Revisiting the passive biocover system at Klintholm landfill, six years after construction

A biocover system was established at Klintholm landfill in Denmark in 2009 to mitigate methane emissions, and the system exhibited high mitigation efficiency during the first year after implementation. The biocover system was revisited in 2016/2017, and a series of field and laboratory tests were carried out to evaluate functionality about six years after establishment. Three field campaigns were executed in three different barometric pressure conditions, namely increasing, stable and decreasing. Local surface flux measurements and gas concentration profiles in the methane oxidation layer showed that barometric pressure changes had a significant effect on gas emission and methane oxidation. Elevated concentrations of oxygen were observed in the gas distribution layer, and field data showed that significant methane oxidation took place in this location. This finding was verified in laboratory-based methane oxidation incubation tests. Temperatures higher than ambient temperature were observed throughout the methane oxidation layer, with average temperatures ranging between 13 and 27 ◦ C, even in the coldest month of the year. Field measurements showed that total methane emissions from the whole landfill cell were at the same level or lower than measurements performed in 2009/2010 after implementation of the biocover system, and laboratory tests showed methane oxidation potential approximately equal to former tests. In spite of an inhomogeneous distribution of landfill gas load to the methane oxidation layer, the performance of the biocover system had not declined over the 6 – 7 years since its establishment, even though no maintenance had been carried out in the intervening years


Introduction
Landfills constitute a significant source of atmospheric methane (CH 4 ), which in turn contributes to climate change (Blanco et al., 2014).At some landfills, the CH 4 in the landfill gas (LFG) is notor cannot beutilised, and the LFG is either flared at the risk of producing toxic combustion products or just emitted into the atmosphere.As an alternative gas emission mitigation option, biocover systems may be established, which use biological active materials to support microbial CH 4 oxidation (Huber-Humer et al., 2008;Scheutz et al., 2009).
The implementation of full-scale biocover systems at landfill sites has been reported in the literature for about 10 different landfills (an overview is given in Kjeldsen and Scheutz, 2019).Only in two of the 10 cases (Klintholm and Faxe landfills, DK) was CH 4 oxidation efficiency documented by measuring the entire sites' emissions, using the mobile tracer gas dispersion method (Mønster et al., 2014).In these two cases, whole-landfill emission measurements were performed prior to and after the establishment of the biocover system, to quantify mitigation efficiency (Scheutz et al., 2011a;2014).The full-scale biocover systems reported in the literature all consist of a gas distribution layer (GDL) overlain by a methane oxidation layer (MOL).Nearly all of them use compost as an active material for the MOL and as an alternative to using permeable soils.Some of the advantages of using compost as active CH 4 oxidation material are a) the large surface area of the compost, which supports bacterial growth, b) a high gas permeability, even at elevated water levels, and c) the ability to maintain elevated temperatures due to the high insulation properties of compost, in combination with heat generation via compost respiration and CH 4 oxidation, which both are exothermic processes (Scheutz et al., 2014).There is also an economic advantage, since compost often is available as a low cost material.A potential drawback is that compost is a living material consisting primarily of organic particles, which break down continuously.Consequently, this may a) change physical-chemical properties (specific surface area, nutrient content, etc.), potentially leading to a lower CH 4 oxidation rate, and b) result in lower porosity, permeability and diffusivity, thus leading to unfavourable transport conditions for the reactants (CH 4 and oxygen (O 2 )).In addition, at O 2 depleted conditions, the compost may break down anaerobically to CH 4 .Due to the novelty of the technology, there is limited knowledge about the long-term efficiency and maintenance needs of full-scale biocover systems, but these parameters are nevertheless important when assessing overall costs.Previously, the efficiency of a six-year-old experimental biocover at St-Nicéphore landfill in Quebec, Canada, has been investigated (Capanema et al., 2013).This study showed significant CH 4 oxidation of the six years old system; however no comparison to initial performance was included.
Mineral-based material, such as permeable soils has been used as CH 4 oxidation layers in a few full-scale systems as an alternative to compost (Gebert et al., 2015;Gebert and Streese-Kleeberg, 2019;Geck et al., 2016).Soils have mainly been used in these systems to avoid the significant temporal changes in material properties, which can be foreseen for compost.
In 2009, a full-scale biocover system was established at the oldest waste cell at Klintholm landfill, Denmark (Scheutz et al., 2014), as part of its closure plan.The system was designed based on a conceptual model of gas emission patterns established through an initial baseline study.The biocover involved the construction of gas collection trenches along the slopes of the landfill cell, where the majority of the CH 4 emissions occurred.Local compost was used as bioactive CH 4 oxidising material.Whole-site CH 4 emission quantifications, based on combined tracer gas releases on site and downwind measurements, in combination with several site investigation activities (e.g.gas composition within biocover layers, surface flux chamber-based emission measurements and measurement of compost temperatures), proved that the biocover system had an average mitigation efficiency of approximately 80% based on the difference between two whole-site CH 4 emission quantifications prior to the installation of the biocover system, and five quantifications after establishment of the biocover system.The study showed that the system also had high efficiency during winter periods when ambient temperatures dropped to below freezing, due to significantly elevated temperatures in the MOL (typical 10-30 • C above ambient temperature with increasing temperatures with depth).
The biocover system established at Klintholm was revisited in 2016/ 2017, with the objective to evaluate its functionality 6-7 years after its establishment.The landfill owner has not been met with any regulatory monitoring requirements after the initial project period.As part of the revisit, field and laboratory activities were carried out to study CH 4 emission rates and oxidation efficiency, in order to compare these new results with the initial findings taken at the time of initial implementation.Table 1 provides an overview of the performed field activities.This study presents an evaluation of a 6-7 years old compost based biocover system founded on comprehensive and systematic field monitoring and laboratory test and represents an important contribution to the limited -but much in need -scientific literature on long-term performance of full-scale biocover systems.

