Water sampling with drones

Abstract

This paper presents proposals to deal with common limitations in surface water sampling routines. Prototypes 1 and 2 were developed in a way that, in addition to having in common the use of small drones, both of them were designed to be simple to build, operate, and maintain. The results of laboratory and field experiments were satisfactory and showed that especially prototype 2 (a disposable bailer connected to a quadcopter with a camera that transmits images in real-time) has the potential to significantly improve fieldwork. Besides its simple operation and maintenance, prototype 2 was able to collect up to 150mL of water in each manual controlled flight – sufficient for 15 turbidity measurements in a portable turbidimeter – with a 5m vertical distance between the drone and the water surface.

Full text

XXIII Brazilian Symposium on Water Resources (ISSN 2318-0358) 1
XXIII BRAZILIAN SYMPOSIUM ON WATER RESOURCES
WATER SAMPLING WITH DRONES
1
Caio P. Cavalhieri
2
ABSTRACT – This paper presents proposals to deal with common limitations in surface water
sampling routines. Prototypes 1 and 2 were developed in a way that, in addition to having in
common the use of small drones, both of them were designed to be simple to build, operate, and
maintain. The results of laboratory and field experiments were satisfactory and showed that
especially prototype 2 (a disposable bailer connected to a quadcopter with a camera that transmits
images in real-time) has the potential to significantly improve fieldwork. Besides its simple
operation and maintenance, prototype 2 was able to collect up to 150mL of water in each manual
controlled flight – sufficient for 15 turbidity measurements in a portable turbidimeter – with a 5m
vertical distance between the drone and the water surface.
Keywords: drone, Arduino, bailer.
1. INTRODUCTION
Although current surface water sampling procedures and equipment are considerably
sophisticated, none of the existing solutions are satisfactory enough when there are either budgetary
constraints or practical difficulties during fieldwork (GETTEL, 2011). Solutions such as monitoring
buoys for reservoirs usually require significant investments to keep track of a single water control
point, which often makes them financially prohibitive when it is necessary to monitor several areas
simultaneously (ANTTILA, 2012). On the other hand, there are situations in which the budget is not
a limiting factor, but the logistical difficulties associated with the use of certain equipment can
compromise the representativeness of water monitoring data.
In addition to the fact that they have inherent limitations, there are also practical difficulties in
addressing renowned water sample collection procedures. For instance, the Standard Methods for
the Examination of Water and Wastewater (1060 COLLECTION AND PRESERVATION OF
SAMPLES, 2017) preconizes that manual collections must be made from equidistant points along
the cross-section of the water body and, if only one collection can be made, the sample must be
taken from the middle of the main channel of the water body and at half depth. However, following
1
This article was originally published in Portuguese in the Proceedings of the XXIII Brazilian Symposium on Water Resources in 2019 and it is
available at <http://anais.abrh.org.br/works/5012>.
2
Caio P. Cavalhieri has a bachelor´s degree in environmental engineering from Universidade de São Paulo (USP) and a master’s degree in civil
engineering from Universidade Estadual de Campinas (Unicamp), both of them public research universities in Brazil. Since 2008, Caio works as a
researcher at IPT, the Institute for Technological Research of the State of São Paulo, where he uses and develops environmental monitoring tools. He
was a visiting researcher in 2012 at the Department of Soil Sciences at North Carolina State University in the USA. Contact: caiopc@ipt.br

XXIII Brazilian Symposium on Water Resources (ISSN 2318-0358) 2
these guidelines is not always a simple task and leads to water monitoring programs in which not all
sampling procedures can be fully completed.
Seeking alternatives to conventional surface water sampling solutions, the present work
developed two prototypes that collect samples from the top of water bodies using tools such as
small quadcopter drones, embedded microcontrollers, and disposable bailers. As reported below,
the results of this work were satisfactory, and, in general, both prototypes have the potential to
significantly improve fieldwork routines based on surface water sampling.
2. OBJECTIVES
The general objective of this work was to develop solutions for surface water sampling
services that: (i) meet established sampling procedures; (ii) are considered simple to build, operate
and maintain; and (iii) demonstrate the potential to become viable products and services. Therefore,
the specific objective of this research was to create prototypes of products and services integrating
different emerging technologies and that, at the same time, could have their performance evaluated
based on laboratory and field experiments.
3. MATERIALS AND METHODS
Based on the research of Ore (2014) and Schwarzbach (2014) on automation to collect surface
water samples with drones (Figures 1 and 2), the present work assumed that small drones would be
essential to develop alternative approaches to conventional sampling procedures. In fact, water
sampling with drones has the potential to replace stationary devices and significantly reduce the
need for trails, bridges, or boats to access different sampling points. Furthermore, the introduction
of drones in sampling routines can provide faster, safer, and more reliable collections, which would
dramatically change the dynamics of fieldwork.
For these reasons, two radio-controlled drones were used in this work: a DJI Phantom 1
(without any embedded camera) and a DJI Phantom 3 Professional – with an embedded camera
providing real-time image transmission from the aircraft (first-person view – FPV). This work also
applied the concept of modularity that, according to Banzi (2011), has to do with the construction of
complex systems based on simple connected devices. This approach resulted in two prototypes
created simply by combining different modules (such as microcontrollers, sensors, and disposable
bailers), as described below.

