The application of a new particle collection system (PCS) developed in-house
and operated on board a commercially available multicopter unmanned aerial vehicle
(UAV) is presented as a new unmanned aerial system (UAS) approach for in situ
measurement of the concentration of aerosol particles such as pollen grains
and spores in the atmospheric boundary layer (ABL). A newly developed
impactor is used for high-efficiency particle extraction on board the
multicopter UAV. An airflow volume of 0.2
More than 30 aerosol particle collection flights were carried out near
Tübingen in March 2017 at altitudes of up to 300
In situ measurements of the concentration of aerosol particles such as
pollen, spores, and fine particulate matter in the atmospheric boundary layer
(ABL) are of great interest in numerous scientific disciplines
For example, in agricultural science, the concentration and aerial dispersal
of pollen and spores are of interest with regard to an optimization of yield
In meteorology, it is known that mineral dust particles that originated from
Saharan dust storms and were transported, for example, to southern Florida
effectively act as ice nuclei capable of glaciating supercooled
altocumulus clouds
In human medicine, the careful scientific evaluation of the actual
concentration of pollen in the air is an indispensable basis for reliable
pollen risk information. Inadequate forecasts concerning the expected pollen
concentration are regarded as a considerable health risk for pollen allergy
sufferers
In environmental sciences, the pollution of air with fine particulate matter
has been a problem for many years, in particular in urban areas with
unfavourable geographical topography. The PM
For most of these applications, it would be highly desirable not only to
count the number or measure the size of the particles as done with an optical
particle counter (OPC), but also to identify the particles according to their
type and/or chemical composition. In this regard, particle collection with
subsequent particle-type identification and quantification has an advantage
over particle counting, at least as long as reliable in situ particle
identification is not available. First attempts to collect bioaerosol
particles using a pollen trap mounted on a fixed-wing UAV are described in
Here we present the structural design and first application of a new
particle collection system (PCS) developed in-house and operated on board a
commercially available multicopter UAV (Fig.
Multicopter UAV (DJI S900) in hovering flight with components of the particle collection system as indicated: air inlet, impactor, mass flow sensor, and blower. The inlet is arranged about 30 cm above the propeller plane.
An essential part of the present study was the development of a new PCS that
can be operated on board the multicopter UAV despite the weight and power
constraints. One major goal in the development of the PCS was to sample an
air volume of 1
In order to determine the capability of the PCS operated on board the multicopter UAV and to test the reliability of the entire new unmanned aerial system (UAS), several test flights were conducted at different altitudes over several days in March 2017. The collected particles were analysed and counted using light microscopy. Finally, the pollen concentration values determined with the PCS on board the multicopter UAV were compared with corresponding data published by forecast information services such as the Stiftung Deutscher Polleninformationsdienst (PID) or MeteoSwiss.
A DJI S900 hexacopter, commercially available from the Chinese company DJI
Technology Co. Ltd, was selected as multicopter UAV with regard to flight
performance, payload capabilities, and expansion options. The DJI S900 has a
diagonal wheelbase of 900
A DJI A2 flight control system was employed to automatically control the flight attitude, i.e. roll, pitch, and yaw angles as well as the flight altitude, and to maintain the spatial position of the multicopter UAV using a GPS receiver. A remote control of the type T14SG (2.4 GHz band, 14 control channels) by Futaba Corporation was chosen due to its high reliability over long distances. Telemetry data such as battery parameters (voltage, current, and capacity) and the barometrically determined flight altitude above ground level were retransmitted from the remote-control receiver on board the multicopter UAV to the handheld transmitter on the ground.
The DJI S900 was operated with a 6-cell Lithium polymer battery (LiPo, 22.2
A new PCS was developed in order to meet the requirements for aerial use
on board a multicopter UAV. To ensure a number of at least 10 collected
particles, even in the case of a particle concentration in the sampled air
being as low as 5 particles per
Starting from these boundary conditions, an impactor-based PCS was developed
(Fig.
Newly developed particle collection system (PCS) with a complete weight of 600
The geometry and orientation of the air inlet must be chosen in such a way
that the sampled air is representative in terms of its particle load, which
can be achieved by isokinetic sampling
In order to provide omnidirectional air intake under isokinetic or at
least near-isokinetic conditions, a bell mouth was chosen, with a wide end for the air inlet and a narrow end for the
connection to the subsequent particle extraction unit (Fig.