Site characterisation and previous investigations
Klintholm landfill, located on the Danish island Funen, was established in 1978 and consists of several waste cells, some of which still receive waste.The oldest cell is Cell 0, which was in use from 1980 to 1996.It has an estimated waste volume of 485,000 m 3 and contains a mix of combustible waste, sludge and non-combustible waste.Cell 0 was established without bottom liner, leachate collection system or gas recovery system.More information about Klintholm landfill including design of the biocover system and material and methods used is provided in Scheutz et al. (2014).In 2009, a biocover system was established at this particular cell, consisting of gas collection trenches constructed on its slopes to route the collected gas up to nine biofilter sections with a GDL consisting of crushed, sieved concrete (30 cm depth, grain size range: 20-60 mm) situated beneath an MOL consisting of compost (70 cm).The compost originated from a mix of composted garden waste and kitchen waste.The total surface area of the biocover sections is approximately 4800 m 2 .Due to the nature of the material, the permeability of the GDL is expected to be much higher than the permeability of the MOL.Gas was collected from the slopes of the waste cell through dug trenches, routed by natural convection to biofilters constructed on top of the cell and distributed into the biofilters through a horizontal gas distribution pipe placed in the GDL (cf.Scheutz et al., 2014 for further details on the design of the biocover system).No maintenance was carried out in the years between its establishment (September 2009) and the time of revisiting the site (March 2016 -January 2017).After the biocover system was established, activities were initiated to evaluate its performance, including total CH 4 emission measurements and detailed investigations of selected sections to measure gas surface fluxes, pore gas compositions and temperatures in the MOL (Scheutz et al., 2014).

Materials and methods
In order to meet the objectives of the study, several field investigations and laboratory experiments were conducted.The landfill was visited on three occasions, namely in March 2016, April 2016 and January 2017.Based on the original study presented by Scheutz et al. (2014), and an initial overall screening and inspection (section 3.1.1),three biocover sections (2, 4 and 7) were selected for detailed field and lab studies.

Field evaluation of the biocover system's performance
Table 1 provides an overview of the field activities performed in the three campaigns.Details on meteorological conditions during the campaigns can be seen in the Supplementary Material -Part 1 (SM1) together with the exact timing of the field activities.The campaigns were timed to represent different scenarios in respect to barometric pressure conditions, and they also included a scenario with a low atmospheric temperature, which may influence the activity of methanotrophic bacteria and make the biocover system less efficient.However, experiences from previous studies of biocover systems using compost as MOL material have shown that temperature in the MOL is significantly elevated in comparison to ambient temperature at all times of the year (Scheutz et al., 2011a;2014;2017).Therefore we believe that changes in barometric pressure is a much more important factor influencing gas transport and CH 4 oxidation than season.Barometric pressure increased during the March campaign, whilst it remained stable in the April campaign.In the January campaign, pressure decreased (cf.SM1).As part of the first campaign, a visual inspection of the biocover surface was carried out with the aim of evaluating vegetation and any potential decay or irregularities at the biocover.Based on the inspection, as well as previous observations and measurements, three sections (2, 4 and 7, shown in Fig. 1) were chosen for detailed investigations.In Scheutz et al. (2014) detailed investigations were carried out on sections 2, 3 and 4. Due to dense vegetation detailed studies could not be repeated in section 3, so section 7 was chosen instead.In each of the three sections, a test plot consisting of 16 measuring points in a 4 × 4 grid was established (named S2, S4, and S7) with typical spacing between measuring points of 110 cm to 200 cm (see Fig. 2).In section 4 and section 7, the test plots were placed in areas with scarce vegetation, as this could be a sign of the occurrence of emitting LFG.Vegetation damages are signs of elevated CO 2 concentrations and low O 2 concentrations in the MOL resulting from emitting LFG.The visual inspection of the three test plots did not reveal any surface cracks.

Surface methane screening and visual inspection
Surface CH 4 screenings were carried out at all nine biocover sections during the first campaign (March 2016) with increasing barometric pressures, and in the third campaign (January 2017) with decreasing barometric pressure.Near-surface concentrations of CH 4 were measured using a handheld CH 4 detector (Laser One, Huberg, Bolzano, Italy).The measurement principle of the instrument is tuneable diode laser absorption spectroscopy.The instrument is calibrated from factory.Measurement range of CH 4 is 0.5 ppm to 100% by volume.A nozzle attached to the CH 4 detector allowed for sampling close to the ground.The nozzle was held five centimetres above ground.The entire biocover area was covered by walking across each section in straight lines approximately one metre apart.When elevated concentrations were observed, the surrounding area was screened as well, to evaluate the size of the emitting spot.A few areas (approximately 10%) of the landfill surface were omitted, due to dense vegetation.

Thickness of the methane oxidation layer
The depth of the CH 4 oxidation layer was measured for all three test plots at each of the 16 measuring points using a steel probe, which was pushed into the layer until it was not possible to push any further by hand, indicating that the tip had reached the GDL.