XXIII Brazilian Symposium on Water Resources (ISSN 2318-0358) 3
Figures 1 and 2 – Examples of surface water sampling devices attached to drones. Sources: adapted from Ore
(2014), and; adapted from Schwarzbach (2014).
Additionally, due to a large number of existing sampling techniques, this work chose to
address only one monitoring procedure and, therefore it was oriented to meet the specificities of
samplings dedicated to measuring turbidity levels in water bodies such as lakes, rivers, and
reservoirs. For this type of procedure, each sample collected must have a minimum volume of
10mL, which is necessary to measure turbidity levels using a portable turbidimeter.
3.1 Prototype 1: microcontroller and peristaltic pump
Prototype 1 was mainly based on the work of Ore (2014) and Schwarzbach (2014). Like the
solutions developed by these authors, this prototype associated a quadcopter with a hydraulic pump,
and its activation was programmed to happen whenever the drone hovered over any water body at a
previously defined height (Figure 3).
Figure 3
3
– Prototype 1 is a sampler device attached to the drone that sends an ultrasonic signal during the flight (a)
and, when it is 30cm far from the water surface (b), this signal is reflected (c) and sampling starts (d).
3
Illustration by André D. L. Zanchetta.
water sensor
vials
ultrasonic
sensor
tube
1m
pump
passive
safety
tether
(a)
(b)
(c)
(d)

XXIII Brazilian Symposium on Water Resources (ISSN 2318-0358) 4
The water sampling device was developed using an Arduino UNO (a light microcontroller
board ideal to get started with electronics and coding), a peristaltic pump, and an HC-SR04
ultrasonic sensor (Figures 4 and 5). Besides, the circuit (Figure 6) was designed so that the total
mass of the system was as light as possible.
Figures 4, 5, and 6
4
– Prototype 1 has a peristaltic pump, an HC-SR04 ultrasonic distance sensor, and a circuit board.
This prototype also had a plastic case (to allow all devices to be attached to the lower portion
of the DJI Phantom 1), a 35cm flexible tube, and a small acrylic vial attached to it to store 15mL of
water sample (Figures 7 and 8).
Figures 7 and 8 – Plastic case with electronic components, a flexible tube, and an acrylic vial make the sampling
device of prototype 1.
Laboratory tests were made to: (i) verify if the system was working properly; (ii) identify the
maximum suction range of the peristaltic pump (up to 40cm water column); and (iii) define 30cm as
4
Illustration by André D. L. Zanchetta made with open-source software Fritzing.

XXIII Brazilian Symposium on Water Resources (ISSN 2318-0358) 5
the height in the code
5
that activates the pump. After that, manual controlled flights were made over
an artificial pond that was relatively wide and had characteristics of lentic systems (Figures 9 and
10).
Figures 9 and 10 – Prototype 1 collecting a water sample from an artificial pond during a manual controlled
flight. Also available at <https://youtu.be/93jexnFGMC8>.
3.2 Prototype 2: hydrodynamic sampling
Prototype 2 was designed as a solution in which the water sampling device could be entirely
disposable and, at the same time, would be free of electrical components. To do so, the pumping
system of prototype 1 was replaced by a bailer
6
with 1cm in diameter and 60cm high. This
disposable tube was connected to the DJI Phantom 3 Professional by a 5m monofilament nylon line
and two carabiner clips (Figures 11 and 12).
Figures 11 and 12 – Disposable bailer connected to the drone by a nylon line and two carabiner clips during a
laboratory experiment in which prototype 2 collected water samples from a bucket.
Moreover, as the DJI Phantom 3 Professional has an attached camera that transmits real-time
images from the aircraft, prototype 2 was designed in a way that water samples could be collected
even when operating beyond the pilot's visual range (i.e., FPV flights). Finally, a Styrofoam floating
5
Code by André D. L. Zanchetta available at <https://tinyurl.com/y2m4deqw>.
6
Bailer is generally used in hydrogeology to retrieve groundwater samples and it is a hollow tube with a ball check valve insi de that seals its bottom
when it comes to pulling up a sample.
5m
30cm

