Primary biological aerosol particles (PBAPs) are an abundant subset of atmospheric aerosol particles which comprise viruses, bacteria, fungal spores, pollen, and fragments such as plant and animal debris. The abundance and diversity of these particles remain poorly constrained, causing significant uncertainties for modelling scenarios and for understanding the potential implications of these particles in different environments. PBAP concentrations were studied at four different sites in the United Kingdom (Weybourne, Davidstow, Capel Dewi, and Chilbolton) using an ultraviolet light-induced fluorescence (UV-LIF) instrument, the Wideband Integrated Bioaerosol Spectrometer (WIBS), versions 3 and 4.
Using hierarchical agglomerative cluster (HAC) analysis, particles were statistically discriminated. Fluorescent particles and clusters were then analysed by comparing to laboratory data of known particle types, assessing their diurnal variation and examining their relationship to the meteorological variables temperature, relative humidity, wind speed, and wind direction. Using local land cover types, sources of the suspected fluorescent particles and clusters were then identified.
Most sites exhibited a wet discharged fungal spore dominance, with the exception of one site, Davidstow, which had higher concentrations of bacteria, suggested to result from the presence of a local dairy factory and farm. Differences were identified as to the sources of wet discharged fungal spores, with particles originating from arable and horticultural land at Chilbolton, and improved grassland areas at Weybourne. Total fluorescent particles at Capel Dewi were inferred to comprise two sources, with bacteria originating from the broadleaf and coniferous woodland and wet discharged fungal spores from nearby improved grassland areas, similar to Weybourne.
The use of the HAC method and a higher fluorescence threshold (9 standard deviations instead of 3) produced clusters which were considered to be biological following the complete analysis. More published data and information on the reaction of different speciated biological particle types to fluctuations in meteorological conditions, such as relative humidity and temperature, would aid particle type characterisation in studies such as this.
Primary biological aerosol particles (PBAPs), or simply bioaerosols, are a
subset of atmospheric aerosol particles produced from a range of sources
within the biosphere. The constituents of PBAP vary in size and abundance,
and include viruses (0.01–0.3
The abundance and presence of PBAP in different regions is impacted by local-scale
meteorology, including, but not limited to, temperature and relative humidity (RH;
The influence of temperature on both fungal spores and bacteria between differing soil
types illustrated that the optimum temperatures for bacterial growth are higher compared
to fungal spores
The diurnal patterns of biological particles differ and for bacteria these particles have
been found to increase at sunrise, decrease during solar noon hours, gradually increase
until sunset, then decrease into the evening, with lowest concentrations between 21:00
and 05:00 UTC (all times in this paper are UTC)
Temperature has also been found to be the main driver for controlling pollen release, and
for some species it is temperature, alongside precipitation, that controls the amount of
pollen that is produced
Increases in PBAP concentrations are not only related to meteorological conditions; and
instead the proximity of the site to different land cover types, alongside wind speed and
wind direction data, is to be considered in order to identify distinctive emission
patterns and factors. It has been found that in arable and agricultural areas, increases
in bioaerosol concentrations may result from the maturing of crops and tree foliage
Biological particles fluoresce when illuminated with ultraviolet light, owing to the
intrinsic presence of bio-fluorophores, such as nicotinamide adenine dinucleotide
phosphate
UV-LIF instrumentation allows for real-time measurements, providing instantaneous data
without the need for constant maintenance. This allows for continuous monitoring for
extended periods of time as opposed to traditional sampling techniques which, although
allow for accurate identification of particle type and species, are often laborious with
poor time resolution and may suffer from potential identification biases when manually
counting particles
There is a potential interference risk from non-biological fluorescent particles which
can possess similar fluorescence properties to PBAP, and can fluoresce at wavelengths
used by UV-LIF instruments. Chemical pollutants from vehicles, such as diesel
particulates and other secondary organic aerosols (SOA), are known to fluoresce upon
excitation wavelengths, especially in submicron ranges
During atmospheric transport, airborne particles are subject to aging, chemical and
physical transformations, and fragmentation affecting the fluorescent characteristics of
the particle
The use of HAC analysis to distinguish and statistically segregate different types of
biological particles is required as UV-LIF instruments do not provide information on
particle genus or species (described further in Sect. 2.3). The use of this method has
been applied for analysis of data from a Colorado pine forest
Recent work, and ongoing areas of research, has included emission modelling
for pollen particles as based on observed pollen counts within the United
States
This study reports the analysis of measurements taken at four different sites within the
United Kingdom, during different times of the year, using a UV-LIF instrument, the
Wideband Integrated Bioaerosol Spectrometer (WIBS). Using a HAC approach, different
clusters are statistically discriminated. The resulting fluorescent and cluster
concentrations between sites are analysed in relation to meteorological conditions,
focusing specifically on temperature and RH, and in relation to wind speed and direction.