Operation on a multicopter UAV requires a particle extraction unit that has a
low mass and provides a high particle extraction rate, even at large airflow volumes (0.2
The functional principle of an impactor is based on the deflection of a
particle-loaded free-flow gas stream by means of an impaction plate
In order to sample an air volume of 2
Schematic longitudinal cross section through the impactor used as a particle
extractor in the particle collection system. Particles are drawn through the pipe from the
top towards the glycerine gelatine-covered microscope slide. Glycerine gelatine is highlighted
in green, cross section of silicone O ring in red. Mean impaction velocity is about 50
The particle sample carrier is
A reliable determination of the concentration of aerosol particles requires
the precise determination of the sampled air volume. This was achieved by
installing a mass flow sensor that permanently remains in the airflow path
of the PCS, irrespective of whether data from the flow sensor were collected
or not. A SFM 3000-200-C mass flow sensor of the Swiss company Sensirion AG was
used for this purpose. This sensor offers a bidirectional measuring span of
The electrically operated blower must ensure a high airflow volume through the
PCS during flight operations and the associated power and mass limitations.
It is also necessary that the blower performance is substantially independent
of fluctuations of the battery voltage in order to provide a constant airflow volume through the PCS. A blower that meets these demands is
commercially available in handheld vacuum cleaners of the British company Dyson
Ltd. The blower that we used in the PCS has a total weight of 245
An individual particle sample carrier was used for each particle collection
operation (Fig.
The sample carriers were produced in batches, usually a few days prior to the
scheduled particle sampling operation, while the production date of the batches
is being recorded. Production, handling, and storage of the sample carriers were
performed in a portable laminar airflow box under continuous flow of
filtered air. The air was filtered by two pre-filters and finally a
H14-specified HEPA (high-efficiency particulate air) filter removing more
than 99.995 % of the particles in the most critical size range of 0.1 to 0.3
Careful post-sampling treatment is highly necessary to avoid contamination
and allow preservation. Immediately after landing the multicopter UAV, the
particle-loaded sample carrier was carefully removed from the impactor and
placed into its transport box (Fig.
When using a multicopter UAV for aerosol particle collection, the position of
the air intake of the PCS has to be considered. It also
needs to be considered how the air intake should be aligned in relation to
the airflow generated by the propellers of the multicopter UAV in order to
avoid an impairment of the measurement results and to ensure a substantial
isokinetic sampling.
In order to investigate the actual airflow around the multicopter UAV used in
this study under ambient conditions with side wind, a visual airflow test
was performed in January 2017 at the airfield in Poltringen, Germany
(48.54322
Investigation of the airflow pattern caused by the multicopter UAV (DJI S900)
using three coloured pyrotechnical smoke cartridges with
Figure
Figure
In order to examine the effectiveness of the newly developed PCS with respect
to its particle extraction rate, an experiment was carried out using two
identical impactors connected in a cascade (Fig.
Schematic sketch of the extraction efficiency experiment with two identical impactors (impactor 1 and impactor 2) connected in a cascade configuration to investigate particle extraction efficiency. At 100 % efficiency, all particles would be extracted by impactor 1, leaving no particles for impactor 2.
Upon analysing the sample carrier using an optical microscope, it cannot be distinguished whether the particles on the sample carrier were collected during the airborne particle collection operation or inadvertently by contamination before or after the sampling operation. By using a laminar airflow box as previously described, contamination during manufacture and storage can be reliably prevented. And with the experiments described in the following, it was examined whether and, if so, what number of particles were inadvertently applied to the sample carrier by the handling of the sample carrier on the ground at the site of operation as well as during the ascent and descent of the multicopter UAV.
At the site of operation, the particle sample carrier is exposed to atmospheric air during installation in and removal from the impactor. This exposure usually lasts less than 30 s, but could lead to a contamination of the sample carrier with particles, in particular if the particle concentration in the ambient air is exceptionally high. In a first investigation carried out in the afternoon (14:15 to 14:30 local time) of 10 March 2017 at the airfield in Poltringen, a sample carrier was removed from its protective packaging and exposed to ambient air for 15 min on the roof of a car about 1.8 m a.g.l. The sample carrier was then repackaged and transported to the laboratory where it was treated and sealed in a particle-free laminar airflow box to prevent any further contamination. The results are discussed in Sect. 4.3.
As observed during the smoke plume tests, an inflow of air into the air
inlet of the PCS appears during the hovering flight of the multicopter UAV,
even if the blower of the particle collection system is switched off. It is
expected that this inflow incorporates aerosol particles onto the sample
carrier and thus has to be regarded as a potential source of contamination.