XXIII Brazilian Symposium on Water Resources (ISSN 2318-0358) 6
device was attached to the upper portion of the bailer to prevent its complete submersion during the
sampling procedures and also to ensure samples of a predetermined volume of water.
Manual controlled flights using prototype 2 were conducted in a water reservoir. Although
there was no physical barrier between the drone and its operator, water samples were collected at a
point distant enough from the pilot to ensure that sampling procedures could also happen beyond
the operator's visual range (Figures 13 and 14). Thus, the experimental samplings with prototype 2
were performed based on real-time image transmission.
Figures 13 and 14
7
– Drone, camera, disposable bailer, Styrofoam floating device, remote controller, and tablet
during a preliminary test conducted within the pilot’s visual range. A previous test over an artificial pond without
the Styrofoam floating device is available at <https://youtu.be/qpZ4urMb-68>.
4. RESULTS AND DISCUSSION
Tests using prototype 1 achieved satisfactory results mainly due to its accomplishment as a
surface water sampler (Figure 15). However, cleaning or replacing its flexible tube would be an
obstacle to fieldwork routines. Additionally, manual controlled flights using prototype 1 were able
to collect samples only when DJI Phantom 1 was operated within the pilot's line of sight, once this
procedure was necessary to ensure that the drone would reach 30cm of vertical distance from the
pond to activate its pumping system.
On the other hand, samplings with prototype 2 in FPV mode were successful and resulted in a
150mL water sample collected in each manual controlled flight (sufficient for up to 15 turbidity
measurements in a portable turbidimeter). Besides, using a carabiner clip to connect a disposable
bailer to the 5m nylon line made it considerably fast to substitute a bailer filled with water for an
empty one, without having to land the aircraft.
7
Photos by Flavio S. J. de Freitas.
5m

XXIII Brazilian Symposium on Water Resources (ISSN 2318-0358) 7
Figure 15
8
– Acrylic vial partially filled with water collected by prototype 1 during a manual controlled flight
over an artificial pond. Inside the vial, there is also sediment pumped from the bottom of the pond.
After the completion of the experiments, prototypes 1 and 2 were compared to identify which
one had more potential to be easily adopted by water sampling professionals (Table 1). In general,
the performance of prototype 2 was superior in comparison to prototype 1. Regarding water
sampling, the volume of 15mL collected by prototype 1, although sufficient to measure turbidity,
was quite limited compared to the 150mL sample of prototype 2. Likewise, prototype 1 was
considerably complex in terms of customization and cleaning.
Table 1 – Comparison between prototypes 1 and 2.
Prototypes
Characteristics
1
2
Design
Are samples collected according to Standard Methods?
yes
yes
Vertical distance between the drone and the water surface
30cm
5m
Water sample volume collected in each flight
15mL
150mL
Does it collect samples at a pre-set depth?
yes
no
Does it collect integrated samples9?
no
yes
Is it possible to use a camera attached to the drone?
no
yes
Level of complexity of customization by non-designers
high
low
Operation
Level of complexity of operation conducted by sampling professionals
high
low
Can sampling be done beyond the pilot's visual range?
no
yes
Does its sampler have to be cleaned?
yes
no
8
Photo by André D. L. Zanchetta.
9
An integrated sample represents various points that may vary in composition across the depth of a water body.