Gas composition in the gas distribution system and the interior of the landfill
Gas was sampled from the existing vertical gas sampling pipes connected to the GDL in the chosen three biofilter sections, in order to determine gas composition in the GDL.The vertical gas sampling pipe is connected to the horizontal gas distribution pipe located in the GDL of each biofilter section (cf.Fig. 1).This horizontal pipe distribute gas from several of the trenches dug into the landfill slope (cf.Fig. 1).More details are given in Fig. 3 in Scheutz et al., 2014).Measurements were carried out during the March and April campaigns.To evaluate the composition of the raw gas in the interior of the landfill, an existing gas well was sampled in the March campaign (see Scheutz et al. (2014) for exact location) and analysed for the main components (CH 4 , carbon dioxide (CO 2 ), nitrogen (N 2 ) and O 2 ).To measure gas concentrations, a vacuum Fig. 1. Cell 0 seen from above.The grey boxes indicate the biocover sections.Gas collection pipes can be seen on the slopes of the landfill, feeding landfill gas to the biocover sections through slotted piping (shown as two parallel dashed lines).Blue dots represent the approximate locations of the three test plots and associated measuring points, though the area they cover is exaggerated.Orange dots represent the approximate position of the excavation sites.(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)pump was first used to purge the well and the vertical gas sampling pipes, before samples were taken and analysed using a Biogas 5000 Gas Analyser (Geotechnical Instruments, Coventry, UK) recording CH 4 , CO 2 , O 2 (±0.5% accuracy) and a balance gas, assumed to be N 2 .The Biogas 5000 Gas Analyser has an accuracy of ± 0.5% and a detection limit is 0.5% for the recorded gases.

Surface methane and carbon dioxide flux measurements
Fluxes of CO 2 and CH 4 through the biocover surface were measured using static flux chambers (height 21 cm and diameter 32 cm) in all three measurement campaigns.The measurements were taken at the 16 measuring points at each of the test plots established in the three selected biocover sections, using an Innova 1312 photoacoustic multigas monitor (LumaSense Technologies A/S, Denmark).Each flux measurement consisted of a series of five consecutive concentration readings, taken every 50 s.The detection limit was ± 0.13 g m − 2 d − 1 , which is the flux of a concentration measurement series where the last reading is 2 ppm higher/lower than the first reading.For all flux measurements, the concentration-versus-time trend lines showed good linear fits (R 2 greater than 0.87) when non-detectable fluxes were disregarded.More information about the flux chamber method used herein, including details on the static flux chamber, can be seen in Scheutz et al. (2011a).

Gas concentration profiles in the methane oxidation layer
In all campaigns, gas concentrations of CH 4 , CO 2 , O 2 and a balance gas assumed to be N 2 were measured in the filter material in the 16 measuring points at each of the three test plots.At each measuring point, a stainless steel gas probe (inner diameter 6 mm), closed at the bottom and slotted over the lower three cm, was pushed into the MOL just outside the footprint of the flux chamber.The probe was connected to the Biogas 5000 Gas Analyser and pushed 10 cm vertically through the filter material to take the first measurement.The internal pump in the gas analyser (pump rate: 550 cm 3 min − 1 ) was running for 15 sec and moved pore gas from the filter material to the gas analyser.After 15 sec, a reading was taken, and the probe was pushed 10 cm further down into the filter material to take the next measurement.The short purging of 15 sec allowed that standing gas in the probe was removed, but also ensured that the gas sampled was only originating from within the ten cm depth interval.The procedure was followed until the probe hit the GDL, at which point it was not possible to push it further by hand.

Temperature profiles in the methane oxidation layer
Temperature in the CH 4 oxidation layer was measured at the 16 measuring points for each of the three test plots during the March 2016 and January 2017 campaigns.The temperature probe was placed within 10 cm from the location of the gas probe measurements.At every point, a first measurement was taken 10 cm below ground level, followed by measurements every 10 cm down until it was not possible to push the thermometer further by hand.

Quantification of total methane emissions from the biocover cell
To evaluate the overall efficiency of the whole biocover system, the CH 4 emissions from the waste cell with the biocover were measured using a tracer gas dispersion method on March 29, 2016, and January 9, 2017.This remote sensing method involves the continuous release of a gaseous tracer at the source of emission.Acetylene (C 2 H 2 ) was used as a tracer gas, and the release rate was between 0.72 and 0.79 kg C 2 H 2 h − 1 .The gas was released from a single location in the centre of the waste cell with the biocover.During the release phase, concentrations of CH 4 and C 2 H 2 were measured while traversing the plume downwind of the waste cell, using a mobile analytical platform.Assuming that CH 4 from the waste cell and the released tracer gas was well-mixed at the sample location, the ratio of measured concentrations of CH 4 and C 2 H 2 in the plume can be considered equal to the ratio of emission rates.As a result, the emission rate of CH 4 can be deduced.The method and instrumentation are described in detail in Mønster et al. (2014;2015).

Sampling and characterisation of the physical properties of biocover materials
During the March 2016 campaign, the biocover was excavated, and depth-specific samples of the MOL were collected from approximately every 10 cm below ground level (bgl).Depth-specific samples were obtained from section 2 (five depths), section 4 (five depths) and section 7 (seven depths).In January 2017, parts of the GDL were excavated and samples retrieved in each of the three sections.All material sampling was done in the vicinity of the test plots, but more than two meters away to avoid disturbance of the flow patterns in the test plots.All sampled materials were stored in a cooled room at 10 • C for less than two weeks until laboratory experiments were carried out.The MOL and GDL samples were analysed for water content and loss on ignition, following standard procedures.The dry bulk density of the material sampled from the GDL was measured by weighing a packed sample of known volume and then deducting the determined water content.

