Measurement of boundary layer ozone concentrations on-board a Skywalker unmanned aerial vehicle

This study demonstrates novel measurements of in situ ozone (O 3 ) concentrations and thermodynamics sampled on-board an instrumented Skywalker Unmanned Aerial Vehicle (UAV). Small spatial and temporal gradients were observed over a localized region, which nearby ground-based in situ measurements lack the ability to resolve. It was found that the UAV-measured O 3 concentrations provided a useful additional indicator of O 3 variability at the sub-urban scale. The ability to sample subtle variability over a localized area highlights the important and novel capabilities of UAVs to rapidly characterize local area micrometeorology and chemistry.


Introduction
Unmanned Aerial Vehicles (UAVs) are remotely or autonomously piloted aircraft.While UAVs have recently been most associated with military applications, their use in the field of atmospheric science and environmental monitoring is rapidly growing, from the monitoring of carbon dioxide (CO 2 ) concentrations (Watai et al., 2006) to observing the spatial distribution of evapotranspiration (Rauneker and Lischeid, 2012).With the continued miniaturization of highly accurate and precise sensors, their potential effectiveness to make low-cost measurements at high spatial resolution is the subject of much scientific and technological interest.
For smaller UAVs the limiting factor in their utility is typically the availability of high quality, miniaturized sensors necessary for their reduced payload capacity, as well as relatively short-duration flight times (typically of the order 1-2 h); while one of the biggest current challenges for larger platforms concerns permission to fly by the appropriate regulatory bodies, such as the Civil Aviation Authority (CAA) for UK airspace.
Low Altitude, Short Endurance (LASE) UAVs are relatively simple to operate, with simple ground-control stations and control mechanisms, requiring only a small crew (Watts et al., 2012).Their small size means that they can be hand-launched from a variety of terrains, and in the UK UAVs with an operating mass of 7 kg or less are exempt from the majority of the regulations that are normally applicable to large and manned aircraft (CAA, 2010).In the atmospheric sciences, UAVs have now been used for a variety of purposes, from making in situ measurements of thermodynamic properties in the planetary boundary layer (Houston et al., 2012) to studying emissions at active volcano sites (Diaz et al., 2010).Measurements of reactive gases such as ozone (O 3 ) from UAVs offer a novel opportunity to sample the three-dimensional (3D) spatial and temporal variability of such gases to enable process analysis.
Although only about 10% of all atmospheric O 3 is located in the troposphere, it is a principal driver of the photochemical processes regulating many of the gases that are emitted into the atmosphere by either natural or anthropogenic processes.In addition to this, tropospheric O 3 is itself a pollutant with impacts on human health and the environment.In Europe alone it is estimated that O 3 contributes to over 20 000 premature deaths per annum (EEA, 2007).Tropospheric O 3 is also a greenhouse gas and has been reported to contribute a net warming effect to the climate system, with a radiative forcing estimated at 0.35 W m −2 (Forster et al., 2007).
Tropospheric O 3 is not emitted; rather it is produced in the atmosphere from reactions involving precursor pollutants such as volatile organic compounds (VOCs) and NO X (nitrogen oxide and nitrogen dioxide).These rapid chemical interactions can result in large spatial and temporal gradients of O 3 in urban environments, which usually peak in sub-urban areas downwind, with the O 3 production rate generally increasing with NO X concentrations (Fowler, 2008).However, in urbanized centres, characterized by even higher concentrations of atmospheric NO X , O 3 production can be inhibited, as a result of the reaction of O 3 with NO and the formation Boundary layer ozone concentrations from a UAV 253 of NO y (i.e. the sum of NO X plus the reservoirs of NO X , such as nitric acid and peroxyacetyl nitrate).Thus, in an urban environment, a reduction in NO X concentrations can often lead to an increase in tropospheric O 3 .With future estimates of NO X emissions in the UK predicting reductions of 45% over the next decade (Hall et al., 2006), an even greater emphasis might be placed on monitoring urban hazardous O 3 concentrations on the local scale.
Ozone concentrations depend strongly on the spatial distributions of NO X and VOCs (and their associated emission sources), which vary both from one city to another as well as within the urban environment itself.Prevailing meteorological conditions can also vary rapidly, and mixing by micrometeorological processes such as the urban heat island effect and street-canyon-scale dynamics all act to modulate the chemistry and transport of ozone and its tracers across a range of temporal and spatial scales.Ground-based in situ measurements such as the Automatic Urban and Rural Network (AURN) in the UK operated by the Department for Environment, Food and Rural Affairs (Defra) can provide surface-level O 3 concentrations, although such fixed in situ measurements are limited in their spatial coverage.AURN currently has 103 active sites across the UK (Figure 1), which while providing high-resolution hourly information, is not able to provide information about surface-level O 3 concentrations on the sub-urban scale.Measurements from ground-based monitoring sites such as AURN are often used as validation datasets for regional air quality models such as the Met Office Air Quality Unified Model.Clearly, the inability of such sites to inform on the sub-urban scale can lead to a poor interpretation (and validation) of urban environments (Savage et al., 2012).
An alternative to ground-based measurements of tropospheric O 3 can be to use aircraft.Typically, airborne measurements around urban environments are made using large aircraft.However, such flight campaigns are not only expensive, but still also lack the required spatial resolution for many applications, e.g.Manchester city centre has a diameter of approximately 2 km, which combined with the science speed of the UK's atmospheric research aircraft (∼100 m s −1 ), means that a typical 1 Hz instrument would only be able to make approximately 20 measurements during an overpass of the city.Large research aircraft are also restricted by the CAA, which often means that they are unable to fly around urban centres or within the lower boundary layer.
However, UAVs offer an ideal alternative at such scales, bridging the gap between ground-based and traditional airborne methods, with the potential to deliver detailed, high-resolution, and precise measurements of tropospheric and near-surface O 3 concentrations at the local scale.Potentially synergistic instruments such as the CityScan ground-based NO 2 remote sensing system (Roland Leigh, pers.comm.) in development at the University of Leicester could be used in conjunction with UAV sampling to fully characterize the sub-urban scale.
In this paper, we present measurements of planetary boundary layer O 3 concentrations measured in situ, thereby demonstrating the development of a system with the capabilities of providing high spatial and temporal resolution sampling, which can be deployed in urban environments.This work describes this development and the first field measurements at a site near to Manchester city centre.

