Studying boundary layer methane isotopy and vertical mixing processes at a rewetted peatland site by unmanned aircraft system

Abstract. A proof of concept study was performed to demonstrate the capabilities of quadrocopter air sampling for analysing the methane isotopic composition in the laboratory. Boundary layer mixing processes and the methane isotopic composition were studied with a quadrocopter system at Polder Zarnekow in Mecklenburg–West Pomerania in the North East of Germany, which has become a strong source of biogenically produced methane after rewetting the drained and degraded peatland. Methane fluxes are measured continuously at the site. They show high emissions from May to September, and a strong diurnal variability with maximum methane fluxes up to 2 μmol m−2 s−1 for summer 2018. For two case studies on 23 May 2018 and 5 September 2018, vertical profiles of temperature and humidity were recorded up to an altitude of 650 m and 1000 m, respectively, during the morning transition. Air samples were taken at different altitudes and analysed in the laboratory for methane isotopic composition. The values showed a different isotopic signature in the vertical distribution during stable conditions in the morning (−51.5 ‰ below the temperature inversion at an altitude of 150 m on 23 May 2018 and at an altitude of 50 m on 5 September 2018, −50.1 ‰ above). After the onset of turbulent mixing, the isotopic signature was the same throughout the vertical column with a mean value of −49.9 &pm; 0.45 ‰. During the September study, water samples were analysed as well for methane concentration and isotopic composition in order to provide a link between surface and atmosphere. The water samples reveal high variability on scales of few 10 m for this particular case. The airborne sampling system and consecutive analysis chain were shown to provide reliable and reproducible results for two samples obtained simultaneously. The method presents a powerful tool for constraining the origin of methane by analysing its isotopic signature, and for measuring the vertical distribution of methane isotopic signature, which is based on mixing processes of methane within the atmospheric boundary layer.