Batch incubation test to determine methane oxidation and respiration potential
The MOL and GDL samples were tested in duplicate for CH 4 oxidation potential as well as respiration activity at room temperature (22 • C).For all tests, control batches with no material added were also carried out.Methane oxidation potential tests and respiration tests for the 17 MOL samples were set up as batch incubation tests with 35 g of filter material added to a 500 mL infusion bottle sealed with a butyl rubber stopper and an aluminium screw cap.No additional water was added to the batches.For the oxidation tests, gas concentration in the infusion bottle headspace was adjusted from atmospheric conditions to approximately 15% vol.CH 4 and 35% vol.O 2 by injection of pure CH 4 and O 2 .The initial high O 2 content should assure sufficient O 2 for both CH 4 oxidation and compost respiration.To avoid an extensive lag phase, the bacteria were allowed to acclimatise to these conditions for approximately 12 h.The batches were then ventilated and the aforementioned gas concentrations recreated before the test began.The respiration tests were carried out by re-using the bottles for the CH 4 oxidation potential tests after they were terminated.For the respiration tests, gas concentration in the headspace was adjusted from atmospheric conditions to approximately 35% vol.O 2 by injection of pure O 2 .
A subset of the MOL samples was also tested for CH 4 oxidation potential at 30 • C and 40 • C. six samples in this subset were taken from sections 2 and 4 at depths of 10 cm, 30 cm and 50 cm.Testing was done to simulate the temperatures in the warmest parts of the biocover and in order to compare the results with maximum oxidation rates found in previous tests of the original compost material (as presented in Scheutz et al., 2014).
The three GDL samples were tested for CH 4 oxidation potential and respiration rates via incubation tests performed as for the material sampled from the MOL.Inorganic matter, mainly consisting of crushed concrete, was removed so that only fines, including the compost material that had migrated into the GDL over time, were tested.Another set of oxidation and respiration tests was made in five liter batches consisting of the original samples taken from the GDL (thus including both fines and crushed concrete).The larger container size was chosen to accommodate the large size of the concrete fractions.
Concentrations of CH 4 , CO 2 and O 2 in gas samples from the batch incubation tests were measured over time by manually injecting 0.2 mL gas samples extracted from the incubation bottles into two TRACE™ 1310 gas chromatographs (Thermo Fisher Scientific, Waltham, USA).Zero-order trend lines were fitted to the initial linear part of the gas concentration curves representing 90% conversion in the incubation tests, and the slopes of the trend lines were used to calculate maximum CH 4 oxidation rates and respiration rates (see Scheutz et al. (2017) for additional methodological details).

Cover inspection and methane surface screening
A visual inspection of the biocover surface revealed that after 6½ years, weeds and grass dominated the surface of the biofilter sections.Some small trees and bushes were also present.In a few areas, bare spots without or with very little vegetation were observed.Surface screening using the handheld CH 4 detector (Laser One) showed elevated CH 4 in these bare spots indicating emission hotspots with elevated CH 4 fluxes.Rodents had made tunnels in the filter material in several places, and burrows made by larger animals were also found.
Near-surface measurements showed background-level concentrations of CH 4 during the March 2016 campaign when barometric pressure increased (dP/dt ~ 0.5 hPa h − 1 ), and concentrations above 10 ppm were found only at three points (two in section 2, and one in section 3 -see Fig. SM2.1 in Supporting Material (SM)).During the January 2017 campaign, which witnessed decreasing barometric pressure (dP/dt ~ -0.65 hPa h − 1 ), concentrations above 10 ppm were found more frequently, and points with concentrations above 50 ppm were seen in sections 1, 3 and 9 (Fig. SM2.1 and SM2.2 in SM).The highest concentration, at 217 ppm, was seen in section 3, in an area where concentrations were generally above background level.
In spite of the varying pressure conditions, surface CH 4 concentrations were in general very low for all three campaigns indicating only few and minor emission hotspots.This was also the case when compared to surface screenings performed back in 2010, where larger areas with increased surface CH 4 concentrations were observed, especially near the edges of some of the biocover sections (cf.Fig. 7 in Scheutz et al., 2014).

Thickness of the methane oxidation layer
Based on measurements in sections 2, 4 and 7, the depth of the MOL was found to be 15 cm shallower on average than when the biocover system was established (see Table 2) most likely caused by natural settlement and compaction of the MOL, which had taken place over the years.In addition, samples retrieved from the MOL yielded a lower loss on ignition on average (11.6 ± 3.3 g (g DM) − 1 ) than at the establishment of the biocover system (18.8 ± 0.3 g (g DM) − 1 ) (Table 2) suggesting that some of the organic matter had been converted to gaseous CO 2 through microbial compost degradation processes.There were clear indications that microbial respiration processes continued the degradation of organic matter, since significant CO 2 fluxes were observed at locations with no CH 4 load.

Gas composition in the interior of the landfill and the gas distribution system
Gas concentrations from the LFG well in the March campaign were CH 4 50.1%,CO 2 19.8% and O 2 0.3%.The balance gas (assumed to be N 2 ) had a concentration of 29.7%.Concentrations of CH 4 and CO 2 were lower than found in 2009 (CH 4 69%, CO 2 27%), while N 2 was higher than found in 2009 (N 2 below 5%) (Scheutz et al., 2014).However, the CH 4 /CO 2 -ratio was unchanged (a value of 2.5), which indicates some dilution in the well with atmospheric air, probably as a result of increasing barometric pressure three days before sampling the well.It has previously been observed that gas compositions in LFG wells can be significantly affected by changes in barometric pressure (Aghdam et al., 2019;Fredenslund et al., 2010).However, an unchanged CH 4 /CO 2 -ratio of 2.5 indicates that changes in the degree of waste oxidation in the deeper waste layers are minimal, and have not changed over the 6½ years.
The twelve existing vertical gas sampling pipes installed in the biocover and connected to the GDL layer were measured for CH 4 , CO 2 and O 2 .The measurements showed low concentrations of CH 4 (0-5%) and CO 2 (0-12%).O 2 concentrations were high (8-21%), albeit in most cases they were a bit lower than O 2 concentrations in atmospheric air.The CH 4 /CO 2 -ratio (0.01-0.90) was much lower than for the gas sampled in the interior of the landfilled waste indicating significant CH 4 oxidation in the GDL.This was supported by field observations and lab incubation tests.Excavations showed that material from the MOL by gravity or by water had relocated into the GDL identified by a significant content of blackish fines in the upper 15 cm of the GDL.When testing the fines, i.e. without the crushed concrete, the material had an oxidation rate of 98 ± 14 µg CH 4 (g DM) − 1 h − 1 , which was much higher than otherwise seen in the deepest parts of the MOL (Table 2).This may be due to the fact that the fine fraction has a much higher specific surface area than the original GDL material, thus supporting bacterial growth better.For the five litre batch test, including all material from the upper 15 cm of the GDL, a lower CH 4 oxidation rate was found (on average 19 µg CH 4 (g DM) − 1 h − 1 -see Table 2).This was expected, as the inorganic material (crushed concrete) had a high mass but no substantial oxidation capacity.Overall, the tests revealed that the fine compost particles washed into the GDL actively oxidise CH 4 and that the GDL today may play an important role in mitigation of CH 4 in presence of O 2 (see section 5.1 for a discussion).