HAC solutions and responses to meteorological drivers from each site are compared with
local land cover types to identify distinctive emission patterns and factors. This is the
first comparison study of measurements taken from four different sites in the United
Kingdom using a UV-LIF instrument, and attempts to classify particles from these sites
using the HAC method, meteorological data, and land cover mapping collectively and infer
emission types in association with different land cover types. Contrary to previous work,
this is additionally the first use of a differing fluorescent threshold of 9 standard
deviations (9 SD) compared to traditionally 3 SD, in an ambient setting, to reduce the
impact of interferents from potential anthropogenic sources, following
Bioaerosol measurements were conducted at four different sites within the United Kingdom
during different years and different times of the year (Table 1). These sites were
Chilbolton in Hampshire, Davidstow in North Cornwall, Weybourne on the coast of north
Norfolk, and Capel Dewi near Aberystwyth, Wales (Fig. 1). At each site, the instruments
were connected to a PM
To identify the land cover characteristics of each site, the Centre for Ecology and
Hydrology Land Cover Map 2015 (LCM2015) was used to provide background information at
each site
Here, using the LCM2015 at each site, a land cover class which is common is “improved grassland”, which is distinguished from semi-natural grasslands owing to its higher productivity and lack of winter senescence, whilst those defined as “arable and horticulture” comprise annual crops, perennial crops (e.g. orchards), and freshly ploughed land. Though the LCM2015 provides up-to-date data on the land cover characteristics of each site, it is acknowledged that there is a risk that smaller features are not identified and thereby not considered potential sources. To provide geographical context to each site, the LCM2015 has been overlaid on an Ordnance Survey (OS) base map. Additionally, the presence of local farming activity and other potential influencing sources are presented in Fig. 1.
Chilbolton is situated within Hampshire on the southern coast of England, and data
collection was conducted at Chilbolton Observatory situated at the edge of Chilbolton
village. Measurements were conducted from the 20 January to 20 March 2009, with the
instrument connected to the PM
Data collection from Davidstow, in North Cornwall, was conducted from the 25 June to 28 August 2013, in which the instrument was connected to a 10 m high sampling line. The land cover around the ground site predominantly comprises improved grassland coverage, with some urban and suburban land cover, arable and horticultural land, and heather covered areas. The Davidstow Airfield runway is located to the south-east from the ground site and in a north-westerly direction there is a dairy factory, the largest producers of cheddar cheese in Britain. The sampling site is in close proximity to a number of farming sites, including a dairy farm located closest to the site from the west. In addition, a small garden centre is located close to the dairy farm to the west (represented by the suburban land cover), and immediately south of the site there is a slaughter house.
Land cover maps at each site produced using the Centre for Ecology and Hydrology Land Cover Map 2015, OS data (© Crown Copyright and database right 2018), and the World Imagery ArcMap Layer in ArcGIS version 10.4.1.