Introduction
Methane's (CH 4 ) global warming potential is 32 times higher than carbon dioxide (CO 2 ) on a century timescale (Etminan et al., 2016), and 72 times higher on a decadal time scale (Solomon et al., 2007) producing a near-future, greater overall impact on the atmospheric radiative balance (Solomon et al., 2007). As a result, CH 4 regulation has a far higher potential for nearterm climate change mitigation than CO 2 . Moreover, global warming feedbacks and rising anthropogenic emissions will likely 5 increase CH 4 emissions (Wunch et al., 2009). Yet, current knowledge of CH 4 sources remains inadequate. This is in part caused by the complex processes of production, transformation and transport, and in part caused by the lack of sufficiently accurate assessments of the vertical distribution of methane and the temporal and spatial behaviour of highly variable anthropogenic and natural CH 4 surface fluxes (Solomon et al., 2007). Global mean CH 4 concentration increased until the early 1990s, then mostly stabilized for about a decade (Dlugokencky et al., 2003). Since 2007, growth has resumed (Dlugokencky et al., 2013), which 10 is visible in the remote polar areas as well (Nisbet et al., 2016). The stabilization was proposed to have been a consequence of reduction in methane emissions from Siberian gas fields and other sources inside the Soviet Union after its collapse, changes in rice agriculture, and changes in microbial emissions (Dlugokencky et al., 2009). The recent increase has been proposed to be related to enhanced emissions from tropical wetlands associated with the extremely wet season in 2009 in the Amazon region (Dlugokencky et al., 2009;Chen et al., 2010). 15 For separating natural and anthropogenic emission, the methane isotopic signature δ 13 C ‰ is of particular interest: Biologically produced methane has a typical δ 13 C ratio of -70 to -55 ‰ (France et al., 2016;Nisbet et al., 2016), methane from fossil fuels has a typical value of -55 to -25 ‰ ( Kirschke et al., 2013), and methane from fires has values of -25 to -13 ‰ (Kirschke et al., 2013;Nisbet et al., 2016). Further, smaller differences in the isotopic signature are observed for different geographic regions.
The backgound value of air in the free troposphere currently has a value of around -47.4 ‰ at northern latitudes (Nisbet et al.,20 2016). The isotopic shift towards depleted (lower, more negative) δ 13 C values provides evidence of predominantly biogenic origin (Nisbet et al., 2016).
The origin of methane and the importance of different natural sources is under discussion for various locations worldwide.
Known sources in northern latitudes are permafrost areas (Sachs et al., 2010;France et al., 2016;Sasaki et al., 2016;Kohnert et al., 2018), the Arctic ocean (Yu et al., 2015;Mau et al., 2017) and wetlands (Bellisario et al., 1999). Each source region has a 25 unique isotopic signature, and mixing of different air masses results in a linear combination of the corresponding δ 13 C signature (France et al., 2016). In the Arctic, inter-annual shifts in the sea ice drift patterns generate an inter-annually patchy methane excess in polar surface water and methane efflux (Damm et al., 2018). The origin of this enhanced methane concentration and the exchange processes between ocean, atmosphere and sea ice are subject to current investigation (Mau et al., 2017;Uhlig and Loose, 2017;Platt et al., 2018). The exchange of gas between air and sea strongly depends on the water stratification 30 (Andersson et al., 2017). Further, isotopic fractionation towards depletion of δ 13 C in the range of few to several ‰ is observed at the water-air interface for diffusion processes (Happell et al., 1995) In this context, unmanned aerial systems (UAS) fill an observational gap for methane mixing processes. They are able to sample small scales with a typical horizontal distance of 1 km, if they are required to be operated in the line of sight, and they reach the top of the atmospheric boundary layer, with a maximum altitude of typically around 1 km. First applications of measuring the methane concentration with UAS have been demonstrated: The air sampling inlet integrated into multirotor systems is either directly connected to the ground-based methane analyser via a sampling line (Brosy et al., 2017), or the air is stored in a tubing, which is analysed after the flight with a cavity ring down spectrometer (Andersen et al., 2018). The limiting element for both techniques is the length and weight of the sampling line or tube, and sampling altitudes up to 50 m have been published 5 (Brosy et al., 2017;Wolf et al., 2017;Andersen et al., 2018). An air sampling concept based on filling evacuated stainless steel containers by remotely opening a valve and subsequent chemical analyses of trace gases and first applications on multicopter systems have been shown (Chang et al., 2016(Chang et al., , 2018. In-situ methane analysers small enough to be integrated on a UAS have been presented (Gurlit et al., 2005;Miftah El Khair et al., 2017;Graf et al., 2018), and first scientific articles showing field measurements of in-situ multicopter borne methane concentrations up to an altitude of 600 m have been published (Golston et 10 al., 2017). The method of air sampling and methane isotopic analyses has been applied to (manned) airborne measurements with high payload capacity in the lower troposphere, e.g. above Siberian peatlands to distinguish emissions from fossil and biogenic emissions (Umezawa et al., 2012), up to balloon borne observations to study sink mechanisms of methane in the stratosphere (Sugawara et al., 1997).
To contribute to constraining the source and vertical mixing processes of methane at different locations, a quadrocopter borne 15 sampling system has been developed. The aim is in-situ air sampling for obtaining the vertical distribution of the methane isotopic ratio in the lower troposphere (up to 1 km) for subsequent laboratory analyses. The goal of the study is two-fold: On the one hand to show the reliability of the measurement chain from the airborne sampling to the laboratory methane isotopic analyses, on the other hand to provide new insights into small-scale methane dynamics at the study site. In order to test the system's capabilities of providing reliable vertical profiles of the isotopic signature, measurements were performed at a rewetted 20 peatland site, Polder Zarnekow (Zerbe et al., 2013), which is known as a source of biologically produced methane. Under stable stratification, it is expected that the methane isotopic ratio is depleted for this source, whereas above the temperature inversion, an isotopic ratio typical of mixed air masses is expected. During the morning transition, when the stable stratification is gradually replaced by a convectively mixed atmospheric boundary layer, the isotopic signature should adjust to a consistent value throughout the profile within the uncertainties of laboratory air sample isotopic analyses. 25

Methods
In the following, the quadrocopter ALICE (Airborne Tool for Methane Isotopic Composition and Polar Meteorological Experiments) as the carrier system, the payload consisting of the air sampling subsystem and the meteorological sensors and data acquisition are described. Further, the laboratory air analysis procedures, and the measurement site for system tests are introduced. For evaluating the whole measurement chain, a local source of methane of particular isotopic composition, and 30 atmospheric conditions that first inhibit and then enforce vertical mixing processes (morning transition) were required. These conditions were met at Polder Zarnekow on 23 May 2018. For confirming the results, the same measurement strategy was applied to the same site on 5 September 2018.