System design
In this study, we have adapted an Electrochemical Concentration Cell (ECC) ozonesonde to fly on a fixed-wing LASE UAV.We now describe the components and integration of this system.
The ECC ozonesonde (manufactured by Science Pump Ltd) comprises a motor, a Teflon pump, and the ECC module.The cell is made up of two chambers (anode and cathode) containing electrodes made of bright platinum and a potassium iodide solution with    radio to a dedicated ground station.The accuracy and precision for these measurements are given in Table I.
Tropospheric O 3 concentrations measured using this system have a typical relative precision of ±3-6% (Smit and Sträter, 2004), corresponding to ∼1-2 ppb at concentrations measured here, with a sampling frequency of 0.5 Hz.This was also confirmed in the laboratory before flight.Anode and cathode solutions for the ECC were prepared in the laboratory at known concentrations in triple-distilled water, and the cell current was calibrated and validated for these solutions under ambient and high (saturated) O 3 concentrations.A calibrated Thermo Scientific TE49C photometric ozone analyser measured the ambient concentration in the lab, and a Science Pump Ltd TSC-1 ozoniser unit generated saturated flows.For a further description of the ozonesonde and radiosonde system, see Skrivankova (2004) and references therein.
The Skywalker airframe (the fuselage and wings) was designed and manufactured by Skywalker Technologies, before being assembled at the University of Manchester; its specifications are summarized in Table II.As well as providing the frame onto which the ozonesonde was attached, the UAV also provided GPS information.
Combining an ozonesonde with a UAV controlled via radio therefore allows the collection of real-time O 3 , GPS, pressure, temperature, and humidity data.
When integrating the ozonesonde into the UAV, the electro-mechanical integration of the payload equipment needed to be taken into consideration.In addition to incorporating the ozonesonde, the final payload had to house the inlet pump motor to draw air into the ozonesonde for measurement, as well as discrete voltage sources for both the pump motor and ozonesonde.The final integration of the Vaisala probe and Skywalker airframe is shown in Figures 2 and 3