Capel Dewi is situated near Aberystwyth in west Wales, and from the 18 February to 3 June 2013 data were collected from the Natural Environment Research Council (NERC) Mesosphere–Stratosphere–Troposphere (MST) radar site, at an inlet height of 3 m. The location of the site is within, and mostly surrounded by, an improved grassland area, with some broadleaf woodland to the north of the site and coniferous woodland predominantly to the south, and some arable and horticultural land close by, in addition to a livestock breeding farm located in a north-westerly direction from the site.
Unlike the other sites, Weybourne is a coastal site, located north of Norfolk, at the Weybourne Atmospheric Observatory. Data collection was conducted for a 1-week period from the 17 to 25 August 2009. The instrument was connected to a high-volume sampling line, with an inlet at a height of 10 m. The site is located in an improved grassland area, with gorse heath observed immediately inland from the Observatory. To the south-east of the Observatory there is arable and horticultural land, to the north of the site there is some littoral sediment, and to the south there is some broadleaf woodland.
Measurements were recorded at each site using a model 3 WIBS (WIBS-3;
The WIBS is able to detect particles ranging from 0.5 to 20
Once a particle has been sized, two xenon flash lamps are triggered at excitation wavelengths 280 and 370 nm to excite tryptophan and NADH fluorescence, respectively. Two photomultiplier tubes (PMTs) then record the fluorescence emitted from the particle. Emitted fluorescence is measured using three detector channels which record the fluorescence over two wavelength ranges: FL1 (fluorescence between 300 and 400 nm, once excited at 280 nm), FL2 (fluorescence between 410 and 650 nm, once excited at 280 nm), and FL3 (fluorescence between 410 and 650 nm, once excited at 370 nm).
The HAC method was used to distinguish and statistically segregate different biological
particle types, using the approach of
Data pre-processing, prior to clustering the data, is required to remove particles which
saturate the PMT, as their true fluorescence cannot be measured. This is also required to
remove particles smaller than 0.8
Each site was analysed as a whole dataset including fluorescent and non-fluorescent particles, with the exception of Capel Dewi. Here, due to the size of the dataset, fluorescent particles only were selected due to computer memory limitations. The size of each dataset differed, owing to different field campaign durations (Table 1).
The site name, geographic location (latitude, longitude), sampling period of each campaign, examples of land cover at each site, size of the data, number of particles clustered at each site, and the instrument used for the campaign presented.
The data were normalised using the z-score method (in which the mean is subtracted and
the data are divided by the SD) and the Ward linkage (in which clusters are merged by
finding the clusters which yield the minimum increase in total within-cluster variance
once merged
Meteorological data were recorded during each campaign, apart from Capel Dewi in which
data were obtained from the NERC MST Radar Site. Land cover data, as obtained from the
Centre for Ecology and Hydrology LCM2015, were analysed for each site using ArcGIS
(version 10.4.1). Additionally, using the R package “openair”, polar plots of wind
direction and wind speed were produced for each site to be used in relation to the
LCM2015
The average diurnal variation in total fluorescent and non-fluorescent particles were plotted for each site (Sect. 3.1), and following the HAC method (Sect. 3.2), the diurnal variation in each cluster was plotted (Sect. 3.3). The data were then considered in terms of meteorological influences (Sect. 3.4), focusing on temperature and RH (Sect. 3.4.1) and wind speed and wind direction (Sect. 3.4.2).
