See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/313720615

Arduino-based system for soil moisture measurement

Conference Paper · February 2017

CITATIONS READS

3 11,391

2 authors:

Vuk Radman Milutin Radonjic

2 PUBLICATIONS   3 CITATIONS   
University of Montenegro
78 PUBLICATIONS   253 CITATIONS   
SEE PROFILE
SEE PROFILE

Some of the authors of this publication are also working on these related projects:

ARCHITECTURE, DESIGN AND PERFORMANCE OF DCQ SWITCH View project

DEVELOPMENT OF FUTURE INTERNET ROUTER ON PROGRAMMABLE HARDWARE AND OPEN SOURCE SOFTWARE PLATFORMS View project

All content following this page was uploaded by Milutin Radonjic on 15 February 2017.

The user has requested enhancement of the downloaded file.

 22nd Internacionalna Naučno-Stručna Konferencija Informacione Tehnologije 2017

Arduino-based system for soil moisture

measurement
Vuk Radman, Milutin Radonjić, Member, IEEE

 excessive soaking, as well as to prevent runoff of

Abstract — A substantial number of modern systems for fertilizers and other chemicals into the deeper layers of the
soil moisture measurement exploit two properties of soils: soil.
moisture content and water potential. Sensors for measuring Purpose of this work was to design a simple system for
these parameters vary substantially in precision and cost.
Moisture content sensors provide data that is more suitable
soil moisture monitoring, using affordable components
for human reading. However, these sensors are less reliable that provide reliable and accurate measurement. The
or very expensive. In this paper we describe the system that system is intended primarily for application in agricultural
utilizes inexpensive and precise sensors, which provide a activities. Although there are a number of soil moisture
convenient way of data presentation. For this purpose, the monitoring systems on the market, numerous
Watermark 200SS granular matrix water potential sensor is disadvantages limit the scope of their application. Systems
used, supported by a scaled MV-VG model for soil moisture
retention curve prediction. This model provides a correlation
that provide data in the form suitable for agricultural
between the soil moisture content and water potential. Since applications (soil moisture volume percentage / soil
our proposed system relies on inexpensive components, it is a moisture content) are very expensive and require
suitable alternative to similar expensive systems, especially sophisticated electronic equipment for their operation. On
for multiple monitoring locations. the other hand, affordable systems usually display data in
unsuitable form (voltage, resistivity, frequency, suction
Keywords — soil moisture measurement, Arduino UNO, force, ...).
Watermark 200SS, soil moisture retention curve.
The sensor used in our system measures suction force
(water potential), while the water content mostly depends
on physical characteristics of the soil. To draw a
I. INTRODUCTION
correlation between these two parameters (water potential

D EVELOPMENTS in technology, science and

medicine have allowed a much better quality and
longevity of a man’s life, leading to a continuous increase
and moisture content), we need to determine the soil
moisture retention curve (SMR). The accurate SMR
determination is usually performed by laboratory analysis,
in population. Consequently, requirements for adequate which we will substitute with a complex mathematical
quantities of food and drinking water has increased, as model. Due to certain model approximations and the
well as the need to improve its quality, especially in third specificities of different soil types, additional calibrations
world countries where each of these recourses are limited. are necessary for very precise measurements.
Through the usage of the "smart" irrigation systems and Paper is organized as follows. Determination of SMR is
monitoring of soils we are heading in the right direction to explained in Section II. Hardware design of proposed
meet these requirements. system is presented in Section III. Calibration of obtained
Soil moisture measurement is of great importance for results is elaborated in Section IV. Paper is concluded in
applications in the field of agriculture, allowing farmers to Section V.
more efficiently manage irrigation systems. Knowing the
exact properties of the soil, farmers are not only able to II. SOIL MOISTURE RETENTION CURVE
use less water to grow crops, but also increase production
Measuring the soil parameters, such as the soil moisture
and quality, irrigating the plants in the critical periods of
retention curve (SMR) is expensive and time consuming,
its growth. Moisture sensors are commonly used in
very difficult and requires laboratory conditions and
combination with irrigation system controllers. Connecting
equipment. Therefore, many scientists have made an effort
a moisture sensor to an ordinary timer irrigation system, it
to develop indirect methods as an alternative. These
gets some "smart" features, stopping irrigation when the
indirect methods are classified into empirical, conceptual
soil is already wet enough. Soil moisture sensors also
and semi-physical.
found diverse applications in scientific research of
Semi-physical methods are based on curve similarity
agriculture and horticulture, climate and environment
between the SMR and the particle size distribution curve
related research and many other areas of science. For
(PSD), implying that the pore size distribution (PoSD) is
example, on golf courses these sensors are used to prevent
closely related to the PSD. This PoSD-PSD relation is
determined by scaling the pore length using a scaling
V. Radman is with the Faculty of Electrical engineering, University of
Montenegro, Podgorica (e-mail: [email protected]) factor α, of an average value of 1.38. However, it is later
M. Radonjić is with the Faculty of Electrical engineering, University revealed that α value varies between 1.02 and 2.97 for
of Montenegro, Podgorica (e-mail: [email protected])
 22nd Internacionalna Naučno-Stručna Konferencija Informacione Tehnologije 2017