Methane and carbon dioxide surface fluxes measured by the flux chamber
During a period of increasing barometric pressure (March 2016), fluxes of CH 4 through the biofilter surface were observed at only a few of the measuring points on the three test plots (7 out of 48 measuring points).Positive CH 4 fluxes were observed more frequently during stable (20 out of 48 measuring points -April 2016) and decreasing barometric pressure (21 out of 48 measuring points -January 2017) events, as shown in Fig. 2, where it is also evident that the spatial and temporal variability of the fluxes was pronounced.Points with CH 4 fluxes greater than 50 g m − 2 d − 1 were found close to points with negative fluxes.One example was established during the April 2016 campaign in section 4, where the CH 4 flux at point 2 was − 0.05 g m − 2 d − 1 , while the CH 4 flux at point 7, situated a few metres away, was 115 g m − 2 d − 1 .The largest CH 4 flux was observed during the April 2016 campaign in section 2, point 1, at 872 g m − 2 d − 1 .The results for the CH 4 and CO 2 flux measurement for all points, test plots and campaigns are given in Table SM3.1 in SM.
Flux chamber measurements carried out in section 2 and section 4 in 2010 showed positive CH 4 fluxes at the edge of the biocover toward its slope, while all other measurements were below detection limits (Scheutz et al., 2014).There was no such recognisable pattern in the 2016/2017 measurements, though positive CH 4 fluxes were observed more frequently.
Carbon dioxide emissions were also highly variable in both time and space; however, significant levels were observed at nearly all measuring points in the three test plots and at all times, except for at a few points (see Table SM3.1).On a molar basis, CO 2 emissions were in all cases higher than the CH 4 emissions, which is most likely due to a combination of CO 2 generated from CH 4 oxidation and originating from respiration processes in the MOL.

Gas concentration profiles in the methane oxidation layer
When measuring gas concentrations in the MOL, CH 4 concentrations greater than 1% were only observed in 5 of the 48 measuring points during the March 2016 campaign under increasing barometric pressure conditions.In contrast, CH 4 was observed in the MOL more frequently in the April 2016 (32 out of 48 points) (stable barometric pressure) and January 2017 (21 out of 48 points) (decreasing barometric pressure) campaigns.At the majority of the measuring points, O 2 concentrations above 10% were found throughout the MOL.Examples of representative gas concentration profiles can be seen in Fig. 3, while all gas concentration profiles are presented in SM4.At locations where CH 4 were observed in the deeper part of the MOL, gas concentration profiles clearly indicated CH 4 oxidation in the above laying MOL layers (profile MO1 and MO2 in Fig. 3).Methane oxidation tests on MOL samples showed high CH 4 oxidation rates also compared to similar compost tests run prior to the establishment of the biocover system (Table 2).In general, the highest CH 4 oxidation rates were observed in samples taken from the shallower parts of the filter (Fig. SM7.1 in SM), which is in accordance with the gas concentration profiles indicating CH 4 oxidation in this zone.Other authors have shown similar results of shallow CH 4 oxidation zones (Lee et al., 2014;Scheutz et al., 2004).Many gas concentration profiles showed respiration throughout the MOL with decreasing O 2 concentration and increasing CO 2 towards the surface (R in Fig. 4).High respiration activity in terms of O 2 consumption and CO 2 production in the MOL was confirmed in incubations tests (Table 2).
When comparing gas concentration profiles, it became evident that the profile connected to one point was not necessarily indicative of neighbouring points, indicating high spatial variability.One example was the gas concentration profiles at points 7 and 8 measured in section 7 in January 2017 (data shown in SM4).Although the points were measured within a few minutes and were situated only metres apart, one profile shows no apparent CH 4 and near-atmospheric levels of O 2 , while the other shows an increasing CH 4 concentration towards the bottom of the MOL and an O 2 level that decreases to approximately zero.Similar to the surface flux observations, the gas concentration profiles showed high spatial and temporal variability.The different types of gas profiles shown in Fig. 3 are the result of several transport processes (diffusion/ advection), transformation processes (CH 4 oxidation and respiration), CH 4 cover loadings and cover properties, which are discussed further in section 5.2.

