Application of Ultralight Aircraft for Aerosol Measurement Within and Above the Planetary Boundary Layer Above the City of Copenhagen

Abstract

An ultralight aircraft was equipped with atmospheric monitoring instruments and flown above Copenhagen on the 17 June 2022 to measure a range of aerosol parameters and meteorology. Three flights were carried out from sunrise to early afternoon with the aim to capture the boundary layer structure and evolution due to surface warming, emissions from the city, and atmospheric mixing. The data show clear evidence of the boundary layer which expanded from 400–600 m in height at around 07:30 to 1200–1400 m by around 14:30. Additionally, a residual boundary layer was observed in the early morning, and an entrainment of pollution at the top of the boundary layer in the early afternoon. The observed atmospheric features were consistent between monitoring instruments and meteorological sensors, supporting the reliability of the data and aircraft setup. These results demonstrate the merits and limitations of the use of small aircraft for scientific research and monitoring of aerosols in the vertical dimension, especially in densely populated areas and high-traffic airspaces.

1. Introduction

Atmospheric aerosols have significant and complex effects on global climate forcing and regional air quality, and it is vital that we understand the consequences of anthropogenic aerosols, which are produced by a range of human activities [1,2,3,4]. Aerosols can affect the radiative balance of the Earth by directly scattering or absorbing light. This aerosol–radiation interaction is expected to have a net global cooling effect, but locally, it is dependent on many factors including the wavelength dependence of scattering and absorption coefficients and the vertical distribution of the aerosol, leading to large uncertainties [3,5]. For example, it is expected that a significant fraction of the uncertainty of black carbon (BC) radiative forcing is due to uncertainties in its vertical profile, particularly in relation to low clouds [6,7]. The indirect aerosol–cloud interaction is responsible for the greatest uncertainty in aerosol forcing, and it is also highly dependent on the vertical profile of aerosols [8,9]. Additionally, it is well established that air pollution, especially the particulates, affect human health negatively [10]. A better understanding of aerosol evolution, transport and vertical mixing is therefore critical to improve model predictions over the atmospheric column and at the surface, which will allow more informed estimates of aerosol climate forcing, regional air quality and human health impacts.

The core aim of this study was to test the applicability of ultralight aircraft for the investigation of planetary boundary layer (PBL) properties and dynamics. To achieve this, a day was chosen for which the PBL was forecast to be well-defined throughout the day and to develop from shallow at sunrise to deep later in the day. This allowed the investigation of PBL structure over the course of the day and how the distribution of aerosols evolved both within and above the PBL. Understanding the impact of high vertical mixing coupled to the development of a deep boundary layer on air pollution concentrations is important to resolve the vertical distribution of aerosols. The in situ measurements were then compared against modelled results from the Danish Eulerian Hemispheric Model (DEHM) to assess the merits and limitations of the measurement technique, by comparing observed features in the spatiotemporal profile to those predicted by DEHM.

Several previous studies have been conducted that have used instruments installed on light aircraft to measure atmospheric parameters [11,12,13]. In one such study, ultra-fine particles within the lower troposphere were measured using a microlight aircraft and a motorized glider, the latter carrying the instruments in pods mounted under the wings [14]. However, there are presently very few studies involving the flying of an aircraft at low altitudes over an urban area to measure boundary layer properties directly above a city.

Initially, this study was conceived to use Unmanned Aerial Vehicles (UAVs), more colloquially known as drones, as the platform on which to mount aerosol sampling instruments. However, challenges arose that forced a change of approach. UAVs have previously been used to measure atmospheric properties and air pollution [15,16], and there are several benefits over manned aircraft, including the capability to fly at lower speeds, allowing for higher spatial resolution of measurements, and at lower altitudes, facilitating near-surface vertical profiling. They are also typically much cheaper than manned aircraft and require less intensive training to gain a pilot’s license. The rules governing the civil use of UAV have been solidified during the last decade [17], including the regulation that BVLOS (beyond the visual line of sight) flight with the 10–25 kg UAV class is restricted to sparsely populated areas and to a maximum height of 150 m in uncontrolled airspace, and 50 m in controlled airspace (for example, close to airports). Only flights conducted in segregated airspace (airspace set aside for flights with UAVs) are excluded from these restrictions. In practice, these new restrictions almost preclude airborne atmospheric measurement flights with UAV at relevant altitudes over urban areas.

In contrast, permissions to fly manned ultralight aircraft were easy to obtain, even above the city of Copenhagen and in proximity to an international airport (flight paths discussed in Section 2.2). Regarding the choice to use an ultralight aircraft specifically: in Europe, the relatively new ultralight aircraft class is experimental by definition, which means that the procedure for approval of modifications, such as attaching instruments to the aircraft, is straightforward. Conversely, for light aircraft, the experimental classification is rarely used in Europe, which severely limits their applicability in a dynamic research environment. The ultralight aircraft was therefore found to represent a “sweet spot” as a research platform, combining the flight permission access of a manned aircraft with the necessary portability and customisability due to its size and class.