fine- and coarse- textured soils, respectively [1]. Even the continuous SMC parameters such as n, m and .
slightest error in the estimation of α may result in immense Since the MV-VG model assumes that the residual
error while estimating the SMR. The essential problem of moisture content is zero, leading to an underestimation in
the semi-physical methods is leading to the dry range dry range of the SMR, especially in clayey soils, we used a
underestimation of the SMR. conceptual scaling approach to reduce the model bias.
To predict the continuous SMR from the PSD data and
B. Scaling Approach
porosity, we used the parameterized Van Genuchten model
(VG), which was integrated with the Mohammadi and Let us assume that the reference soil consists of
Meskini-Vishkae (MV) model. Model was tested on 82 particles, which are spherical and uniform in size and are
soils samples from the UNSODA database, which has arranged in random close packing state leading to minimal
shown that the scaling approach improves the MV-VG porosity. This minimal porosity is known as the "Kepler
model estimations on average by 30% [2]. Results show conjecture" with value of 0.259 for this packing state or, in
that the scaled method properly estimates the SMR for all our case, for the reference soil. On the other hand, the
soil samples and may be used at larger field scale for maximum value of porosity for the reference soil
predicting continuous SMR from basic soil data. would equal to 1.41432. Thus we can suggest the scaling
factor λ, for each soil sample as:
A. Theory
. (10)
Model of the SMR developed by the Van Genuchten [3]
is widely used and mathematically defined as: Now being that the packing parameter ξ and pore-size
distribution index n are small for the fine-textured and
, (1) large for the coarse-textured soils, F. Meskini-Vishkaee,
, (2) M. H. Mohammadi and M. Vanclooster suggest that the
scaling factor can scale the index n as follows:
where is effective saturation degree, is the soil ∗
∗ , (11)
moisture content, and are saturated and residual soil where n* is scaled to the PoSD index in the VG model [2].
moisture contents, respectively. Parameter h represents the Finally, the scaled model is determined as:
suction head, n represents pore-size distribution index, is
the pore-length scaling factor and m is a fitting coefficient.
The suction head, hi is given as: ∗ . (12)
. ∗
, (3)
III. HARDWARE DESIGN
and corresponds to the particle radius of the i-th fraction Ri
Our design is based on the Arduino UNO
[4]. Parameter is defined as:
. microcontroller platform, equipped with the soil moisture
, (4) sensor Watermark 200SS and the GSM/GPRS SIMCOM
and represents a coefficient depending on the state of soil 900 mobile communication module (Fig. 1). The sensor is
particle packaging. Parameter e is the void ratio given by: connected to a built-in 10-bit analog-to-digital converter
, (5) (ADC) of the Arduino platform. The platform measures
where is soil particle density and is bulk density. output voltage of the moisture sensor, processes obtained
Because of the similarity between the SMR and PSD data and sends it over the GPRS connection to the cloud.
curves, many scientists expressed a SMR model in terms
of a PSD model. Arya and Paris (AP) [5] suggested that
the soil moisture content θ can be obtained from PSD and
as
∑ ; i = 1, 2, 3, ..., k, (6)
where is the mass fraction of particles in the j-th
particle size fraction. Let us now consider that
∑ , (7)
where is the cumulative mass fraction of soil particles,
would result in
, (8) Fig. 1. Experimental prototype of the system.
where S is the saturation degree. If = 0, then = S and
subsequently S = . However, the AP model ignores the In order to properly measure the water potential using
residual moisture content, which has considerable values the Watermark sensor, it is necessary to provide an
for many types of soil. Combining (1) and (3) and alternating current (AC) power source, because long
invoking them in (7) yields exposure to the direct current polarize sensor electrodes.
This has a negative impact on the measured results, and
. (9) leads to the degradation of the sensor over time. The AC
. ∗
power source is emulated by digital output ports of the
Fitting (9) to the PSD data enables us to predict the microcontroller, as shown in Fig. 2. Alternating the status
 22nd Internacionalna Naučno-Stručna Konferencija Informacione Tehnologije 2017