The average diurnal variation in fluorescent and non-fluorescent particle concentrations
differs between the four sites, potentially indicating the differing types of biological
particles present (Fig. 2). Weybourne exhibits increased fluorescent concentrations prior
to
Total fluorescent and non-fluorescent diurnal particle concentration variation
at each site (number of particles, N
The standard threshold for defining fluorescent and non-fluorescent particles is
calculated by using the instrument forced trigger (FT) measurement
HAC Ward linkage analysis at each site displaying the average fluorescent
intensities per channel (FL1, FL2, and FL3); the average optical diameter, D
(
The dominant cluster at Weybourne is Cluster 3, representing 50.6 % of total
fluorescent particle concentration, and exhibiting the greatest fluorescence signal in
channel FL3, similar to Cluster 1. Cluster 1 displays a similar fluorescence profile and
similar high abundance (33.7 % of total fluorescent particle concentration), but with
a much larger size and shape (5
Cluster 2 is the least abundant cluster and represents only 6.9 % of the total
fluorescent particle concentration. Fluorescence signal is greatest in channel FL1, with
some signal in channel FL3, and very low signal in channel FL2. The size and shape of
this cluster indicate a fairly small aspherical particle (3.1
Cluster 4 also accounts for a small proportion of fluorescent particles (8.9 % of the
total) but displays high fluorescence in channel FL3, with some fluorescence in FL2, and
less in FL1. Particle size is relatively large, and similar to the other clusters, the
shape value indicates an aspherical particle (4.4
Cluster analysis of the Chilbolton dataset similarly produced a four-cluster solution, of
which Cluster 2 accounts for 58.6 % of the fluorescent particle concentration.
Cluster 2 displays low fluorescence in channels FL1 and FL2, with higher fluorescence in
channel FL3. Not only is this fluorescence profile and cluster dominance similar to
Weybourne Cluster 3, but Chilbolton Cluster 2 also exhibits the smallest size and shape
(1.8
Chilbolton Cluster 1 displays a similar fluorescence profile to Chilbolton
Cluster 2. However, Chilbolton Cluster 1 is less abundant (23.2 %), and
exhibits the largest size and shape (3.6
Chilbolton Cluster 3 is the smallest cluster (representing 5.6 % of the total
fluorescent population) but exhibits the greatest fluorescence in channel FL1 compared to
the other clusters, and exhibits a similar fluorescent profile to Weybourne Cluster 2.
Both Weybourne Cluster 2 and Chilbolton Cluster 3 make up the smallest cluster group of
each dataset, with the same size (3.1
Cluster 4 from the Chilbolton dataset displays similar fluorescent profile
characteristics to Weybourne Cluster 4, in which there is a high fluorescence signal in
channel FL3, with some in FL2 and less so in FL1. Chilbolton Cluster 4 represents
12.6 % of the total fluorescent particle population, and has a size of
2.4
The fluorescent profiles at Chilbolton show much similarity to Weybourne, even though Weybourne is a coastal site. This suggests that the sources of these clusters originate from the same land cover type, which for both sites is either the improved grassland or arable and horticultural land. Using the fluorescent profiles for each cluster, the clusters at Chilbolton are initially assumed to comprise fungal spores (Cluster 1 and Cluster 2), bacterial or fungal spores (Cluster 3), and a pollen fragment (Cluster 4).
Cluster analysis of the MST Capel Dewi dataset produced a five-cluster solution, of which
Cluster 5 accounts for 61.5 % of the total fluorescent cluster population. Cluster 5
exhibits relatively low fluorescent channel signals in comparison to other clusters, with
the exception of Cluster 4 which presents a similar fluorescent profile and represents
18.7 % of the total fluorescent particle population, the second largest cluster.
Incidentally, Cluster 5 represents the smallest size (1.6
Cluster 3 exhibits a large fluorescence signal in channel FL1, with only some in FL2, and
less in channel FL3. The larger fluorescence in channel FL1 is similar to Chilbolton
Cluster 3 and Weybourne Cluster 2. Though these two sites have similar signals in
channels FL2 and FL3, Capel Dewi Cluster 3 signals decrease in channel FL2 and even more
so in channel FL3. However, comparing the size range of the particles between these
sites, they are remarkably similar, with a size range of 3.1
Cluster 2 represents 8.7 % of the total fluorescent population, and
displays a high fluorescence signal in channel FL2, with some signal in
channel FL1 and FL2. This cluster shows some similarities to Cluster 1, the
smallest cluster (0.8 % of the total fluorescent cluster population).