During vertical ascent of the multicopter UAV with a typical speed of 6
Numerous aerosol particle collection flights were carried out in March 2017
to evaluate the scientific potential of a multicopter UAV equipped with the
newly developed PCS. The major aim of developing such a PCS was the
collection of aerosol particles at different altitudes and their quantitative
determination. For the present study we focused at first on the quantitative
determination of the concentration of pollen grains. The airfield in
Poltringen near Tübingen in Germany was chosen as the launch site with regard
to an existing official flight permit for UAV flights up to an altitude of
300
Three series of aerosol particles collection flights were carried out on
3, 10, and 16 March 2017, at the airfield in Poltringen with three flights
each day. Table 1 gives an overview of these aerosol particle collection
flights, including data concerning the hovering altitude above ground level,
at which the blower of the particle collection system was activated, the
airborne particle collection start time, as well as the measured air
temperature, wind direction, and wind speed on the ground. On 3 March 2017, the
blower of the PCS was activated during hovering in 25, 100,
and 200
Prior to each day of aerosol particle collection flights, the
bell-mouth-shaped air inlet, the tube leading to the impactor, the O ring and
the two housing halves were cleaned in an ultrasonic bath with soapy water
for 15
In the field again, shortly before the particle collection operation, the
impactor was taken out of the sealed storage bag, the sample carrier was
installed in the impactor and the inlet was plugged onto the tube leading to
the impactor. In between the sampling flights, shortly before the next flight
operation and shortly before installation of the next unloaded sample
carrier, the impactor and the bell-mouth-shaped inlet were flushed with
filtered air using an battery-operated electric blower with a medical
ventilation filter installed on its inlet (type Pall Ultipor 100,
The sample carriers were treated post-flight as described previously.
Identification and counting of the collected particles were visually
performed using a Olympus transmitted light microscope BX50 at 400 times
magnification. The entire area of the slides was counted row by row.
Identification was assisted by a reference collection and literature
Aerosol particle collection flights carried out on 3, 10, and 16 March 2017 performed at different hovering altitudes.
The smoke plume tests allow a quantitative determination of the airflow
velocities. Despite their limited resolution, the results obtained here are
in good agreement with the CFD calculations reported by
With regard to the isokinetic sampling conditions concerning the
direction of the airflow velocity vectors, it was observed that a
plume of smoke approaching horizontally (due to prevailing side wind) 50
With regard to the isokinetic sampling conditions concerning the
magnitude of the velocity vectors, successive frames of the video
sequences recorded during the visual airflow tests were evaluated. A
horizontally approaching smoke plume begins to deflect in a vertical
direction. Within three frames of the recorded video sequences, corresponding
to 0.12
The circular opening of the free (wider) end of the bell-mouth-shaped
air inlet has an inner diameter of 69
As a result, positioning the air inlet of the PCS 30
The extraction efficiency of the impactor was determined by visual analysis
of sample carriers of two identical impactors connected in a cascade and
filled with the same airflow as shown schematically in Fig.
Number of pollen grains collected in impactor 1 and
impactor 2 of the arrangement of Fig.
The particle extraction and retention capability of the newly developed PCS
was demonstrated for pollen of the genera
With regard to the question of whether this high extraction and retention rate
also applies to other particles, i.e. to smaller particles, it should be
noted that in the widely used Burkard pollen trap a mean jet velocity of 6
The PCS, the visual identification, and the counting of particles are subject
to various influences, which potentially form a source of errors with regard
to the determination of the actual concentration of particles in the ambient
air. An overview of these influences in the different components of the PCS,
namely air inlet, impactor, and mass flow sensor, is given in Fig.
Overview of the possible influences of the different components of the newly developed particle collection system (PCS) on the final determined particle concentration. The components of the PCS, in which the influences can occur, namely air inlet, impactor, and mass flow sensor, are arranged along the horizontal axis. Influences that can lead to the determination of a particle concentration higher than the actual particle concentration are shown in the upper half of the figure (blue background), whereas the influences that can lead to the determination of a particle concentration lower than the actual particle concentration are shown in the lower half of the figure (red background).
The first source of measurement error might occur during the air intake. If
the ambient air is not drawn in under isokinetic conditions, i.e. with the
same velocity (by magnitude and direction) as the air approaching the air
inlet, then the drawn-in air might be enriched or depleted with particles
due to mass inertia effects. The multicopter UAV airflow tests have shown
that the suitable placement and design of the bell-mouth-shaped air inlet,
in combination with the operation of the PCS on board the multicopter UAV in
hovering flight mode, result in almost isokinetic sampling conditions provided
there are no excessive side winds. In order to be able to give an estimate of
the error caused by non-100 % ideal isokinetic sampling, further
investigations are required. A loss of particles, which have already been
drawn in, could occur due to adhesion to the wall of the air inlet as well
as to the wall of the downstream connecting pipes (“wall losses”, Fig.
