi.org/10.1061/ Visser, A., Stuurman, R., and Bierkens, M.F. (2006). “Real-time
(ASCE)0733-9410(1990)116:7(1073). forecasting of water table depth and soil moisture profiles.” Ad-
Mitchell, P.W. (1979). The Structural Analysis of Footings on vances in Water Resources, 29(5), 692706, doi.org/10.1016/
Expansive Soil. Kenneth W. G. Smith & Associates. Research j.advwatres.2005.07.011.
Report No. I, Webb & Son, Adelaide, Australia. Wray, W.K. (1987). “Evaluation of static equilibrium soil suction
Morris, P.H., Graham, J., and Williams, D.J. (1992). “Cracking in envelopes for predicting climate-induced soil suction changes
drying soils.” Canadian Geotechnical Journal, 29(2), 263277, occurring beneath covered surfaces.” Proceedings of the 6th
http://doi.org/10.1139/t92-030. International Conference on Expansive Soils, 1, 235240.