of the digital pins D0 and D1 to HIGH (5V) and LOW the linear function:
(0V) digital levels, rectangular signal is generated. 20 ∗ ∗ 1 0.018 ∗ 24 0.55 (14)
Initially, D0 is set to HIGH, while D1 is set to LOW, so where P (kPa) is a potential humidity, R (kΩ) is the sensor
the current flows in one direction. We then measure the resistance and T (°C) is the soil temperature. The value
voltage drop on the sensor and perform the necessary 0.018*(T-24) represents the 1.8% resistance change for
calculations. The voltage polarity is then reversed, putting each degree of Celsius change in soil temperature from the
D0 to LOW and D1 to HIGH. The current now flows in reference value of 24°C.
the opposite direction and the measurement starts again.
TABLE 1. CALIBRATION POINTS OF THE WATERMARK 200SS
SENSOR
Moisture potential [kPa/cbar] Resistivity [Ω]
0 550
9 1000
10 1100
15 2000
35 6000
55 9200
75 12200
100 15575
200 28075
Fig. 2. Watermark 200SS sensor connection to Arduino.

Voltage drop (resistance of the Watermark 200SS

sensor) is measured using A0 input of the Arduino ADC.
Arduino uses the 10-bit ADC converter which has a 4.88
mV resolution. This means that the final resistance value is
to be calculated with a certain error due to the imprecision
of the ADC. The sensor voltage is calculated according to
the voltage divider as follows:
5∗ , (13)
where V is the sensor voltage and Rw is sensor resistance.
After calculation of the resistance using (13), we
proceed to the soil moisture potential calculation using Fig. 3. Transfer function of the Watermark 200SS
(14), (15), and (16). Finally, after applying the moisture sensor.
potential into scaled MV-VG model using (12), we acquire
the value of the moisture content for the desired soil type. The second segment projection, which covers the range
The final value is then transmitted using the GPRS module from 1000Ω to 8000Ω, was developed by Shock [7]. This
to the desired location. equation, developed from purely empirical data, deviates
We should take into consideration a flaw of this method from the values in Table 1 for moisture potential values
of measurement. Due to the large difference of the greater than 100 kPa :
. ∗ .
unknown (sensor) resistance and the reference resistors . (15)
. ∗ . ∗
(18kΩ), the result can be fairly inaccurate. This problem
arises due to the entirety of voltage drop occurs across the Finally, for scope over 8kΩ, quadratic equation
resistor of larger value, so consequently the unknown developed by Rick Allen [8] is used:
voltage carries far less useful information during AD 2.246 5.239 ∗ ∗ 1 0.018 ∗ 24
conversion. This problem can be resolved by an auto- 0.06756 ∗ 2 ∗ 1 0.018 24 . (16)
ranging voltage divider. Equation (16) was determined by the simple square
regression of the values taken from Table 1 in the range of
IV. CALIBRATION OF RESULTS 10kPa to 200kPa. The equation has a coefficient of
In order to perform the required conversion of the determination r2=0.9996 and a standard error of 1.07kPa
sensors electrical resistance into the soil moisture potential [8].
it is necessary to perform the calibration. For this purpose, Although the temperature sensor calibration is necessary
a number of calibration points, at a temperature of 24°C, for serious research of soil moisture, our design does not
are provided by the manufacturer [6] (Table 1). implement the temperature sensor. The system was
Based on calibration points given in Table 1 it is originally designed for agricultural uses and in the
possible to draw the transfer function of the Watermark growing season changes in soil temperature rarely vary
200SS sensor (Figure 3). The most efficient way for from the standard 24°C. However, the possibility of a
reproducing the curve in Fig. 3 is to divide the resistance temperature calibration is included in the calculations and
domain into three ranges and project the corresponding may be implemented in the future works.
curves for each of them. The first section of the curve Typical examples of the measured and predicted SMRs
covers the range from 0Ω to 1000Ω, which is estimated by values with the MV-VG model, scaled approach and
 22nd Internacionalna Naučno-Stručna Konferencija Informacione Tehnologije 2017