Particle contamination is a potential error source that leads to higher
particle numbers deposited on the sample carrier. Within the present study,
experiments were
performed concerning potential contamination on the ground as well as particle
contamination during ascent and descent of the multicopter UAV. Concerning the potential particle contamination on the ground, a
total of four pollen grains were identified on the sample carrier, i.e. 2 of the
genus
For the evaluation of these results, the concentration of the pollen grains
in the ambient air must be taken into account. The contamination experiments
were carried out on 10 March 2017 at the same time as the aerosol particle
collection flights. The mean values of the concentrations measured at the
three altitudes (25
More relevant is the contamination of the particle sample carrier during
ascent and descent of the multicopter UAV. During the corresponding
contamination experiment, 450
During the visual identification and counting of the particles, it is
possible that contrast differences when using the transmitted light
microscope are erroneously identified as particles (false positives) and/or
that some particles are counted twice. Furthermore, it is possible that some
particles are not or not correctly identified (false negative) and/or that
some particles are overlooked. This potential source of error was excluded in
the present study by entrusting particularly experienced scientists with the
visual identification and counting of the particles, which still is the
golden standard for pollen concentration measurement
Finally, a potential error source exists with regard to the accuracy of the
mass flow sensor SFM3000-200-C. It is evident that any difference between the
actual and measured air mass flow produces a corresponding error in the
determined particle concentration. According to the data sheet of the mass
flow sensor, within the temperature range of
The number of particles collected during the aerosol particle collection
flights on 3, 10, and 16 March 2017 from 2
Summary of the number of collected particles (from 2
Only pollen of the genera
The amount of collected pollen grains, fungal spores, and large (
Graphical representation of the measured concentrations of particles
(in particles per
Only the numbers of the pollen of the genera
For all sampling altitudes, the concentration of pollen of the genus
For many of the pollen genera collected during the particle collection
flights in March 2017, the pollen grain concentrations measured at altitudes
of 100
Concentrations of pollen of the genera
During the measuring flights on 10 and 16 March 2017, the concentration of
pollen of the genus
The observation that the pollen grain concentration was higher at elevation
than on the ground is in good agreement with the results of
During the measuring flights on 10 March 2017 for both the pollen of the
genera
During the measuring flights on 3 March 2017, in addition to the aerial
sampling at various altitudes, one sample was taken on the ground with the
propellers of the multicopter UAV switched off and only the blower of the PCS
being activated. The concentrations of the most frequently occurring pollen
of the genera
The Stiftung Deutscher Polleninformationsdienst (PID) publishes and stores online weekly forecasts
(
The allergy centre of Switzerland (Allergiezentrum Schweiz) provides
not only online forecast information on expected pollen concentration, but also on the actual daily
pollen concentration
(
The presented multicopter-based UAS with the newly developed impactor-based
particle collection system (PCS) operated in-flight and on board the
multicopter UAV has proven to be a powerful and reliable system for aerosol
particle collection in the ABL. More than 30 particle collection flights
were carried out with this new UAS, each with a sampled air volume of 2
A particle separation efficiency of more than 98 % was determined for the
newly developed impactor-based PCS despite the high airflow volume of 0.2
Subject to a sufficiently high concentration of the corresponding particles
in the air, the number of in-flight collected particles was regularly well
above 100 during a 10 min sampling operation. These large numbers
of collected particles provide the possibility of reducing the volume of
sampled air and thus reducing the aerial sampling period. Accordingly,
particle collection flights at altitudes of up to 500
The particle collection flights carried out during the pollen season in March 2017 at altitudes of 25
The sample carriers of the particle collection operations are available for examination by interested parties.
The authors declare that they have no conflict of interest.
This research was financed through institutional funding by Eberhard Karls Universität Tübingen. Martin Schön supported us in the manufacture of the bell-mouth-shaped air inlet using FDM (fused depositing modelling) three-dimensional printing technology. In addition, we would like to express our gratitude to Barbara Maier, Simone Schafflick, and Wolfgang Kürner for their highly appreciated assistance in the implementation of numerous ideas into functional designs. In the realization of some technical solutions, we experienced kind support from companies established in the relevant technical fields; therefore we would like to express many thanks to Helmut Memmel and Alexander Post of the Daldrop + Dr. Ing. Huber GmbH & Co. KG in 72666 Neckartailfingen, Germany for the provided insights into cleanroom technology and the generous provision of high efficiency air filter elements, Manuel Meier of the Sensirion AG in 8712 Stäfa, Switzerland for a kind introduction in air mass flow measuring and the provision of sensor samples, Julia Ganter of the ebm-Papst GmbH & Co. KG in 78112 St. Georgen, Germany and Ernst Scherrer of the Micronel AG in 8317 Tagelswangen, Switzerland for the provision of miniature blower samples, and Jörg Haus and Rouven Möller of the Helmut Hund GmbH in 35580 Wetzlar, Germany as well as Matthias Werchan of Stiftung Deutscher Polleninformationsdienst for the provided insights into pollen and spores measurement and identification. Finally, we would also like to thank the editor and the anonymous reviewer for their kind and highly qualified comments, which have improved the quality of our manuscript. Edited by: Francis Pope Reviewed by: one anonymous referee