2. Materials and Methods

2.1. Aircraft and Instrumentation

Airborne sampling was performed using an ultralight fixed-wing aircraft (type AP32; Aeroprakt, Kiev, Ukraine). The aircraft type was chosen for its heated cabin, its high inside-cabin payload (2 persons plus a maximum of 35 kg of instruments), in addition to its legal status facilitating modifications by its owner. Modifications to the airframe were only implemented in places that the manufacturer had defined as structurally non-critical. The aircraft was equipped with the flight instruments required for operating inside controlled airspace. In sampling configuration, fuel consumption was approximately 15 L h⁻¹ of unleaded gasoline, resulting in a theoretical endurance of 6 h on full tanks. Air speed was 160 km h⁻¹ and the enroute climb rate, 200 m min⁻¹. The maximum permitted altitude was 9500 ft (2900 m), and the minimum height when overflying residential areas was 1000 ft (300 m).

For sampling the outside air, an isokinetic air inlet, designed and delivered by Brechtel (Hayward, CA, USA), was made of stainless steel and consisted of a diffuser cone (inlet tip 4.42 mm inside diameter), a sampling tube, a water trap, a scarf tube assembly (internal-external pressure equaliser) and 2- and 4-way splitters. The air inlet was mounted 195 mm under the left wing’s lower surface and 2080 mm from the aircraft’s centreline, in an area that had been defined by the manufacturer as undisturbed by propeller slipstream and engine exhaust. The inlet tip protruded 100 mm in front and 260 mm below the wing’s leading edge, thus remaining free of turbulence generated by the wing. Connectors were made of conductive silicone tubing. The analytical instruments individually obtained pressure-free sample air from the water trap by action of their own pumps. Flow time from tip of inlet tube to analytical instruments was approximately 0.2 s, estimated from the sample flow rate through the main inlet tube. Free flux of air from the inlet tube to scarf tube assembly was calculated to be between 28 and 47 L min⁻¹ depending on air speed. Effectively, total sample flux through the inlet exceeded the combined sampling rates of all instruments by an order of magnitude, the surplus of air being dissipated pressure-free through the scarf tube assembly at aircraft speeds between 30 and 50 m s⁻¹.

The instruments chosen for this study were a mixing Condensation Particle Counter (MCPC; Brechtel Model 9400) was used to measure total particle number con-centration, an Optical Particle Counter (OPC; Mini Laser Aerosol Spectrometer, Mini-LAS, Model 11-E, Grimm-Durag, Hamburg, Germany) to measure coarse particle number size distribution and estimate PM2.5, and a Single-channel Tricolor Absorption Photometer (STAP; Brechtel Model 9406) to measure absorption coefficients to deduct BC mass concentrations. Based on assumptions of aerosol particle density, the OPC measurements were converted to PM measurements, and assuming the mass-absorption cross-section (MAC) value of BC, the STAP measurements were converted into equivalent BC (eBC) values.

The Model 9400 MCPC is a mixing condensation particle counter designed to be compact while still having fast response times and is similar to the Brechtel Model 1720 MCPC with minor alterations. It uses gentle turbulent mixing to accelerate the supersaturation process, allowing for significantly faster particle growth. It uses butanol to grow the particles and has a 50% detection threshold of 7 nm in particle diameter. This model and similar models have previously been used in studies to obtain in situ measurements using aerial systems [18]. In this study, outside relative humidity was below 80% throughout. The temperature measured at the MCPC inlet was measured consistently above 20 °C during flights, and was consistently at least 6 °C warmer than outside temperatures, according to instrument sensors. This confirms that sample air was warmed while entering the aircraft. Therefore, the aerosol was not supersaturated before entering the MCPC.

The STAP is a lightweight absorption photometer that continuously draws sample air first through a Brechtel model ACC dryer and then through a filter. It measures the attenuation of light and has been used on many airborne platforms [18,19,20]. The instrument measures at three wavelengths simultaneously: 450, 525 and 624 nm. The measured attenuations are then used to calculate the absorption coefficients at 1-s time resolution. There are several corrections that can be applied to the STAP data to account for filter loading effects and sensitivity to aerosol scattering [21] to retrieve final results. However, minimal post-processing was done to the STAP data here, as it would rely on assumptions that may not be valid for this dataset and preserving the signal of the raw data also allows a clearer assessment of the feasibility of this platform and methodology. Specifically, without an instrument to measure the scattering coefficient, the scattering correction could not be included in the analysis of the STAP data. In order to account for filter loading experimentally, the STAP was run alongside another identical instrument at ground level for 60 min before the first flight and 20 min after the last flight. The STAP used on the aircraft was found to produce values on average 36% higher compared to the second STAP during the first intercomparison, and 30% higher compared to that during the second intercomparison. Temporal correlation between the instruments was also lower during the second intercomparison, so it could not be confidently concluded that there was an observed reduction due to filter loading. Visual inspection of the STAP filters after the flights also revealed no discernible darkening of the sample filter, supporting the conclusion of minimal filter loading. Therefore, to avoid the unnecessary over-processing of data, a filter loading correction was not applied. The absorption coefficients at the 624 nm wavelength were also converted into concentrations of eBC, which is a form of BC defined optically. The term eBC will be used henceforth when referring to STAP measurements only, whereas the term BC will be used for DEHM results and cases where both are compared. The longest wavelength was chosen because although BC is the most abundant absorbing aerosol component, other components such as brown carbon and dust also contribute to absorption, and their contribution is more significant at shorter wavelengths [22]. In the absence of chemical speciation instruments, the MAC value must be estimated. This was achieved utilising the MAC values used by the AE33 aethalometer, another commonly used absorption photometer used to measure eBC concentrations [23]. These MAC values are based on the Ångström law, and so the governing equation was used to calculate a new MAC value of 10.95 m² g⁻¹ for 624 nm.