XXIII Brazilian Symposium on Water Resources (ISSN 2318-0358) 8
Any change in sampling volumes or pump activation distance would require a code change or
even different components in the circuit. Besides, cleaning the pumping system is not practical
especially because the peristaltic pump had to be disassembled at the end of each collection so that
the flexible tube could be cleaned or replaced.
However, prototype 1 was able to collect samples from a specific depth, which tends to be
very useful in some situations (e.g., collecting a water sample under a thick layer of oil). On the
other hand, samples collected using prototype 2 can be considered integrated samples (1060B.1C
COLLECTION OF SAMPLES, 2017), once each of them consists of a mixture of the same amount
of water collected from different points along with the depth during a single bailer sampling.
Collecting integrated samples is important to analyze suspended materials that are not
homogeneously distributed throughout a water body.
Regarding the safety of both devices, prototype 2 stood out for allowing a 5m vertical
distance between the drone and the water surface, while in prototype 1 the same distance was up to
30 cm. One additional positive point in prototype 2 has to do with the camera attached to the drone.
This accessory showed its potential to optimize fieldwork because it allows operators to conduct
two activities during the same flight: collecting water samples and making photographic records of
the sampling campaigns. Finally, the option for a disposable sampler in prototype 2 tends to create
simpler and faster fieldwork routines and its overall design can be perceived by non-designers as a
user-friendly device and easy to customize.
5. CONCLUSIONS
Based on the results of laboratory and field experiments, it is reasonable to say that prototypes
1 and 2 have the potential to significantly improve surface water sampling routines. The simple fact
that both of them are solutions based on the use of small drones is already an advantage over
conventional sampling procedures. In fact, water sampling with drones reduces the need for boats or
bridges to access different sampling points and also allows samples to be collected according to
renowned procedures, such as those preconized by the Standard Methods for the Examination of
Water and Wastewater.
When compared to each other, prototype 2 stood out for: (i) its simplicity of operation and
maintenance; (ii) being able to collect up to 150mL of water sample on each flight; and (iii) having
a disposable bailer as its sampling device. Moreover, by using a camera that provides real-time

XXIII Brazilian Symposium on Water Resources (ISSN 2318-0358) 9
images, prototype 2 can keep a 5m vertical distance from the water surface throughout the sampling
procedure and the operation can happen even when the drone is beyond the pilot's visual range.
The results of this article also hint out a series of opportunities and challenges when it comes
to water sampling with drones. It would be opportune to use similar solutions to prototype 1 and 2
during, for instance, the construction of large infrastructure projects to assess the performance of
stormwater Best Management Practices (BMPs) such as sediment basins. An interesting aspect of
using these solutions in construction sites is that only two flights per monitoring campaign would be
enough to assess this type of BMP: one to collect a sample from the basin and the other one to do
the same at its outflow point. Besides, it could address procedures required by the Standard
Methods for the Examination of Water and Wastewater without using any boat.
Finally, the biggest challenge related to the results of this article is collecting water samples
greater than 150mL, which is the maximum capacity of prototype 2. If the volume exceeds this
amount, there are at least two ways to assure water sampling. One of them would involve drones
designed to carry loads compatible with the specifics of each sampling campaign. A second
alternative could involve the use of autonomous micro laboratories attached to drones to collect
water samples, analyze them, and share the results in real-time.
ACKNOWLEDGMENTS
To the Institute for Technological Research the São Paulo State (IPT), the Foundation for
Support of the Institute for Technological Research (FIPT), the São Paulo State Water Resources
Fund (FEHIDRO), professor Marcelo De Julio (in memoriam), professor Dione Morita (USP),
Alvaro Kopezynski, Antônio Catib, Benedito Nachbal, Kaique Justino, Marcelo Gramani, André
Zanchetta, and Ernesto Agostini.
REFERENCES
ANTTILA, S. Assessing temporal representativeness of water quality monitoring data. Journal of
Environment Monitoring, v. 14, n. 2, p. 589-595, 2012.
BANZI, M. Getting Started with Arduino. O'Reilly Media, Inc., 2011, accessed 13 May 2020.
Available at <https://tinyurl.com/ydc9wgjr>.
1060 COLLECTION AND PRESERVATION OF SAMPLES. Standard methods for the
examination of water and wastewater, 2017, DOI: 10.2105/SMWW.2882.009, accessed 13 May
2020. Available at <https://www.standardmethods.org/doi/abs/10.2105/SMWW.2882.009>.

XXIII Brazilian Symposium on Water Resources (ISSN 2318-0358) 10
1060B.1C COLLECTION OF SAMPLES. Standard methods for the examination of water and
wastewater, 2017, DOI: 10.2105/SMWW.2882.009.
GETTEL, M. Improving suspended sediment measurements by automatic samplers. Journal of
Environment Monitoring, v. 13, n. 10, p. 2703-2709, 2011.
ORE, J. P. Autonomous aerial water sampling. Journal of Field Robotics, v. 32, n. 8, p. 1095-1113,
2014.
SCHWARZBACH, M. Remote water sampling using flying robots. In: INTERNATIONAL
CONFERENCE ON UNMANNED AIRCRAFT SYSTEMS, 2014, Anchorage. Proceedings…
Anchorage: ICUAS, 2014.