Temperature in the methane oxidation layer
At all measuring points, temperatures were lowest at the shallowest part of the MOL and highest at the deepest part at the interface to the GDL (Fig. 3).The average temperature found in the MOL was considerably higher (on average 17.5 • C higher) than the ambient temperature Fig. 4. Conceptual model of behaviour of the biocover system at Klintholm Landfill, resulting in different observed profile types.LFG is supplied to the GDL, where CH 4 oxidation takes place due to air intrusion through more permeable regions of the biofilter, driven by a downward pressure gradient created by gas volume reduction by the CH 4 oxidation in combination with barometric pressure pumping (AI).In other regions of the biofilter, we find CH 4 , CO 2 and O 2 (MO2), CH 4 and CO 2 (MO1/AE), or only CO 2 (R).A further explanation of this figure can be found in the text.
C. Scheutz et al. (see Table 3).Even in January, during the Danish winter, the average temperature showed elevated levels, which could support CH 4 oxidation processes.Elevated temperatures were also seen in the initial study (Scheutz et al., 2014), and these have also been observed in other biocover systems (Scheutz et al., 2017;Dever et al., 2011;Philopoulos et al., 2008).Heat is provided to the MOL by the landfill gas, as temperatures inside landfills can reach more than 50 • C (Hanson et al., 2010).Furthermore, heat is generated by microbial CH 4 oxidation and compost respiration processes.As the MOL consisted of compost, and the gas concentration curves suggest CH 4 oxidation and compost respiration, both of these processes are expected to have contributed to the increased temperatures measured in the MOL at Klintholm landfill.

Quantification of total methane emissions from the biocover cell
In March 2016, during a period of increasing barometric pressure, the average emission from Cell 0 was measured at 0.59 ± 0.25 kg CH 4 h − 1 (average of five plume traverses ± 1 standard deviation), using the tracer gas dispersion method.In January 2017, during decreasing barometric pressure, the average emission from Cell 0 was measured at 0.97 ± 0.47 kg CH 4 h − 1 (average of seven plume traverses ± 1 standard deviation), using the tracer gas dispersion method.In both cases, the CH 4 plume from Cell 0 was well-separated from plumes from other waste cells.The emissions were however likely to have been overestimated, due to potential CH 4 emission from compost storage piles located close to the biocover cell.The average CH 4 emission from Cell 0 prior to biocover was 5.4 ± 0.6 kg CH 4 h − 1 .After establishment of the biocover system, the emission was measured to be reduced to 1.8 ± 0.2 kg CH 4 h − 1 in 2010 and 0.7 ± 0.1 kg CH 4 h − 1 in 2011/2012 (Scheutz et al., 2014).The 2016/2017 results thus indicate that the overall CH 4 oxidation efficiency of the biocover system is still high 6½ year after installation.It should be noted that the change in CH 4 generation rate from the waste in Cell 0 probably has decreased by approximately 20-30% over the 6½ years as estimated by gas generation modelling (Rambøll, 2008).The decline in gas generation therefore cannot be the only cause of the low emission from the landfill cell.

Emission and oxidation processes
The low CH 4 /CO 2 -ratios observed in the GDL (0.01-0.9), in comparison to the CH 4 /CO 2 -ratio of landfill gas sampled from the interior of the landfill (2.5), indicated significant CH 4 oxidation in the GDL.Gas samples from the GDL showed elevated O 2 concentrations at most times and places providing the conditions for CH 4 oxidation.In addition, batch experiments performed on samples from the GDL showed a high potential for CH 4 oxidation (with rates in the same order as for the MOL at 22 • C) (cf.section 4.1.3).Based on the rates obtained in the incubation studies (at 22 • C), the combined CH 4 oxidation potential of the GDL layer was calculated for sections 2, 4 and 7, showing a total potential of 3.8 kg CH 4 h − 1 (cf.SM5).Assuming that these three biocover sections are representative of all sections, the CH 4 oxidation potential for the whole GDL layer is 14 kg CH 4 h − 1 .The CH 4 emission prior to biocover implementation was 5.4 kg CH 4 h − 1 ; thus, in comparison, the CH 4 oxidation potential of the GDL is significant, and the GDL very likely play a very important role in the overall CH 4 oxidation efficiency of the biocover system.
In Table 4, average CH 4 and CO 2 emissions (in g m − 2 d − 1 ) are calculated based on the flux chamber measurements in the three test plots.SM3 illustrates the calculated CH 4 and CO 2 emissions for all points in the test plots and campaigns.Table 4 shows that in March, with increasing barometric pressure, CH 4 emission levels for all three test plots are very low.This occurs because the increasing barometric pressure either reduce advective transportation of LFG to the MOL from below or of LFG from the gas supply trenches to the GDL, leading to low CH 4 loading, and as a result low or even no CH 4 emission.This will result in CO 2 surface emissions only originating from compost respiration of about 70-110 g CO 2 m − 2 d − 1 , based on the measurements in locations where no CH 4 is observed (see SM3 in SM).The observed range seems higher that seen in non-loaded locations in the biocover system back in 2010, where the observed end-valuesafter some maturationwere on average 32 g CO 2 m − 2 d − 1 (Scheutz et al., 2014).The higher CO 2 emissions in March 2016 could be due to the mobilisation and respiration of carbon from the compost, and the respiration of dead microbial biomass and residual plant material accumulated in the upper part from MOL. Incubation tests of the MOL also showed higher respiration rates in comparison to when the filter materials were put in place (Table 2).Increasing respiration activity after prolonged CH 4 oxidation has previously been observed (Scheutz et al., 2011b).
For April 16 and January 17, CH 4 emissions were much higher, due to the stable or decreasing barometric pressure experienced during the measurement campaigns (4.2 -84 g CH 4 m − 2 d − 1 ) (Table 4).For the test plot in section 2, the average CH 4 emission was lower than expectedeven lower than the emission in April 2016.The reason for this discrepancy has not been revealed.Also, higher CO 2 emissions (273 -283 g CO 2 m − 2 d − 1 ) were evident, and based on gas concentration profiles and CH 4 /CO 2 ratios, it is clear that additional CO 2 resulted from oxidation of the higher CH 4 load to the biofilters.Many of the measured gas profiles from April 16 and January 17 reflect a higher CH 4 load and greater CH 4 oxidation (confer SM4).
Overall, the investigations on the three test plots showed significant temporal as well as spatial variability in the surface fluxes.Especially changes in barometric pressure had a profound effect on the magnitude of surface emissions.However, in general CH 4 fluxes were modest at most measured locations.