Rosetta software (computational modeling and analysis

software based on a purely empirical data) for clay and silt
loam textures, are shown in Fig. 4 and Fig. 5 [2]. For both
textures, scaled approach well corresponds to the
measured data and gives better prediction than the MV-
VG model in the whole range of the SMR. For silty loam
(Fig. 5) scaled approach overestimates the measured data
slightly while MV-VG model underestimates the wet part
of the SMR [2]. The graphs also shows that Rosetta
software performs very badly while predicting the wet
parts of the SMR. Fig. 6. Measured moisture content for clay soil.

V. CONCLUSION
In this paper we described design of the microcontroller
system for soil moisture measurement using Watermark
200SS soil moisture sensor. The idea was to develop a
modular, flexible system using affordable components.
Proposed solution is designed to measure water potential
from 0kPa to 200kPa, and to be used for agricultural
purposes. System is flexible, low cost, simple to use and
offers real-time monitoring. It is a suitable alternative to
similar expensive systems, especially in case of multiple
monitoring locations.
Future work on the system will include several different
Fig. 4. Example of measured vs. predicted SMRs using activities. Main point of interest will be to evaluate the
integrated MV-VG model, scaling approach and Rosetta recorded data and compare it to high accuracy systems.
software for clay textured soils. Attention will also be focused to autonomy of the system,
which can be achieved through autonomous power supply
consisting of solar energy or batteries and implementation
of attention (AT) commands. This greatly improves
flexibility and manageability of the system as it allows us
to control the system from a central point wirelessly,
without the need of field work. Another important issue is
to implement a temperature sensor for increased precision
and calibration of the moisture sensor.

LITERATURE
[1] A. Basile and G. D’Urso, “Experimental corrections of simplified
methods for predicting water retention curves in clay-loamy
soilsfrom particle-size determination”, Soil Technol., 10, 261–
272,1997.
[2] F. Meskini-Vishkaee, M. H. Mohammadi, and M. Vanclooster.
Fig. 5. Example of measured vs. predicted SMRs using Hydrol. Earth Syst. Sci., 18, 4053–4063, 2014. "Predicting the soil
integrated MV-VG model, scaling approach and Rosetta moisture retention curve, from soil particle size distribution and
bulk density data using a packing density scaling factor".
software for silty loam textured soils. [3] Van Genuchten, M. Th. 1980. "A closed-form equation for
predicting the hydraulic conductivity of unsaturated soils". Soil Sci.
The proposed system has been tested in a sample of clay Soc. Am. J. 44:892-898.
[4] Mohammadi, M. H. and Vanclooster, M.: Predicting the soil
soil in a large flower pot at a depth of 20 cm. The results moisture characteristic curve from particle size distribution with a
shown in Fig. 6 were obtained during a short test of tens of simple conceptual model, Vadose Zone J., 10, 594–602, 2011.
minutes of the previously sprinkled soil, with periods of [5] Arya, L. M. and Paris, J. F.: "A physicoempirical model to predict
the soil moisture characteristic from particle-size distribution and
measurement of a little more than a minute. The initial soil bulk density data". Soil Sci. Soc. Am. J., 45, 1023–1030, 1981.
moisture was at 50.9%. The soil is then briefly moistened [6] From Irrometer Company, Inc., Riverside, CA via M.K. Hansen
and allowed to dry over a period of several minutes. Company, E.Wanatchee, WA, October, 2000.
Increase in soil moisture (from first to fifth point) can be [7] Shock, C.C., J.M. Barnum, and M. Seddigh. 1998. Calibration of
Watermark soil moisture sensors for irrigation management.
noticed until the soil has reached the maximum humidity Proceedings of the 1998 Annual Meeting of the Irrigation
saturation of 51%. The excess water is then removed and Association. p. 139-146.
slight pressure is applied to the soil to remove a certain [8] Rick Allen (October 13, 2000)." Calibration for the Watermark
200SS Soil Water Potential Sensor to fit the 7-19-96 “Calibration
amount of water for the purposes of demonstration (fifth to #3” Table from Irrometer". University of Idaho.
tenth point). The rest of the graph shows a constant value
since the drying of clay soils is a time consuming process.

View publication stats