Results
Test flights were carried out at Hough End Fields, Manchester on 28 June 2012 at midday, with partially cloudy conditions and gusting winds of up to 10 mph as recorded by a handheld anemometer.While the Skywalker UAV (including ozonesonde payload) weighs less than the 20 kg for which a certificate of airworthiness becomes a prerequisite, there are still a number of restrictions that are enforced by the CAA for those UAVs without an approved detect and avoid capability (as was the case here), namely that: the UAV is not allowed to fly in controlled airspace; the UAV can not be flown at an altitude exceeding 400 ft; the UAV cannot be flown within 150 m of any congested area of a city or town, or within 30 m of any person; and that a direct line of sight (500 m maximum range) must be maintained (CAA, 2010).Working under these restrictions, the Hough End Fields site, which is an existing model flying location (Figure 4), was chosen.The ozonesonde sampled for two flying periods: from 12:25:36 to 12:50:26 and 15:09:50 to 15:41:30.These data are shown in Figure 5, with the corresponding meteorological data (pressure, temperature, and RH) shown in Figure 6.As seen from these plots, the O 3 appears to be reasonably well mixed in the lower boundary layer with a mean and standard deviation across this data set being 30.11 and 0.70 ppb, respectively.From Figure 6, the periods where the pressure was constant at roughly 1005 hPa (e.g. at approximately 12:40) indicate the periods of time when the UAV was grounded.
Peak concentrations of approximately 39 ppb were observed at 15:10.This peak was associated with a short-term shift in the prevailing wind direction to a more north-easterly direction, bringing air from the nearby main road (A5103, Princess Road) and Manchester 'city centre'.Such changes in wind direction were measured on this day using a handheld anemometer and were noted to be coincident with turbulent downdrafts associated with passing non-precipitating cloud.Our ability to sample this variability illustrates the potential for future process studies that relate to air quality, micrometeorology, and local dynamics in regions of strong local sources.
For comparative purposes, the data from two Manchester-based AURN sites is also shown in Figure 5.These sites are situated in Manchester city centre (Manchester Piccadilly: 53.481520 ∘ N -2.237881 ∘ E), and near to Manchester airport (Manchester South: 53.369026 ∘ N −2.243280 ∘ E). Figure 5 illustrates the importance of being able to make numerous measurements of O 3 at a high and temporal resolution for process studies and air quality model validation, because while the AURN sites provide a well-calibrated and consistent dataset they are only able to inform on concentrations of O 3 in their immediate vicinity, at a relatively poor temporal resolution (1 h).Furthermore, as shown in Figure 5(c), the AURN stations show rising ozone concentration throughout this day, especially between the two periods of measurement used in this study (green boxes in Figure 5(c)).Unlike at the AURN sites, the Skywalker measurements do not show a consistent increase in O 3 concentration between the midday and mid-afternoon measurement periods.These large local differences and gradients in species such as O 3 illustrate the need for dense sampling in order to inform and validate air quality models on the sub-urban scale.
Back trajectories can serve to provide information about the general airmass history of air sampled in a particular location.To probe the history of the air masses encountered by the Skywalker system, we used multiple single-particle 3D (vertical motion enabled) back trajectories from the offline Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT) model (Draxler and Rolph, 2003), initialized using the National Centre for Environmental Prediction (NCEP) reanalysis meteorological wind fields at 1 ∘ spatial resolution, at 200 m vertical intervals between the ground and 2 km.48-h back trajectories (with half-hourly outputs) were calculated with endpoints corresponding to the Hough End Fields site (Figure 7).We would expect that the observed variability in O 3 over the relatively short timeframe of the measurements in this study would be driven more by local emissions, chemistry, and dynamics than regional scale transport, with the trajectories shown in Figure 7 demonstrating that the prevailing wind direction at the time of measurement was approximately south-easterly (consistent with observations noted at Hough End Fields at the time of flight).This suggests that air sampled at Hough End would have passed near to the Manchester South AURN site; however, the poor sampling resolution (1 h) of the AURN site (and the lack of information between the two sites) means that we are unable to comment on any localized chemistry, although this might be possible in the future using simultaneous measurements of other active chemical species with UAVs.The use of the UAV also enabled a spatially dense map of surface-level and lower boundary layer O 3 concentrations to be generated, as shown in Figure 8.It should be noted that not all of the data shown in Figure 5 is plotted in Figure 8, as there was a period of time for which the GPS data was not recorded due to a logging fault.The red colours (larger symbols) in Figure 8 show the enhancement in O 3 associated with the change of wind direction noted earlier.This highlights the potential for sampling 3D variability that may enable micrometeorological process analysis in the future using simultaneous measurements of thermodynamics and other trace gases from UAVs.

Conclusions
This work has presented the design, data, and analysis of a Skywalker UAV system incorporating an ozonesonde and thermodynamic measurement capability.These measurements demonstrate the capabilities of UAVs to make high-density, 3D measurements of tropospheric O 3 , and (more widely) other trace gases on a sub-urban scale.The measurements recorded in this study demonstrate the localized spatial and temporal variability that exists in urban environments, and the difficulty that traditional ground-based instrumentation has in capturing these gradients at sufficient resolution.
Simultaneous measurements of pollutant trace gases are necessary to inform and validate air quality models to provide real-time data for monitoring, and to alert the public to dangerous levels.UAVs equipped with such a suite of instruments could conceivably help to provide such a network, giving vertical as well as ground-based information, and enabling a much clearer 3D picture to be developed.

Figure 3 .
Figure 3.A picture of the integrated ozonesonde and Skywalker airframe at the Hough End Fields site.

Figure 5 .
Figure 5. (a) Measured O 3 concentrations from 12:25 until 12:50.UAV measurements are given in black, Piccadilly Gardens AURN site measurements in red, and Manchester South AURN site measurements in blue.(b) Same as (a) but for 15:10 until 15:40.(c) AURN site measurements for the 28 June 2012 between 01:00 and 23:00.The Piccadilly Gardens site is shown in red, and the Manchester South site in blue.The green boxes indicate the data shown in (a) and (b).

Figure 7 .
Figure 7. Forty eight-hour ensemble 3D back trajectories from the HYSPLIT Lagrangian model, ending at Hough End Fields at midday on 28 June 2012, initialized at 200 m vertical intervals between the ground and 2 km.

Figure 8 .
Figure 8. Plot of O 3 along the latitudinal flight track of the UAV.For ease of reference, the sizes of the plotted circles are proportional to the O 3 concentration.

Table I .
Accuracy and precision of the Vaisala RS92-KE radiosonde, when operating from 1080 to 100 hPa.

Table II .
Specifications of Skywalker UAV.