The Mini-LAS is a portable, battery-powered OPC that counts coarse mode aerosol particles and classifies them by diameter into one of 31 size bins, which range from 0.25 μm to 32 μm. The particle number size distribution and particle mass size distribution are calculated, as are the PM mass fractions PM10, PM2.5 and PM1. Measurement intervals between 6 s and 1 h are available, of which an interval of 1 min was chosen for this study. The instrument also includes a sensor to measure RH, which we used to measure the RH of the sampled aerosol (without drying). Unfortunately, RH data were unavailable during Flight 3 due to technical difficulties.

2.2. Flight Details

The pre-requisite for the desired atmospheric scenario is a cool, clear night leading into a clear day, which matched the forecast for Friday 17 June 2022. Three flights were performed on this day at the local times (UTC+2): 07:08–08:33 (Flight 1), 10:15–11:35 (Flight 2) and 13:37–14:59 (Flight 3). The aircraft took off and landed each time from Roskilde Airport, Denmark, and the flight paths were primarily conducted over a north-south stretch of land in the centre of the region Storkøbenhavn (Greater Copenhagen area), around 6 km west of Copenhagen city centre. This path represents a gradient from the urban-suburban boundary at the southern point of the city to the suburban-rural boundary at the north of Copenhagen. All flights were performed over densely populated areas and inside the controlled airspaces of Roskilde and Copenhagen international airports. Critical parts of our flight path were pre-agreed with Air Traffic Control (ATC), as they were as close as 6 km from the runways at Copenhagen Airport, the extended flight paths of commercial aircraft intersecting horizontally with our own. Permission by ATC to operate inside controlled airspace was obtained via telephone prior to departure, and by maintaining radio contact with ATC during all phases of flight. Altitude changes on the north–south flight path also required stepwise permissions, while ATC continuously ensured sufficient separation between aircraft. Each flight began at Roskilde Airport before heading east towards Avedøreværket and then repeating north–south paths at a selection of altitudes in the range 1200–5000 feet (around 370–1500 m) over a straight line between Avedøreværket and Bagsværd Sø, as seen in Figure 1.

Weather data from the Danish Meteorological Institute Weather Archive for Rødovre were chosen to give insight into the meteorological conditions during the flights, as this municipality lies near the centre of the north–south flight paths [24]. The night between the 16th and 17th of June was clear, with a minimum surface air temperature of 12 °C. Sunrise occurred at 06:45 UTC+2 and temperatures subsequently rose gradually to a peak of around 22 °C at 13:00. Conditions remained clear until early afternoon when clouds began to move in, resulting in cloudy conditions from around 15:00. Wind was recorded from the west between 266° and 289° throughout the flight periods, with mean wind speeds between 3.4 m s⁻¹ and 5.0 m s⁻¹ and maximum wind speeds between 6.2 m s⁻¹ and 7.9 m s⁻¹.

The north–south flight path was chosen for several reasons. Firstly, the proximity of Copenhagen Airport, Kastrup to the south-east placed some limitations on possible flight paths, and it was vital that a path was devised that could be consistently and reliably repeated at different altitudes within the limitations agreed upon with ATC. Secondly, a near-exact north–south line was chosen because it allowed for simpler data visualisation; longitude was near-constant and therefore the spatial extent of the vertical profile sections of the flights could be represented with only two dimensions (latitude and altitude). The path between Avedøreværket and Bagsværd Sø also covers the transition between an urban/suburban region of Greater Copenhagen area towards the southernmost point, and a suburban/rural region of Greater Copenhagen area towards the north, allowing for the investigation of urban–regional horizontal gradients. Care was taken to ensure that the aircraft exhaust would not be sampled by the instruments onboard, by consulting wind speed and direction forecasts while making flight plans. During this study, the altitudinal separation between consecutive north–south passes was at least 500 feet (~150 m) and winds were >5 ms⁻¹ from the west. It was therefore concluded that even considering turbulence and diffusion, it would be sufficiently far below (at least 150 m) and eastward (at least 600 m) on each successive pass to exclude the exhaust from contributing to the results.

All flights taken in this study—and more broadly, with this platform—have avoided flying through clouds. This is primarily on the grounds of safety, but it additionally ensures the health of the instruments, and that they will not measure highly supersaturated air.

2.3. The Danish Eulerian Hemispheric Model

The Danish Eulerian Hemispheric Model (DEHM) is a 3-D atmospheric chemistry transport model covering the area of the Northern Hemisphere and the lowest ~18 km of the atmosphere [25,26]. DEHM includes a comprehensive chemical scheme, and natural emissions of e.g., sea salt and biogenic VOCs are calculated on-line in the model, while maps of anthropogenic emissions are based on international and national inventories (see Frohn et al., 2022 for details [27]). The data obtained from the model for the present study are based on the setup applied in the EU-funded CAMS2-40 project ‘Regional production’, providing air quality forecasts for Europe based on 11 European air pollution models (https://ads.atmosphere.copernicus.eu/datasets/cams-europe-air-quality-forecasts?tab=overview, accessed on 1 September 2022). These emission inventories are annual averages from 2018, with temporal variation achieved by overlaying seasonal, weekly and hourly activity patterns. For sectors where emissions decline, e.g., due to legislation, this may result in an overestimation of modelled concentrations.