Though Cluster 1 displays a higher fluorescent signal than Cluster 2, a
similar fluorescence signal profile is noticeable. The sizes of these two
clusters differ, with Cluster 2 reasonably small (1.8
The cluster types at Capel Dewi are suggested to comprise fungal spores (Cluster 4 and Cluster 5), and it is inferred that clusters 1 and 2 are interferents given the larger signal in channel FL2. Cluster 3 is assumed to be a fungal spore or bacterial particle, which shows some slight similarity to Chilbolton Cluster 3 and Weybourne Cluster 2 in terms of the high FL1 channel.
Similar to the Capel Dewi dataset, cluster analysis of the Davidstow data produced a
five-cluster solution. Of which, Cluster 5 represents 44.9 % of the total fluorescent
particle population, but displays a very low fluorescent signal and small size and shape
(1.6
Again, similarities between Davidstow Cluster 4 and Capel Dewi Cluster 3 can be seen,
though additionally with some similarities to Weybourne Cluster 2 and Chilbolton Cluster
3 in terms of the high fluorescence signal in channel FL1. Davidstow Cluster 4 and Capel
Dewi Cluster 3 display a very similar fluorescence signal profile, albeit at different
fluorescent intensities. Davidstow Cluster 4 represents 24.5 % of the total
fluorescent particle population, with an average size and shape (4
Davidstow Cluster 2 displays a similar fluorescent profile to Capel Dewi Cluster 2, and
represents a small proportion of the fluorescent particle population (4.3 %) similar
to Capel Dewi Cluster 2 (8.7 %). However, there are differences in size and shape, as
Davidstow Cluster 2 has a larger average particle size and asymmetry factor compared to
Capel Dewi Cluster 2. It can be seen that Davidstow Cluster 2 also displays some
similarity to Capel Dewi Cluster 1, though at lower fluorescence signal intensities, but
similar in terms of average particle size and asymmetry factor (Capel Dewi Cluster 1, 5.3
Davidstow Cluster 1 is a remarkable cluster, owing to its large size (16.5
Though the HAC method proved successful in segregating the fluorescent particle data into distinctive clusters, this method alone does not allow for easy particle identification, as demonstrated during the first attempt to assign particle type to each cluster. For ease of analysis, a summary table of the similar clusters is presented (Table 3), identifying the clusters which share a similar fluorescent profile between the four different sites, and were speculated to belong to the same particle type.
Summary table of similar clusters (CL) between sites with the inferred cluster grouping and type.
The average diurnal variation in each cluster at each site was plotted to identify any diurnal patterns which influence cluster concentration at each hour of the day (Fig. 3). These have been collated together according to the suggested groupings made based on the HAC analysis (Table 3), rather than at each site, in order to assess the ability to characterise particles based on the fluorescent profiles, particle size and shape, and percentage contribution to the overall fluorescent particle population as in Sect 3.2.
Average cluster diurnal variation at each site (
The stability in diurnal concentrations is also experienced for Capel Dewi Cluster 4 and Cluster 5, in which both clusters show small diurnal variation and remain relatively static. One exception to this is that Cluster 5 concentrations increase after 16:00, whereas Cluster 4 very gradually increases in concentration throughout the day, and then decreases after 21:00.
It can be concluded that these clusters have been accurately collected together and can be assumed to be the same type of particle. However, the behaviour of these clusters do not reflect typical fungal spore behaviour, and instead show some characteristics of bacterial particles given the lack of diurnal variation.
The diurnal profiles of these two clusters do not match, even though the fluorescent profiles are similar. However, It is suggested that these clusters are both fungal spores, and it is proposed that, due to the size of the Capel Dewi dataset, the diurnal variation for this cluster is not as clear.
