by future generations. In order to use this resource consciously, it is known in the area of agricultural irrigation

the need to evaluate the amount of water to be used. Among the soil moisture content determination methods,

sensors, coupled to a programmable logic controller, emerge as an alternative to conventional laboratory

methods, making the procedure faster and less labor intensive. In this sense, the present work aimed to evaluate

gerenciado pelas gerações futuras. De forma a utilizar este recurso de forma consciente, estudos na área de

irrigação agrícola apontam a necessidade de se estimar com precisão a quantidade de água a ser usada. Dentre

os métodos de determinação da umidade do solo, os sensores, acoplados a um controlador lógico programável,

surgem como uma alternativa aos métodos convencionais de laboratório, tornando o procedimento mais rápido

Portanto, é necessário calibrar cada sensor individualmente.

The World Economic Forum listed the water crisis as the

most devastating global risk. Considered an essential

component for the preservation of the present and future

generations, it is necessary that water is used with increasingly

parsimony (ROCCARO; VERLICCHI, 2018; JINDAL et al.,

2017). Thus, the design of agricultural production systems

Currently, there are several methods used to determine

soil moisture content (at laboratory and field levels), however

there is no consensus among the field experts. According to

Pouso (2012), the choice of method should consider the

objectives desired by the researcher and/or producer, the

project or product, the desired precision level, and other

factors that may limit their choice. Klar (1988) reports that all

Within this context, the emergence of faster alternative

methods, based on sensors, has been gaining more and more

prominence. In general, soil moisture measurement methods

using sensors are based on the measurement of soil dielectric

properties, which are directly correlated to soil water content

around sensor probes (WILL; ROLFES, 2013).

The coupling of a Programmable Logic Controller (PLC)

to the soil moisture sensor allows the development of more

frequency, since measurements can be taken at any time, with

no need to go to the field. Another advantage of the use of

sensors is the possibility of performing sampling in large

areas, requiring only a greater number of sensors connected

to a monitoring system through wireless connection.

The soil moisture content sensors model FC-28 have

been used in several studies in the last few years, emphasizing

the implementation in precision agriculture systems for the

monitoring of soil moisture content in agriculture

(GADDAM et al., 2014; VANI; RAO, 2016). Some of the

mainly due to the local characteristics where the calibrations

presence of plant roots, rocks, climatic conditions and

different soil textures directly affect the calibration of the

sensors (EVETT et al., 2006; LOGSDON, 2009; RÜDIGER

et al., 2010; MITTELBACH et al., 2011).

In this sense, this work aimed to evaluate the reliability

and precision of a low-cost soil moisture sensor. For this, the

following hypotheses were evaluated: a) Does soil texture

influences the sensor reading, which would imply the need to

The project was developed in the Laboratory of

Drainage, Irrigation and Environmental Sanitation

(LaDISan) of the Department of Agricultural and

Environmental Engineering (TER1), located at Praia

Vermelha Campus, Fluminense Federal University, Niterói-

RJ, Brazil. Located approximately in the geographical

coordinates 22. 90º S; 43. 13º W.

microcontroller chip ATmega328 and 3.3 or 5.0 V operating

voltage (MOWAD et al., 2014).

In this work, three soil moisture sensors of the model FC-

28 from Glyduino manufacturer, were used. The FC-28

sensors are composed of 4 pins: a digital communication pin,

processing and sending it to the microcontroller.

The electrodes are responsible to measure the electric

potential difference, caused by the resistance to the electric

current flow in the ground. The data obtained are measured

in bits (analog values) by the sensor. (KOLAPKAR et al.,

2016; RATHORE; SINGH, 2015). The sensor principle of

operation is based on the application of an electrical signal,

through one of the electrodes of the probe through the

ground, which travels through the soil and reaches the

second electrode of the probe. The electric potential

were used to calibrate the sensors and to evaluate their

accuracy. In addition, to ease the data acquisition an LCD

display model RT162-7 was installed to inform the measured

soil moisture content in real-time.

Three soil types with different textures were used (sandy

soil, medium texture soil and clay soil). The granulometric

analysis and textural classification of the samples were

performed by the pipette method, according to the

was a 2x2 factorial with blocking by sensor. There were two

factors, being factor A the soil type (sandy, medium and

clayey); and factor B the soil moisture content (15, 28, 38 and

52%); the experimental units were the crucibles, with

approximately 300 g of soil. There were performed 15

replications for each factor-level combination. The

content by means of the MPUE was carried out by the

following equation (Equation 1).

After filtering the data as a function of the mean values

In general, the values read by the sensors differ

significantly depending on the soil texture, supporting the

initial hypothesis (Table 2). Therefore, in order to use this

sensor properly, the user would need to have prior

knowledge of the soil type in which the sensor will be

installed.

presented in Table 3.

The voltage values measured by the sensors and the

values determined by the MPUE were classified as very

strong correlation (-1.0 <r <0.90 or 0.90 <r <1.0) according

to the classification proposed by Hinkle et al. al. (2003). The

values in parentheses (p-value) located below the values of

the correlation coefficient indicated that all correlations were

significant by the t test (p <0.05), proving that there is a

strong correlation between the values measured by the sensor

and the expected moisture content of the soil.

with their respective values of moisture content, determined

by the standard official methodology in Brazil for the

determination of the moisture content, the results are

presented in Figure 1.