In this setup, DEHM covers Copenhagen with a horizontal resolution of 0.1° × 0.1° and outputs results for 8 vertical layers (5, 50, 250, 500, 1000, 2000, 3000, and 5000 m above ground) covering the lowest 5 km with an hourly temporal resolution. A 3D-Var data assimilation was conducted to improve the simulation of 3D PM2.5 concentration field using the ground observations combined with modelled vertical correlations of PM2.5 at an hourly basis [28]. This is achieved through the model creating a background field, which is then corrected using ground observations, and this results in a 3D PM2.5 field that is consistent with both observations and model physics. In the model, total BC is estimated as the sum of fresh and aged partitions, where fresh represents the BC aerosols right after emission, and aged BC reflects the changes in physical properties with respect to condensation of soluble material occurring after some time in the atmosphere. Fresh and aged BC concentrations from residential and wildfire sources are also calculated individually. Total modelled PM2.5 includes elemental and primary organic carbonaceous particles (BC/OC), mineral dust (combustion process emission, but without organic content), sea salt, secondary inorganic aerosols (i.e., NO₃⁻, NH₄⁺, SO₄²⁻), secondary organic aerosols, and error corrections from the 3D-var data assimilation.

3. Data Processing

Data from the STAP, MCPC, OPC and aircraft flight log were synchronised and then harmonised to 1-min time resolution. Quality control was performed to remove unphysical data, and data were given several tags to differentiate flight paths and flag any points that corresponded to suspected strong point sources. For example, there is an aircraft holding point in the airspace above the heat and power plant Avedøreværket, and it was common in the flights performed throughout the year that the aircraft would pass over the plant. In several of these cases, clear spikes in PN and PM were observed reaching an order of magnitude above background levels, due to the sampling of aerosols emitted from the plant. Similarly, during earlier circling maneuvers, brief peaks of PN and PM were observed in response to the aircraft crossing its own flight path. However, on the 17 June 2022, there was no evidence of a signal from Avedøreværket, nor any other aerosol emission point sources or aircraft exhaust.

In this study, data presented from the aircraft flight log include GPS co-ordinates in three dimensions (latitude, longitude, altitude) and outside air temperature (OAT). An offset in the aircraft OAT data was observed, and therefore a comparison was performed against an external thermometer over the course of one flight. It was found that the correlation was described by a constant offset of −3.4 °C (95% confidence bounds: −3.7 °C, −3.0 °C) with an R-square value of 0.98. This correction was therefore applied to all OAT data.

Total BC and PM2.5 data from DEHM were interpolated in both space (from 0.1° × 0.1° grid cells) and time (from 1-h time resolution) to extract values at the measurement data points. Altitudes were also harmonised such that all data refers to altitude above mean sea level (MSL).

Vertical profile contour plots as shown in Figure 2 and Figure 3 were created using the meshgrid and griddata functions in MATLAB 2022b, used to interpolate the irregular 2-dimensional data in latitude and altitude. Of the several methods offered by griddata, the linear interpolation algorithm was chosen, as it was found to produce the most faithful reflection of the data points, as opposed to falsely exaggerating smaller features (observed using cubic interpolation) or obscuring them entirely (observed using nearest neighbour and natural neighbour methods). Latitude steps of 0.01° and altitude steps of 50 m were used. The data and script used to produce all figures are available; see Open Research.

4. Results

The flight paths over Greater Copenhagen area and corresponding time series of altitude, OAT, RH, PN_tot, PM2.5 and BC mass concentration are shown in Figure 1. PM2.5 and BC concentration data extracted from DEHM are also shown in blue. The measurements reveal a well-defined PBL that gradually expands over the course of the day, while retaining a distinct boundary separating it from the free troposphere. This boundary can be seen in the sudden reductions in PN, PM2.5 and eBC concentration measurements as the aircraft climbs (such as at 10:35), and sudden increases in these parameters as the aircraft descends (such as at 11:15). The instruments were monitored during flight to ensure sampling of both PBL and free-tropospheric air masses, which is the reason that later flights reach higher altitudes. It was found that the STAP often reported slightly negative values of eBC while sampling the clean free-tropospheric air masses, which is unphysical, though the values were most often within the expected uncertainty of the sampling method. Moreover, the STAP demonstrated close correlation to PM2.5 measurements from the OPC while inside the PBL, with measured concentrations substantially larger than the negative values in the free troposphere, and therefore the dataset is evaluated to be reliable within the PBL.

The difference in homogeneity of the PBL can also be deduced from the variability of the measurements in Figure 1. During Flights 1 and 3, there is significant variation in PN, PM2.5 and eBC concentrations while the aircraft is located in the PBL. In contrast, Flight 2 features much less variability, indicating a more homogenous, well-mixed boundary layer during the late morning. An oscillating pattern is seen between around 14:00 and 14:15 in the PN, PM2.5 and eBC data, indicating that the aerosol concentration was not simply decreasing monotonically with increasing altitude between 1 km and 1.5 km.

PM2.5 and BC data from DEHM exhibit many of the same features observed in the aircraft PM2.5 and eBC measurements, though the magnitude is overestimated in both parameters by the model. The patterns relating to boundary layer structure and corresponding evolution appear to be recreated well, resulting in distinct decreases and increases in particulate concentrations as the aircraft ascends and descends, respectively. A minor exception is the BC concentration during the first half of Flight 1, where DEHM predicts BC to follow PM2.5, but no measurable eBC was detected by the STAP.