The trend is similar for both Chilbolton and Weybourne Cluster 4, in which there is a drop in concentrations during the morning, with lower concentrations during the day, which then increase in the evening. This illustrates that these are fungal spores, unlike Davidstow Cluster 1, which shows more variation during the day. Considering this, it is possible that only Davidstow Cluster 1 is a pollen fragment, and Chilbolton and Davidstow Cluster 4 are both fungal spores. This potentially illustrates that the dominance of channel FL2, when accompanied by a high signal in channel FL3, is what separates pollen from fungal spores.
The diurnal variation in Capel Dewi Cluster 1 and Cluster 2 is reminiscent of other clusters which have been suggested to be fungal spores, owing to the lower concentrations throughout the day, the decrease in concentrations in the morning, and subsequent increase in concentrations in the early evening. The identity of Davidstow Cluster 2 is uncertain, owing to the lack of fluctuations throughout the day, and a fairly static concentration profile; therefore, further analysis with the meteorological data is required.
The influence of local scale meteorology, in particular the relationships
between cluster abundance and temperature (
Group 6 was considered to be non-biological interferent particles following the
HAC analysis, following
Total fluorescent particle concentration at each site for each
campaign period (number of particles, N
In the previous section, the data were split according to the fluorescent signal profiles, using HAC analysis to initially assign a cluster class, and cluster diurnal time series and meteorological data to confirm or contradict these assumptions (Table 4).
Following this, the relationship between total fluorescent particles and
clusters at each site were analysed as a function of wind speed (m s
Suspected particle class types at each site showing the inferred cluster types following fluorescence signal analysis, and the suggested particle types following the cluster diurnal time series, and meteorological data analysis (temperature and RH). (Note: Weybourne and Chilbolton comprise 4 clusters, hence unassigned Cluster 5). WD represents wet discharged.
The sources of the total fluorescent particles at Weybourne appear to originate from a
south-south-west (SSW) direction at low wind speeds (2–4 m s
Weybourne Cluster 1 polar plot presenting two potential sources of biological particles, which when using Fig. 1 are likely to comprise improved grassland and arable and horticultural land.
Total fluorescent particle concentrations at Chilbolton originate from a local source at
low wind speeds, with lower concentrations of particles originating from a
west-south-west (WSW) direction at wind speeds of
The source of fluorescent particles at Davidstow originate from a WSW wind direction, at
wind speeds from
The source of total fluorescent particles at Capel Dewi mostly originate locally, with
some contribution from a north-north-westerly (NNW) wind direction at wind speeds of
10–14 m s
The data from each site were initially analysed using the original
recommendation of a baseline FT
Sensitivity of the HAC threshold using 3 SD and 9 SD for analysis of the Chilbolton dataset.
When adopting the 9 SD threshold, potential interferents were removed, which represents a more robust solution for biological particle detection, although low activity within aged biological particles may mean these are removed as well.
Hierachical agglomerative cluster (HAC) analysis was used to characterise biological
particles from three sites in a rural environment (Davidstow, Chilbolton, and Capel Dewi)
and one site on the coast of England (Weybourne). The use of forced trigger
(FT)
Often, clusters which displayed a similar fluorescence profile were segregated due to differences in particle size and shape. Though most clusters could be attributed to a certain biological particle type, Weybourne Cluster 1 was unique, and was inferred to be a mixture of two different particles, owing to different meteorological activity compared to other clusters and origin from potentially two sources as identified using polar plots in association with the LCM2015.
When using fluorescent signal intensities for particle identification, this would, for the most part, provide incorrect particle identification (Table 4). The use of meteorological data allowed for the initial assumptions made to be either discredited or improved upon. For the clusters which were considered to be interferents, owing to the high signal in channel FL2, meteorological data analysis instead indicated clusters of biological origin (Table 4).