Suggested conceptual model for biocover system performance
Integrating observations from the performed laboratory and field activities, especially the gas profiles and flux chamber measurements performed at three different campaigns covering different barometric pressure scenarios at 48 locations in three test plots, has led to a  In the previous section, it was illustrated that CH 4 oxidation is a prevailing process taking place in the GDL.Theoretical considerations (Kjeldsen, 1996) and several experimental lab batch experiments have shown that the process leads to a volume reduction, subsequently resulting in a pressure drop (e.g.Maurice and Lagerkvist, 2004).If some regions of the MOL consist of material with higher permeability (due to more coarse material or lower water content), air may be sucked through the MOL and into the GDL as a result of the pressure drop caused by the oxidation process in the GDL.The transport of LFG from gas collection trenches on the landfill slope may also be enhanced, leading to an increased routing of gas to the GDL, thus supporting CH 4 oxidation in the GDL.The intrusion of air into the MOL may be enhanced even further by increasing barometric pressure.If the intruding air is cold, the process may cool down the MOL and create gas compositions close to atmospheric conditions and low MOL temperatures (Gas Profile AI (Air Intrusion) in Fig. 4).The intruded air may later, when the barometric pressure decreases, be transported upward into the MOL in nearby regions: O 2 will be supplied to the MOL from both below and above (Gas Profile MO2 (Methane Oxidation (O 2 from above and below))).In regions with high permeability, decreasing barometric pressure may force the fast transport of raw LFG containing high concentrations of CH 4 and CO 2 .In-diffusing O 2 will only be present in the shallowest part (Gas Profile AE (Anaerobic Conditions due to high Emission)), resulting in CH 4 oxidation only in the shallow part of the MOL.If the upward transport of LFG is more modest, O 2 will be able to diffuse deeper into the MOL from above, and a more traditional landfill cover soil profile is observed (Gas Profile MO1 (Methane Oxidation (O 2 only from above))).At locations and times where the GDL does not contain CH 4 -either as a result of CH 4 oxidation in the GDL or the absence of LFG loading -, gas profiles with no CH 4 , elevated CO 2 concentrations and O 2 diffusing in from above are apparent.Temperatures are elevated due to heat generated by the respiration process taking place in the MOL (Gas Profile R (Respiration, only)).If the GDL below contains elevated CO 2 concentrations (maybe as a result of CH 4 oxidation in the GDL), the CO 2 profile gradient may be positive at the lowest point, to drive the upward transport of the CO 2 from the GDL.If the gradient is close to zero, only CO 2 generated by the respiration process in the MOL will emit from the surface.
The biocover system stands out six years after its establishment as a much more heterogeneous and dynamic system in comparison to the "traditional" biofilter, where the MOL is homogenously loaded from the GDL with pure LFG not containing any O 2 , and O 2 supply occurs through diffusion from the atmosphere (i.e. the profile type MO1 will be seen throughout the entire MOL).In such a case, CH 4 oxidation efficiency can be limited by O 2 availability, which may not be the case in the "developed" biocover system with clear O 2 intrusion into the GDL.It can be speculated that changes to the physical properties of the MOL, i.e. change over time in water content, grain size distribution etc., can be the reason for the observed heterogeneous and dynamic behaviour of the MOL with several different profile types.As discussed above, the changes over time may create a more heterogeneous biocover system, however, still efficient in terms of CH 4 oxidation.

Methane oxidation efficiencies based on local carbon mass balances
An overall evaluation of the performance of the three test plots showed a good correlation between the observed concentrations in gas profiles and gas emissions measured by flux chambers.In nearly all cases, CH 4 fluxes were very low or below detection when CH 4 concentrations in the gas profiles were very low or below detection, and the opposite.
The carbon mass balance approach (Christophersen et al., 2001) can be used to evaluate local CH 4 oxidation efficiencies.The carbon mass balance approach assumes stationary conditions.Our results show a degree of short-term variability in gas fluxes and pore gas concentration.This imply that the obtained results are somewhat uncertain and should mostly be used in relative terms.The approach has lately been extended also to take into account CO 2 generation as a result of MOL respiration (Kjeldsen and Scheutz, 2019).The approach relies on the local measurement of CH 4 and CO 2 surface fluxes, aligned with CH 4 and CO 2 gas concentrations in the lowest part of the MOL.The CO 2 flux originating from respiration is estimated.The approach provided by Kjeldsen and Scheutz (2019) (also explained in SM6) is used to calculate local CH 4 oxidation efficiencies, as presented in Table SM6.1.The approach also assumes that O 2 is present in excess so no competition for O 2 between CH 4 oxidation and respiration processes will occur.This is believed to be a valid assumption in most cases.The CO 2 flux originating from respiration has been estimated at 28.5 g CO 2 m − 2 d − 1 (0.65 mol C m − 2 d − 1 ), based on measured CO 2 flux at an identified R-type from the April campaign, where barometric pressure was stable.Table SM6.1 shows that only a few locations are home to active CH 4 oxidation during increasing pressure (Campaign March 16), with relatively low CH 4 oxidation rates (<13 g CH 4 m − 2 d − 1 ).This may be a result of low LFG loading from the landfill body, due to increasing barometric pressure in combination with higher air intrusion into the GDL, thereby supporting significant CH 4 oxidation in the GDL.In the stable pressure scenario (Campaign April 16) and decreasing pressure scenario (Campaign January 17), many more locations exhibit CH 4 oxidation rates with maximum rates above 50 g CH 4 m − 2 d − 1 .In both campaigns, no CH 4 loading is observed at several locations.In some locations, however, CH 4 oxidation rates cannot be calculated by the C-balance approach, because the calculated bottom CH 4 load is lower than surface CH 4 emissions.The reason for this is most likely the very dynamic behaviour of the biocover system, exemplified by changing gas composition in the MOL over a short time, due to barometric pressure changes.Similar behaviour has been reported in other biocover systems (Gebert and Grönhöft, 2006a;2006b).The fact that the gas profiles and the flux measurements are done at different times for a specific location (typically 30-90 min between the profile and flux measurements at specific locations) may to some extent violate the steady-state assumption of the C-balance approach, and render CH 4 oxidation rates more uncertain.