In order to analyse the vertical profiles of each measured parameter, the north–south sections of the flights were isolated, and the parameters were interpolated over latitude and altitude. The resulting contour plots are shown in Figure 2. It should be noted that the y axes begin at 300 m altitude and relationship between latitude and altitude is not to scale; the x axis spans an equivalent of around 22 km, such that the x:y axis ratio is in reality around 17:1. The vertical profiles illustrate more intricately the structure and evolution of the PBL.

In order to estimate the PBL height, both meteorological and aerosol parameters can be consulted. During Flight 1, the temperature maximum lies at an altitude of around 600 m, rather than at the Earth’s surface. This is likely due to a deep residual boundary layer containing warmer air from the previous day, and a shallow, cooler emerging boundary layer whose temperature more closely follows the surface temperature. This is supported by the aerosol measurements, which also suggest a PBL height of 400–600 m. This is an artefact of a stable boundary layer with little vertical mixing, such that the cooler surface only interacts with a thin layer of the atmosphere. The residual boundary layer also results in a scenario in which certain altitudes exhibit temperature changes that do not follow the rise and fall of the diurnal surface temperature, and aerosol concentration changes that do not follow daily emission patterns. At around 1000 m in altitude, the temperature remained at around 10 °C during all three flights, even as surface temperatures increased from 12 °C to 22 °C. The heat island effect can also be seen towards the surface, as the southern, more urban area warms slightly faster than the northern, more rural area. RH during Flight 1 was between 50% and 70%, and it can be seen in Figure 2 that the presumed shallow PBL carries more humid air than the layer above.

Data from Flight 2 in the late morning show that the surface has begun to warm, and the moister PBL has increased in depth. RH shows a maximum of 73% between around 600 m to 800 m. though the PBL height is difficult to ascertain from temperature measurements alone due to the complex system with a warming PBL and dissipating residual boundary layer. RH and aerosol measurements can be used to provide more context, indicating a PBL height of around 700–900 m. During Flight 1, the aerosol concentration is primarily confined to the emerging PBL below 600 m. However, Flight 2 shows a slight decrease in particulate concentration at certain altitudes, such as around 1000 m, demonstrating that the residual boundary layer retained a slightly elevated concentration that diminished as the residual layer dissipated. This also leads to lower minimum and higher maximum particulate concentrations as the day progresses, which can also be seen in Figure 1.

During Flight 3, the temperature profile shows a gradual decrease with increasing altitude. Unfortunately, due to technical problems with the sensor, RH is not available for Flight 3. The PBL height can be estimated using aerosol profiles, indicating that mixing occurred up to around 1200–1400 m. It can also be seen from the aerosol concentrations that the PBL became less stable and homogenous. As the surface temperature increased, the PBL will have expanded and vertical mixing within this layer will have become stronger. This will have resulted in the upward transport of emissions from the morning rush hour and other urban sources. This is likely the cause of features such as the maximum in aerosol concentration around 600–1000 m. The BC data maximum is observed to increase from around 75 ng cm⁻³ to 95 ng cm⁻³, representing a concentration increase of around 25%. Meanwhile the PM2.5 exhibits an increase of around 18% from 5.5 µg cm⁻³ to 6.5 µg cm⁻³. This indicates that local urban emissions, and transport from other sources, are increasing the BC fraction of the aerosol throughout the day. There is also a local maximum seen in aerosol concentrations during Flight 3, at the southernmost latitudes and around 1200 m altitude. An explanation for this local maximum is an entrainment of pollution, transported to Copenhagen from further afield.

Data for PM2.5 and BC concentrations were extracted from DEHM results at the measurement sample points, and these were interpolated using the same methods as for the measurements. The comparison between measured and modelled data is shown in Figure 3. It is evident that DEHM overestimated the concentrations, and depicted a more gradual transition between the PBL and the free troposphere. The primary cause of the overestimation is likely the limitation of the emission inventories, which are based on annual totals from 2018 overlaid with temporal distribution functions describing seasonal, weekly and daily variations. The vertical resolution of the model also limits its ability to render a sharp boundary like the observed transition. Nevertheless, DEHM recreated certain features well. The PM2.5 concentration above around 700 m decreases from Flight 1 to Flight 2, resembling the evidence of a residual boundary layer observed experimentally. During Flight 2, DEHM also showed the steepest gradient in aerosol concentrations within the 700–900 m altitude range, mirroring the measurements and demonstrating that DEHM recreated the PBL height well. Finally, although DEHM did not recreate the aerosol concentration maximum in Flight 3, the horizontal gradient observed in eBC is also seen in the modelled results.

5. Conclusions

The emergence of the new class of ultralight aircraft, along with the recent availability of miniaturised lightweight aerosol monitoring instruments facilitates significant advances for in situ atmospheric measurements. Almost 60 years ago, an expensive light twin-engine aircraft with a maximum take-off mass of 1769 kg was able to carry 193 kg of aerosol measuring equipment and 2 persons, the generators providing 120 A of 28 VDC electrical power to meet instrument needs [11]. In this study, using an economical ultralight 2-seated aircraft with a maximum take-off mass of 600 kg, instruments with a total weight of 5 kg and requiring 7.5 A of 12 VDC power were used to measure aerosol particle number concentration, PM2.5, eBC concentration and meteorological parameters. This platform could be extended to investigate other atmospheric properties, meteorological parameters or aerosol chemical speciation. This study highlighted several merits and limitations of the method, which will be more or less relevant for further studies depending on the climate and atmospheric conditions, measurement site, location-based flight limitations and necessary permissions, desired vertical flight range, and choice of scientific instrumentation.