It can be seen from Table 4 that all soil types presented a

negative slope for regression (β1), which confirms the results

obtained previously. In addition, constant variance (p-value>

α) of the sample residuals was observed by the Breush-Pagan

test, indicating the adequacy of the adopted model. The

normality assumption was not met for any type of soil (p-

tested on soils with different texture, physical and chemical

characteristics, which explains the results found in this work.

Based on the data displayed at table 1, when looking

within the same texture, for sand 28% had two readings equal

but different for A1; at medium texture S1 kept this behavior,

moistures (neither S1). At clay soil S1 had no voltage

variation at any moisture, same behavior for S2 and S3 but at

different levels.

This information could indicate the presence of a

systematic error, probably due to manufacture process

unable to give all sensors the same features, from what

adopting a single global calibration is not recommended for

the sensors.

moisture measurement with less them 10% threshold, as they

would be statically equal (p <0.05), and this could be higher

depending on the soil texture.

For table 2, grouping sensor readings by moisture level

(where was expected to have reading not differing within the

same moisture, regardless the soil texture); all three sensors

were able to read 15% moisture for all three textures (did not

differ significantly), but at different voltage for each texture,

indicating that even when they translate well the moisture

(into voltage), the use of a single calibration model

disregarding the soil texture, is a mistake.

When calibrating and building the model, it was verified

that the voltage values presented an inverse relation to the

soil moisture, which was already expected, since low voltage

values are associated with low values of resistance to the

passage of electric current in the soil. Results obtained by

Vernandhes et al. (2017) corroborated this finding.

The R² values obtained by the calibration curves (0.841,

0.858 and 0.872) indicated an efficient calibration, validating

the calibration of these sensors. PAYERO et al. (2017), using

Decagon EC-5 model soil moisture sensors, which have a

0.556 for sandy, medium texture and clay soils, respectively.

The values found were relatively high, as there are

commercial sensors with errors lower than 3%, such as the

Fieldscout 350 TDR probe (manufacturer Spectrum

Technologies, Inc.). Such results cast doubt on the accuracy

and reliability of the sensors studied in this work.

Therefore, the use of sensors should be restricted to

operations that do not require high accuracy (i.e. lower than

10%). In addition, the sensors should not be used in

experiments that require data acquisition in several locations

The evaluated sensors did not present good precision to

estimate soil moisture content without previous calibration

procedure.

Soil texture directly influenced the readings obtained by

the sensors at a given value of moisture content, so – once

information, resulting in waste of resources or an inefficient

irrigation.

The evaluated sensors did not present replicability, so it

is necessary to calibrate each sensor individually. Therefore,

the use of this sensor model for monitoring soil conditions

in larger areas is not feasible (due to soil texture variation and

large moisture range).

Despite its low cost, flexibility and fast data acquisition,

this sensor model requires a time-consuming calibration

procedure.

The authors thank Professor Dario de Andrade Prata

Filho for allowing the use of the LaDISan laboratory where

the experiment was developed and for his aid to the research.

EUSTAQUIO, J. F. L. L.; SANTANA, G. M.; SILVA, E. L.

Bradesco, 2016. 4p. Disponível em:

http://sistemaolimpo.org/midias/uploads/c85c65e0c68

fcaf7e6ccdfbbd7e847e5.pdf. Acesso em: 25 fev. 2021.

EVETT, S. R.; TOLK, J. A.; HOWELL, T. A. Soil profile

water content determination: Sensor accuracy, axial

response, calibration, temperature dependence, and

2006. DOI: 10.2136/VZJ2005.0149

https://doi.org/10.2136/sssaj2008.0146

NY: Apress Inc., 2010. 475p.

MITTELBACH, H.; CASINI, F.; LEHNER, I.; TEULING,

A. J.; SENEVIRATNE, S. I. Soil moisture monitoring for

climate research: evaluation of a low-cost sensor in the

framework of the swiss soil moisture experiment

935-939, 2014.

PAYERO, J. O.; NAFCHI, A. M.; DAVIS, R.;

KHALILIAN, A. An Arduino-Based Wireless Sensor

Network for Soil Moisture Monitoring Using Decagon

p. 288, 2017. DOI: 10.4236/ojss.2017.710021

POUSO, M. T. P. Sistema de automação e controle de um

sistema de irrigação. Brasília: UniCEUB, 2012. 85p.

RATHORE, J.; SINGH, J. Review on Wireless Sensor

RÜDIGER, C.; WESTERN, A. W.; WALKER, J. P.;

SMITH, A. B.; KALMA, J. D.; WILLGOOSE, G. R.

Towards a general equation for frequency domain

319-329, 2010. DOI:

https://doi.org/10.1016/j.jhydrol.2009.12.046

SHARMA, H.; SHUKLA, M. K.; BOSLAND, P. W.;

STEINER, R. Soil moisture sensor calibration, actual

evapotranspiration, and crop coefficients for drip

2016. DOI: 10.17485/ijst/2016/v9i31/95340

VERNANDHES, W.; SALAHUDDIN, N. S.;

KOWANDA, A.; SARI, S. P. Smart aquaponic with

monitoring and control system based on iot.

In: Informatics and Computing (ICIC) and Second

USA, 2013. Available in:

https://ieeexplore.ieee.org/abstract/document/668852

9