Quadrocopter ALICE and instrumentation
The quadrocopter ALICE ( Fig. 1) was designed to carry meteorological sensors and 12 glass bottles for air sampling. It has dimensions of 1.56 m x 1.56 m x 0.38 m without scientific payload. The arms are quickly removable (rotating lock) for convenient storage and transport. It was developed for a maximum take-off weight of 25 kg. For the operations presented here, the total weight was 19 kg, with 4 kg the quadrocopter system itself, 7.2 kg of LiPo batteries with a total capacity of 21 Ah and a The quadrocopter is constructed with a thrust-to-weight ratio of 2:1. The system was designed for wind speeds up to 35 km h −1 10 during take-off and landing, and up to 70 km h −1 during free flight. At a wind speed of 60 km h −1 , a climb rate of the current system of 8 m s −1 was still possible. ALICE is electrically powered with four motors U11 KV120 of T-Motor, China.
As the system was intended for operations in the polar regions, the design point of the system is -30 • C. All parts have been tested extensively in a climate chamber down to -30 • C, taking into account vibrations of the carrier, which were simulated with a shaker after in-flight measurements. The batteries are insulated passively, and the temperatures of the batteries and the Further, standard long-term stable sensors were used (Pt1000 "Humicap" HMP110, Vaisala, Finland, digital sensor TSYS01, Measurement Specialities, US). Relative humidity is measured with two different sensors, the Humicap HMP110 (Vaisala, 20 Finland), and the Rapid P14 (Innovative Sensor Technology, Switzerland). Absolute pressure is recorded with AMSYS 5812-0150-B sensors, AMSYS, Germany. Two pyranometers ML-01 of EKO Instruments, Japan, with nadir and zenith viewing geometry are integrated, which allow to estimate cloudiness and surface properties at a sampling rate of 100 Hz. Further, a surface temperature sensor Melexis MLX90614, Belgium, is mounted into the airframe and fixed at the bottom of the UAS. A Global Navigation Satellite System (GNSS) receiver and an inertial measurement unit (IMU) iµVRU of iMAR, Germany, are 25 integrated. Data is recorded at a sampling frequency of 100 Hz. The air sampling system consists of 12 glass flasks (sample containers) evacuated before take-off. They are equipped with two manual valves. Directly before the mission, each glass flask is linked with a vacuum pump RE5 of Vacuubrand, Germany. One valve is left open, and an electromagnetic valve is connected, which is normally closed. Then the flask is evacuated, and the pressure is controlled by a pressure sensor integrated in the electromagnetic valve. The flasks are opened during the flight with 30 magnet valves, that are triggered either manually by remote control or automatically at altitudes predefined by the operator.
After triggering, ambient pressure is reached within less than 2 s. The pressure sensors integrated in the valves are used to monitor air tightness. The most critical point were the manual plastic valves deployed routinely for the glass flasks in open position. They had to be treated individually and controlled to make sure that no leakage occurred during the mission. For quality control and redundance, two glass flasks were filled simultaneously, resulting in six possible sampling altitudes during one flight. The whole quadrocopter system with technical details and instrumentation is shown in Fig. 2.
The data are displayed at the ground station. Depending on the atmospheric structure, the operator decides on the altitude of taking samples during the descent, e.g. above/below the temperature inversion, or within altitudes of enhanced humidity, as required for the scientific question. The evacuated flasks are opened by electromagnetical valves via telemetry command at  15 The δ 13 C methane signature in air samples was analyzed using a Delta plus XP mass spectrometer combined with a combustion oven, a gas pressure interface and a pre-concentration device (PreCon) (ThermoFinnigan, Bremen, Germany). All valves and traps are automatically operated by the mass spectrometer software. Following the quadrocopter mission, the sample containers (SC) were transported to the laboratory at the Alfred Wegener Institute in Bremerhaven, Germany. Each SC was installed onto the PreCon and the connections were flushed with helium before the SC was opened. The air sample was first carried by helium 20 carrier gas through a chemical trap and then trapped in a cool box filled with liquid nitrogen (-196 • C) to remove CO 2 , CO and H 2 O. Afterwards the CO 2 free air was carried by helium carrier gas to the combustion oven (1000 • C) for methane conversion into CO 2 which then was purged into a second cool box filled with liquid nitrogen and trapped therein. When the combustion is finished, the gas stream is purged and trapped in a third cool box filled with liquid nitrogen to pre-concentrate the sample, and carried by helium carrier gas into the isotope ratio mass spectrometer. Methane stable carbon isotope data are given in δ 25 notation (in ‰) relative to the Pee Dee Belemnite (PDB) standard: along a transect between the EC system and the shore of the lake for spatially resolved CO 2 and CH 4 flux investigations (Hoffmann et al., 2017).