The new ultralight-based instrument platform proved versatile with respect to payload, and uncomplicated with respect to instrumentation mounted on the outside of the ultralight aircraft. In addition, it was operationally flexible, in terms of flight planning and flight permissions, even in controlled airspace and in an area of high traffic density. During flight, the instrument platform was successfully operated under air traffic rules applicable to General Aviation, which facilitated a high degree of responsiveness by ATC to changes of flight plans. In our aircraft, the instruments were placed in the cabin, the outside air being supplied from in front of the wing via a specially designed inlet (Brechtel, Hayward, CA, USA). This is expected to be useful in cold climate, such as the Arctic, as inside-cabin temperature can be maintained above freezing at all times, accommodating the lower end temperature limits of most instruments. For recording outside temperature and relative humidity during flight, we mounted an appropriate probe close to the air inlet.

The crew size of two allowed for workload distribution between pilot and instrument operator, which was used advantageously for ad hoc changes to the flight path and altitude based on real-time data. Limitations of the method include its maximum altitude limit, and its minimum height limit when overflying residential areas. Also, because the possibility of flying depends on weather conditions, a bias may arise towards calmer and clearer days, as it will be more difficult to obtain data on days with more turbulent or rainy conditions. These biases must be understood and taken into account when building large datasets from flight measurements like those in this study. Despite certain limitations, it is evident that the aircraft platform was able to capture physical and very meaningful data during this study, and would be suitable for a wide range of similar applications.

Based on aircraft measurements, we were able to build a spatiotemporal profile of the PBL as it evolved due to surface warming, emission of gases and pollutants into the PBL, and atmospheric mixing over the course of a day. Flights were performed both in the boundary layer and the free troposphere. A shallow boundary layer was observed during the first flight shortly after sunrise, with a height of around 400–600 m, which expanded to around 1200–1400 m by the third flight in the early afternoon. Evidence of a residual boundary layer was also observed during the first flight, which dissipated by the second flight in the late morning. Data from the aerosol monitoring instruments also indicated an entrainment of pollution near the top of the boundary layer during the third flight. All main features were observed similarly across the three aerosol monitoring instruments, supporting the reliability of the data.

Experimental data from the flights were compared with modelled PM2.5 and BC concentrations from DEHM. Many of the features measured were recreated well, including the evolution of the boundary layer structure and horizontal gradient of BC concentrations. The similarities of the observed structure, features and evolution of the PBL to those predicted by DEHM support the validity of the aircraft measurements. However, DEHM was found to overestimate the aerosol concentrations, likely owing to limitations of its emission inventories, and did not accurately recreate the sharpness of the transition between the boundary layer and free troposphere due to the limited vertical resolution of the model. Further in situ measurements of boundary layer dynamics and aerosol mixing with vertical resolution will allow the improvement of models to simulate aerosol concentrations both in the atmospheric column and at the surface, benefitting our understanding of the impacts of aerosols on radiative forcing and air quality affecting human health.

Author Contributions

Conceptualization, D.C.T., U.G., L.M.F., Z.Y., A.M. and H.S.; Data curation, D.C.T. and U.G.; Formal analysis, D.C.T., L.M.F. and Z.Y.; Funding acquisition, H.S.; Investigation, D.C.T., U.G., L.M.F., Z.Y. and H.S.; Methodology, D.C.T., M.D.C., L.M.F., Z.Y., A.M. and H.S.; Project administration, U.G. and H.S.; Resources, U.G. and M.D.C.; Supervision, U.G., A.M. and H.S.; Validation, A.M. and H.S.; Visualization, D.C.T.; Writing—original draft, D.C.T. and U.G.; Writing—review & editing, D.C.T., U.G., L.M.F., Z.Y., A.M. and H.S. All authors have read and agreed to the published version of the manuscript.

Data Availability Statement

The files containing MCPC, STAP, OPC and flight log data used in this study and the MATLAB script used to analyse and visualise these data are available on Zenodo via https://doi.org/10.5281/zenodo.7913181 (accessed on 9 May 2023), Version 1, with Open Access.

Acknowledgments

Our special thanks go to the air traffic controllers of Naviair (Kastrup, Denmark), for facilitating our measuring flights within the Terminal Control Area of Copenhagen airport. Villum Research Station is acknowledged for support of the STAP and MCPC to this study. Support was also received from The Danish Energy Agency (Energistyrelsen) within the project “Vertical distribution of particle number concentration and black carbon, TAS nr. 4005-510108”. Finally, Fred Brechtel is acknowledged for his assistance in the instrument and aircraft setup.

Conflicts of Interest

Michael Dines Christiansen is director of Dines Avitech Aps. The paper reflects the views of the scientist and not the company.

Footnotes

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Figures

Figure 1. Flight paths and time series of measured parameters: altitude, outside air temperature (OAT), relative humidity (RH, blue lines), total particle number concentration (PNtot), PM2.5 concentration and eBC concentration. Time elapsed is represented in the flight paths and altitude plots by a coloured scale from blue to red chronologically. RH data were not available during Flight 3. Modelled PM2.5 and BC concentrations from DEHM are also shown in red. All times are local (UTC+2). Attribution (basemap): Earthstar Geographics. Two parallel runways of Copenhagen International Airport are recognizable in the bottom right corner of the map of Flight 1.