Clusters that were considered to be pollen fragments (Group 5) were accompanied by high (Davidstow Cluster 1) to moderate (Chilbolton and Davidstow Cluster 4) signal in channel FL2. It was speculated that the intensity of channel FL2, when accompanied by a high signal in channel FL3, is what separates pollen from fungal spores. However, following meteorological data analysis, Group 5 was inferred to comprise wet discharged fungal spores (Chilbolton and Weybourne Cluster 4) and bacteria (Davidstow Cluster 1). Though the original identity inferred using the fluorescent profiles was incorrect, the dominance of FL2 may have provided some distinction between bacteria and fungal spores.
For Weybourne, Chilbolton, and Capel Dewi the most common cluster class was wet discharged fungal spores, which comprised almost all of the clusters at each site. On the contrary, Davidstow, a site comprising mostly improved grassland and some arable and horticultural land, similar to the other sites, has mostly bacterial clusters. Emissions from the dairy factory located close to the site are likely to have influenced the experienced concentrations at Davidstow, in addition to the main source of Cluster 2, Cluster 3, and Cluster 5, which originated from the grassland and suburban land where there is a local dairy farm and small garden centre.
Using land cover maps and associated polar plots proved beneficial for identifying potential sources of clusters and in some cases helped characterise clusters (e.g. Weybourne Cluster 1). However, certain cluster types could not be clearly attributed to a specific land cover type. Overall, more wet discharged fungal spores appeared to be associated with improved grassland areas. More variety was found for bacterial clusters, ranging from broadleaf and coniferous woodland (Capel Dewi Cluster 4) to a mixture of suburban, arable, and improved grassland (Davidstow Cluster 2, Cluster 3, and Cluster 5), which is likely to be influenced by the presence of the local dairy farm and small garden centre at Davidstow.
To our knowledge this is the first use of ArcGIS land cover mapping in association with airborne bioaerosol concentrations collected using a UV-LIF spectrometer to identify distinctive emission patterns and factors. This analysis relied upon the use of the HAC method; and as one of multiple unsupervised clustering approaches, it is possible that this analysis is subject to various levels of misclassification depending on the instrument used. Whilst the ancillary ambient data included in this study supports some of the derived cluster variability, it is not possible to comment further on this issue without additional laboratory data on known bioaerosol types. This is the subject of ongoing work and will be assessed in the near future, with the discussion of results from the referenced “Dstl experiment 2017” to be published in 2019.
For future studies, more published data and information on the reaction of different speciated biological particle types to fluctuations in meteorological conditions, such as RH and temperature, would aid particle type characterisation in studies such as this.
Owing to the variable size and format of the data used in this study, the data is only available by contacting the lead author.
Laboratory data from a WIBS-3D were collected during a series of
characterisation studies at the Defence Science, and Technology Laboratory
(Dstl). This data included bacteria comprising unwashed E. coli and
The supplement related to this article is available online at:
EF wrote the paper and performed the analysis; DT and MG provided advice, supervision, and feedback throughout the drafting and submission process; MG additionally managed the ambient dataset campaigns; laboratory data were collected at the Defence Science and Technology Laboratory, which were provided by VF who also proof-read the manuscript and provided feedback; IC provided software training and support; PK, WS, and RSE proof-read the manuscript and provided constructive feedback.
The authors declare that they have no conflict of interest.
Elizabeth Forde is funded under the Dstl (Defence Science and Technology Laboratory) and DGA (Direction Générale de l'Armement) Anglo–French PhD scheme and affiliated to the NERC EAO Doctoral Training Partnership (ORCID ID: 0000-0003-0719-3258). Data from Chilbolton were collected during the NERC APPRAISE programme (Aerosol Properties, PRocesses And Influences on the Earth's Climate), and from Davidstow during the COnvective Precipitation Experiment (COPE). We would like to thank Professor Geraint Vaughan for the measurements conducted at the NERC MST site. Edited by: Daniel J. Cziczo Reviewed by: two anonymous referees