10
The restoration of the peatland area towards a net sink of greenhouse gases, and in particular CH 4 , is a process of several years to decades, which strongly depends on vegetation and the water level (Couwenberg et al., 2011;Zak et al., 2015).  3 Results

Synoptic conditions and meteorological observations
On 23 May 2018, the synoptic situation was characterized by a pronounced high-pressure system above Scandinavia and the 10 Baltic Sea, leading to conditions of low wind speed below 5 m s −1 , north-easterly wind directions and a cloudless sky, as evdenced from the smooth behaviour of global radiation (Fig. 5).
On 5 September 2018, the synoptic situation was determined by two strong high pressure systems located above the Atlantic Ocean and above Northern Russia, and a low-pressure system above Southern Europe. This resulted in low wind speed below 4 m s −1 of north-easterly direction at the observation site, similar to the conditions on 23 May 2018. In the morning, until 05:30 15 UTC, around 30 min before the first flight, fog was observed, which was denser towards the East. Starting around 09:00, before flight 3, shallow convective cumulus clouds were present. This is also evident in the high variability of the global radiation ( Fig. 6).
On 23 May 2018, the near-surface temperature increased from around 13-14 • C at 6 UTC to around 17 • C at 08:30 UTC (Fig.   7). The profiles of potential temperature show a strongly stable stratification in the morning, with an increase of potential tem-20 perature of around 2 • C (06:04 UTC) from the surface to 250 m. The top of the nocturnal temperature inversion was located at around 150 to 250 m altitude. At this altitude range, the water vapour mixing ratio decreased as well from around 7 g kg −1 in the boundary layer close to the surface to around 5 g kg −1 above. After 07:30 UTC, the atmosphere was less stably stratified.
The water vapour mixing ratio was enhanced in the lowermost 150 m at 06:04 UTC. Above and later, the water vapour mixing ratio was almost constant with height. 25 On 5 September 2018 during the first flight, the profiles show a mixed-layer up to 50 m altitude and a stable stratification above (Fig. 8). The water vapour mixing ratio was lower in the mixed layer (11 g kg −1 ) and increased above up to 14 g kg −1 . For the following four flights, the near surface temperature increased, and the potential temperature increased only slightly with altitude. The water vapour mixing ratio decreased with time throughout the whole profile.

Observations of methane emissions and isotopic signature
For both observation days, the methane concentration recorded at the eddy covariance tower at 2.6 m altitude above the surface shows a high variability during the night and the morning, and a relatively constant concentration during the day. During the time period of the first flight on 23 May 2018, the methane concentration was still slightly enhanced compared to the background value during the day (Fig. 9). The first flight on 5 September 2018 took place during the high fluctuations of the 5 CH 4 concentration (Fig. 10).
The isotopic signature of the air samples was different during the first flight of each day at the altitudes located within the stable stratification. On 23 May 2018, the isotopic ratio was depleted at 10 m and 100 m altitude with a value of -51.5‰ (see Fig. 11).
Above, the mean isotopic ratio was -50.1‰, almost the same as during the next flights throughout the profile, with an average value of -49.9‰. On 5 September 2018, the same systematic behaviour was observed: During the first flight, the isotopic ratio 10 was depleted at an altitude of 10 m, in agreement with a temperature inversion above an altitude of 50 m. Above, the isotopic value of -49.9‰ was the same as during all other flights throughout the profile (Fig. 12).
Aerial pictures obtained during the measurement flights for the two case studies show the difference in water level: On 23 May 2018, the lake was filled with water ( Fig. 13). On 5 September 2018, almost the whole lake had fallen dry, with only few wet areas remaining, and the sediment still saturated with water (Fig. 4). The depth was in the range of few cm.

15
The water samples taken on 5 September 2018 within a radius of 100 m revealed highly different CH 4 concentrations (see Table 1). The highest CH 4 concentration of around 4770 ppm was measured from the water sample Z-4 in one of the small remaining water areas with a depth less than 5 cm within the former polder. Directly next to this, the water sample Z-3 taken at a depth of 5-10 cm had a methane concentration of 2310 ppm (see Figs. 4 for exact locations). The water sample Z-4 contained more suspended sediment load than sample Z-3. The sample sites are located within small remaining water areas, which are 20 not connected.