Conclusions and perspectives
Total methane emission measurements conducted by a gas tracer dispersion method indicated methane emissions at lower or similar levels compared to previous measurements performed just after the establishment of the biocover system in 2009.The gas production in the landfilled waste is expected to have decreased over the 6½ years period, but not to an extent, which alone can explain the low methane emission C. Scheutz et al. from the landfill cell as measured by the gas tracer dispersion method.Gas concentration profiles in some locations showed clear signs of methane oxidation in the methane oxidation layer, while surface screenings following a gridded approach and surface flux measurements revealed that areas with positive methane fluxes were scattered and dynamic.This suggests that landfill gas was not distributed evenly between or throughout the biofilter sections.Temperatures higher than ambient temperature were observed throughout the methane oxidation layer, with average temperatures ranging between 13 and 27 • C, even when measured in the traditionally coldest month of the year.This shows significant microbial activity and demonstrates the insulating effects of the filter material.Both laboratory and field observations showed that the gas distribution layer played a significant role in methane oxidation, mainly as a result of down-wash of fine material from the methane oxidation layer into the gas distribution layer.
In conclusion, the biocover system still worked in an acceptable manner after seven years.Moreover, the results point to no or little need for maintenance in the first 6-7 years of a biocover system's life, except to repair damage caused by burrowing animals.
The original tests in 2009/2010 were carried out using the biofilter material (compost made of a mix of garden waste and kitchen waste), which at the time was placed in the biocover windows; thus, it was quite homogeneous.Years later, it is less homogenous, due to the specialisation of bacterial communities, natural settlement, differences in the supply of water, methane and oxygen and, finally, exposure to different temperatures.The development of a more heterogeneous system is in general not wanted.However, using a living material such as compost, development of a more heterogeneous system is probably unavoidable.
Our study showed that it did not have a detrimental effect on the functionality of the biocover system.
The obtained result may be highly site-specific, thus, extrapolation to other sites should be done with caution.The observations from this study reveals that the properties and homogeneity of the methane oxidation layer and the gas distribution layer may change over time due to natural occurring processes.This development is spatial variability is expected in any compost based biocover system in a long-term perspective.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Fig. 2 .
Fig. 2. 4 surface fluxes from the established test plot in sections 2, 4 and 7 (numbering shown at the top), measured by static flux chambers in the three campaigns (shown to the left).The distances in cm show the length between the outer points in the measuring grid.Typical spacing between measuring points were 110 cm to 200 cm.

Fig. 3 .
Fig. 3. Selected gas concentration and temperature profiles representing different profile types (AE: Anaerobic conditions due to high emission; MO1: Methane oxidation (O 2 only from above); MO2: Methane oxidation (O 2 from above and below); R: Respiration, only; AI: Air intrusion).Depth is in cm below ground level.

Table 1
Overview of the performed field activities in the three campaigns.
a Only in the three sections with test plots.C.Scheutz et al.

Table 2
Data from laboratory tests and selected field measurements.The values presented here are averages; uncertainty represents one standard deviation.Numbers in parentheses represent the numbers of measurements or experiments on which the average and uncertainty are based.

Table 3
Ambient temperatures and temperatures in the MOL.Ambient temperatures were recorded at Hans Christian Andersen Airport situated approximately 40 km from Klintholm landfill.The data was retrieved from www.wunderground.com,b Temperatures at every measured depth for every measuring point in each section were included in the calculation of average temperatures. a

Table 4
Calculation of average CH 4 and CO 2 emissions (in g m − 2 d − 1 ) from the three test plots during the three campaigns.ofaconceptualmodelfor biocover system behaviour.The field observations shown in Figs. 2 and 3 depict a high degree of spatial and temporal variation; changes in barometric pressure especially can explain some of the observed variability.The conceptual model is presented graphically in Fig.4, which is explained in the following.The conceptual model is made under the assumption that gas flow is vertical in the MOL at all locations and times.Since the biofilters are constructed on top of the landfill and are completely flat and horizontal, gas flow along sloping layers introduced by capillary effects is thus not likely to occur.The figure shows the LFG inlet to the left, where LFG is led into the GDL.The MOL is shown as the layer above the GDL.Based on an analysis of the observed CH 4 and CO 2 fluxes and the gas and temperature profiles, five different types of biofilter behaviour is identified: AI (Air Intrusion), AE (Anaerobic conditions due to high Emission), MO1 (Methane Oxidation (O 2 supply only from above), MO2 (Methane Oxidation (O 2 supply from above and below) and R (Respiration (only) (cf.SM4, which identifies the type of each observed gas concentration profile, and registers the quantity of each type for each measuring campaign and test field in TableSM4.1).The different behaviour types are connected to different parts of a specific biofilter, which can change according to time and location, depending on changes in environmental factors (such as meteorological conditions, seasonal effects and others). development