Figure 2. Vertical profiles of measured parameters: outside air temperature (OAT), relative humidity (RH), total particle number concentration (PNtot) from the MCPC, PM2.5 concentration from the OPC, and eBC concentration from the STAP. Data were harmonised to 1-min time resolution and then interpolated. RH data were unavailable during Flight 3. Crosses symbolise the spatial positions of the 1-min resolution harmonised data points.

Figure 3. Vertical profiles PM2.5 concentration from the OPC compared against the same parameter from DEHM, and eBC concentration from the STAP compared to BC concentration from DEHM. Data were harmonised to 1-min time resolution and then interpolated. Crosses symbolise the spatial positions of the 1-min resolution harmonised data points.

References

1. Chen, R.; Kan, H.; Chen, B.; Huang, W.; Bai, Z.; Song, G.; Pan, G. Association of particulate air pollution with daily mortality: The China Air Pollution and Health Effects Study. Am. J. Epidemiol.; 2012; 175, pp. 1173-1181. [DOI: https://dx.doi.org/10.1093/aje/kwr425] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/22510278]

2. Pope, C.A., III; Coleman, N.; Pond, Z.A.; Burnett, R.T. Fine particulate air pollution and human mortality: 25+ years of cohort studies. Environ. Res.; 2020; 183, 108924. [DOI: https://dx.doi.org/10.1016/j.envres.2019.108924] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/31831155]

3. Szopa, S.; Naik, V.; Adhikary, B.; Artaxo, P.; Berntsen, T.; Collins, W.; Klimont, Z. Short-Lived Climate Forcers. Proceedings of the Presented at the AGU Fall Meeting Abstracts; New Orleans, LA, USA, 13–17 December 2021.

4. Tian, P.; Zhang, L.; Cao, X.; Sun, N.; Mo, X.; Liang, J.; Li, X.; Gao, X.; Zhang, B.; Wang, H. Enhanced bottom-of-the-atmosphere cooling and atmosphere heating efficiency by mixed-type aerosols: A classification based on aerosol nonsphericity. J. Atmos. Sci.; 2018; 75, pp. 113-124. [DOI: https://dx.doi.org/10.1175/JAS-D-17-0019.1]

5. Mishra, A.K.; Koren, I.; Rudich, Y. Effect of aerosol vertical distribution on aerosol-radiation interaction: A theoretical prospect. Heliyon; 2015; 1, e00036. [DOI: https://dx.doi.org/10.1016/j.heliyon.2015.e00036] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/27441222]

6. Samset, B.H.; Myhre, G.; Schulz, M.; Balkanski, Y.; Bauer, S.; Berntsen, T.K.; Bian, H.; Bellouin, N.; Diehl, T.; Easter, R.C. et al. Black carbon vertical profiles strongly affect its radiative forcing uncertainty. Atmos. Chem. Phys.; 2013; 13, pp. 2423-2434. [DOI: https://dx.doi.org/10.5194/acp-13-2423-2013]

7. Zarzycki, C.M.; Bond, T.C. How much can the vertical distribution of black carbon affect its global direct radiative forcing?. Geophys. Res. Lett.; 2010; 37, L20807. [DOI: https://dx.doi.org/10.1029/2010GL044555]

8. Costantino, L.; Bréon, F.M. Analysis of aerosol-cloud interaction from multi-sensor satellite observations. Geophys. Res. Lett.; 2010; 37, L11801. [DOI: https://dx.doi.org/10.1029/2009GL041828]

9. Heald, C.L.; Coe, H.; Jimenez, J.L.; Weber, R.J.; Bahreini, R.; Middlebrook, A.M.; Russell, L.M.; Jolleys, M.; Fu, T.-M.; Allan, J.D. et al. Exploring the vertical profile of atmospheric organic aerosol: Comparing 17 aircraft field campaigns with a global model. Atmos. Chem. Phys.; 2011; 11, pp. 12673-12696. [DOI: https://dx.doi.org/10.5194/acp-11-12673-2011]

10. WHO. WHO Global Air Quality Guidelines: Particulate Matter (PM2.5 and PM10), Ozone, Nitrogen Dioxide, Sulfur Dioxide and Carbon Monoxide; World Health Organization: Geneva, Switzerland, 2021.

11. Adams, D.F.; Koppe, R.K. Instrumenting Light Aircraft for Air Pollution Research. J. Air. Pollut. Control. Assoc.; 1969; 19, pp. 410-415. [DOI: https://dx.doi.org/10.1080/00022470.1969.10466504]

12. Pratt, K.A.; Prather, K.A. Aircraft measurements of vertical profiles of aerosol mixing states. J. Geophys. Res.; 2010; 115, D11305. [DOI: https://dx.doi.org/10.1029/2009JD013150]

13. Liu, P.; Zhao, C.; Zhang, Q.; Deng, Z.; Huang, M.; Ma, X.; Tie, X. Aircraft study of aerosol vertical distributions over Beijing and their optical properties. Tellus B Chem. Phys. Meteorol.; 2009; 61, pp. 756-767. [DOI: https://dx.doi.org/10.1111/j.1600-0889.2009.00440.x]