Plausibility of the observed isotopic signature
The methane flux data indicate that the observation site was a source of methane during both measurement days, but had much higher emissions on 23 May 2018 (Fig. 14). During the first flights in the early morning, under stable stratifications, an 25 enhanced methane concentration in the ABL can be expected, as observed by Andersen et al. (2018) above wetlands. This is the case on both measurement days with the UAS. An enhanced methane concentration and high variability was observed until around 06:30 UTC on 23 May 2018 (Fig. 9), and even higher concentrations and stronger fluctuations were observed until 6:45 UTC on 5 September 2018 (Fig. 10). Under stable atmospheric conditions during the night and in the early morning hours, the near-surface methane concentration is enhanced, as no vertical mixing is possible, e.g. Emeis (2008); Brosy et al. (2017). 30 With a local methane source, the near-surface methane concentration above agricultural land therefore increases during stable atmospheric conditions (Wolf et al., 2017), as this prevents dilusion with the free troposphere. However, a high local horizontal inhomogeneity in methane concentrations resulting from mixed-use areas have been reported, an effect which is visible in the lowermost 10 m above the surface, but smeared out above due to an increase in horizontal wind speed (Wolf et al., 2017). A high variability of methane sources is in agreement with the highly variable methane water concentrations measured within a radius of 100 m on 5 September 2018.
The isotopic difference obtained for air samples near ground and above the temperature inversion during stable stratification is 5 around 1.5 ‰, thus significantly higher than the uncertainties (flight 1 for each measurement day). This shows that the observed systematic differences are real measurement features, not measurement uncertainty.
Assuming that the parts of the surface with high methane concentrations, like sample Z-3, Z-4 and Z-6, act as a methane source, with an isotopic ratio of around -48 to -49 ‰ (water sample Z-4 and Z-6), an isotopic fractionation of around 3 ‰ would occur across the water-air boundary. This is in the order of magnitude of carbon isotope depletion observed for emitted relative to 10 floodwater CH 4 , ranging from 1.8 to 3.4‰ (Happell et al., 1995), thus can be considered realistic.

Interpretation of isotopic signature from measurement point of view
Generally, the isotopic ratios of the two air samples taken simultaneously but with a constant horizontal distance of 13 cm at the lowest altitude of 10 m agree to within 0.2 ‰ ( Fig. 11 and Fig. 12). However, during flight 1 on 23 May 2018, parallel to the high variability of the methane concentration, and flight 5 on 5 September 2018, the isotopic signature of the two air samples 15 taken simultaneously at 10 m altitude reveal a significant isotopic difference of 0.75‰. As the repeatability of the isotopic ratio of samples taken within identical air masses is 0.5‰, this difference is larger than the uncertainty, and therefore the systematic differences are treated as features. There are several striking features in the profiles of isotopic signatures: -The isotopy ratio has more negative values in the morning before vertical mixing starts, as long as a temperature inversion is present (first flight on 23 May 2018 below 150 m, and first flight on 5 September 2018 below 70 m). This is in 20 agreement with methane from biologic processes emitted from the surface that are not vertically mixed.
-The difference in isotopic ratio between the two simultaneous samples is systematically smaller at 10 m altitude compared to the higher altitudes on 5 September 2018, except the last profile.
-The differences in isotopic ratio at 100 m altitude increase during the course of the day on 5 September 2018.
The hypothesis is that the air samples obtained at 10 m altitude are influenced by air masses representative of a smaller, 25 thus more homogeneous area. At an altitude of 100 m, air masses are influenced by a similarly small area, as long as stable stratification is present. With increasing turbulent mixing, a different footprint influences the air masses, and small-scale inhomogeneities can impact measurements only 13 cm apart. Episodic CH 4 ourbursts on short time scales of few min have been observed by Schaller et al. (2018). The high spatial and temporal variability of methane emissions reported here confirm their observations.

Conclusions
The measurements serve both as a proof of concept for the system and show the vertical mixing of methane in the ABL by means of its isotopic signature. With ALICE air samples and subsequent laboratory analyses, it is possible to determine differences in the methane isotopic signature in agreement with atmospheric stability. For the future, a combination of in-situ high resolution methane concentration measurements and air sampling for isotopic analyses is required for substantial progress 5 in understanding vertical exchange and trace back methane concentrations to specific sources and sinks in the atmospheric boundary layer.
The differences in isotopic values of water and air, the differences in isotopic values between both flight days and the development during each day emphasize the highly complex and inhomogeneous nature of methane processes on horizontal scales below 1 km in sediments, at the sediment-water and the water-atmosphere border. Therefore, the representativeness 10 of individual methane flux measurements have to be treated with care, and footprint analyses are required to extrapolate the measurements on a larger area. Vertical layering of air masses with different methane properties strongly depends on atmospheric stability, both concerning concentration as well as the isotopic ratio. A holistic approach is needed to approach methane processes from sediments to the atmospheric boundary layer, including dedicated measurements of the isotopic fractionation.