14. Junkermann, W.; Hacker, J.M. Ultrafine Particles in the Lower Troposphere: Major Sources, Invisible Plumes, and Meteorological Transport Processes. Bull. Am. Meteorol. Soc.; 2018; 99, pp. 2587-2602. [DOI: https://dx.doi.org/10.1175/BAMS-D-18-0075.1]

15. Adkins, K.A.; Swinford, C.J.; Wambolt, P.D.; Bease, G. Development of a sensor suite for atmospheric boundary layer measurement with a small multirotor unmanned aerial system. Int. J. Aviat. Aeronaut. Aerosp.; 2020; 7, 4. [DOI: https://dx.doi.org/10.15394/ijaaa.2020.1433]

16. Jońca, J.; Pawnuk, M.; Bezyk, Y.; Arsen, A.; Sówka, I. Drone-Assisted Monitoring of Atmospheric Pollution—A Comprehensive Review. Sustainability; 2022; 14, 11516. [DOI: https://dx.doi.org/10.3390/su141811516]

17. EASA. Easy Access Rules for Unmanned Aircraft Systems (Regulations (EU) 2019/947 and 2019/945)—Revision from July 2024; EASA: Cologne, Germany, 2024.

18. Bates, T.S.; Quinn, P.K.; Johnson, J.E.; Corless, A.; Brechtel, F.J.; Stalin, S.E.; Meinig, C.; Burkhart, J.F. Measurements of atmospheric aerosol vertical distributions above Svalbard, Norway, using unmanned aerial systems (UAS). Atmos. Meas. Tech.; 2013; 6, pp. 2115-2120. [DOI: https://dx.doi.org/10.5194/amt-6-2115-2013]

19. Pikridas, M.; Bezantakos, S.; Močnik, G.; Keleshis, C.; Brechtel, F.; Stavroulas, I.; Argyrides, M. On-flight intercomparison of three miniature aerosol absorption sensors using unmanned aerial systems (UASs). Atmos. Meas. Tech.; 2019; 12, pp. 6425-6447. [DOI: https://dx.doi.org/10.5194/amt-12-6425-2019]

20. Pilz, C.; Düsing, S.; Wehner, B.; Müller, T.; Siebert, H.; Voigtländer, J.; Lonardi, M. CAMP: A balloon-borne platform for aerosol particle studies in the lower atmosphere. Atmos. Meas. Tech. Discuss.; 2022; 2022, pp. 1-23. [DOI: https://dx.doi.org/10.5194/amt-15-6889-2022]

21. Ogren, J.A. Comment on “Calibration and intercomparison of filter-based measurements of visible light absorption by aerosols”. Aerosol Sci. Technol.; 2010; 44, pp. 589-591. [DOI: https://dx.doi.org/10.1080/02786826.2010.482111]

22. Andreae, M.O.; Gelencsér, A. Black carbon or brown carbon? The nature of light-absorbing carbonaceous aerosols. Atmos. Chem. Phys.; 2006; 6, pp. 3131-3148. [DOI: https://dx.doi.org/10.5194/acp-6-3131-2006]

23. Drinovec, L.; Močnik, G.; Zotter, P.; Prévôt, A.S.H.; Ruckstuhl, C.; Coz, E.; Rupakheti, M.; Sciare, J.; Müller, T.; Wiedensohler, A. et al. The “dual-spot” Aethalometer: An improved measurement of aerosol black carbon with real-time loading compensation. Atmos. Meas. Tech.; 2015; 8, pp. 1965-1979. [DOI: https://dx.doi.org/10.5194/amt-8-1965-2015]

24. DMI. Vejrarkivet. 2022; Available online: https://www.dmi.dk/dmis-vejrprodukter/vejrarkivet/ (accessed on 13 September 2023).

25. Brandt, J.; Silver, J.D.; Frohn, L.; Geels, C.; Gross, A.; Hansen, A.B.; Hansen, K.; Hedegaard, G.; Skjøth, C.; Villadsen, H. et al. An integrated model study for Europe and North America using the Danish Eulerian Hemispheric Model with focus on intercontinental transport of air pollution. Atmos. Environ.; 2012; 53, pp. 156-176. [DOI: https://dx.doi.org/10.1016/j.atmosenv.2012.01.011]

26. Christensen, J.H. The Danish Eulerian hemispheric model—A three-dimensional air pollution model used for the Arctic. Atmos. Environ.; 1997; 31, pp. 4169-4191. [DOI: https://dx.doi.org/10.1016/S1352-2310(97)00264-1]

27. Frohn, L.M.; Geels, C.; Andersen, C.; Andersson, C.; Bennet, C.; Christensen, J.H.; Im, U.; Karvosenoja, N.; Kindler, P.A.; Kukkonen, J. et al. Evaluation of multidecadal high-resolution atmospheric chemistry-transport modelling for exposure assessments in the continental Nordic countries. Atmos. Environ.; 2022; 290, 119334. [DOI: https://dx.doi.org/10.1016/j.atmosenv.2022.119334]

28. Silver, J.D.; Christensen, J.H.; Kahnert, M.; Robertson, L.; Rayner, P.J.; Brandt, J. Multi-species chemical data assimilation with the Danish Eulerian hemispheric model: System description and verification. J. Atmos. Chem.; 2016; 73, pp. 261-302. [DOI: https://dx.doi.org/10.1007/s10874-015-9326-0]