This study was conducted at a sewage treatment plant, located in Victoria, Australia. One of two identical treatment lagoons with surface aerators, 25W Pond 1 was chosen for this study. Wastewater after being treated from the anaerobic Pot flows directly into 25W Pond 1, and travels through the pond in an east-west direction. Other research using hoods and mobile survey techniques have shown that the Pond 1 have the highest CH₄ emissions at the sewage treatment plant and therefore are the largest concern with respect to the CH₄ emissions. This is because it is the first pond after the covered anaerobic pot where anaerobic digestion of raw sewage takes place. Biogas is generated underneath the cover and is collected and sent to onsite power generation facilities for energy generation. Effluent from the anaerobic pot is saturated with dissolved methane, once agitated and/or under different atmospheric pressure (in Pond 1), releases CH₄ emissions to the environment. Wastewater in the downstream facultative ponds such as Pond 2 has a much lower chemical oxygen demand (COD) than in Pond 1, and Pond 2 typically emits less than 1/10 of the CH₄ emissions compared to Pond 1. The 25W Pond 1 has dimension of 250 × 1010 × 3 m (width × length × depth) (Fig. 1). There are also several lagoons located on the south of the Pond 1, while to the north of Pond 1 the terrain is flat, covered with short grass, and there are no tree lines or tall buildings nearby. Further north, there is a corn field, about a few hundred meters north of the pond. A sludge drying pan area is located ∼ 200 m to the west of the pond. This layout allows for CH₄ measurements when there are winds tending from the north, because to the north within a radius of a couple of hundred meters, there are no other CH₄ sources interfering with the measurements. The average minimum/maximum ambient temperature was approximately 18 and 26 °C for summer, 12 and 19 °C for winter, respectively. A total of 1.4 and 37.4 mm of precipitation was observed over the summer and winter measurement period, respectively.

Three open-path lasers (OPL) were deployed in each campaign measuring line-average CH₄ concentration (in ppm-m). Two of the three laser sensors used in this study were from Unisearch Associates Inc. Canada (LasIRView, OPL_C33, OPL_C34), the third laser was from Boreal Laser Inc. Canada (Gasfinder 2.0, OPL_C1013). The concentration sensor sends a collimated beam from a tunable infrared laser diode to a retro reflector, from which the beam is reflected back to the receiver optics and a detector. The outgoing beam is absorbed by CH₄ molecules over the measurement path (between the laser and retro reflector), giving a measure of CH₄ concentration. Line-averaged CH₄ concentration is obtained every few seconds. The precision of CH₄ concentration at a 100 m path length is: < 1 ppb for the Unisearch laser, 20 ppb for the Boreal laser.

To avoid surrounding ponds contributing to the emission of target source Pond 1, our experimental layout was designed for only northerly winds, and data collected with other wind directions was not counted for the flux calculation. More details are shown in below filtering criteria section. In each campaign, two lasers were set up at the south of the pond to measure the downwind concentrations and one laser at the north of the pond to measure the upwind concentrations so that the enhance concentrations from the Pond 1 can be determined. Each instrument was deliberately matched to the expected CH₄ concentration range at its location. The higher-precision Unisearch laser was used in the western low-emission zone and the Boreal laser in the eastern high-emission zone: OPL_C33 was located the western side of the pond (west laser) and OPL_C1013 was located the eastern side (east laser), close to the anaerobic Pot cover area. The third laser OPL_C34 (background laser) was located at the north of the pond measuring the upwind concentration. Noted the upwind laser remained at the same location during the two measurement periods as the land cover was similar and the background condition remained stable during the measurements (Fig. 1).

Each laser and retro reflector were mounted on a separate tripod at approximate 1.50 m above the ground, with the pathlength of 100–150 m between the laser and retro reflector. Each OPL was powered by a 12 V battery coupled with solar panels. Prior to the measurements, three lasers were conducted cross-calibration on site for at least 48 h to examine their stability and performance under same climate conditions. Gas emission measurements were begun from 8 February to 12 March 2024 for the summer campaign, and from 15 August to 5 October 2024 for the winter campaign. Measurements from both campaigns are included in this paper.

To examine the Pond 1 effluent chemical and physical property and understand how these factors are associated with the flux measurements, wastewater samples were collected in September using a drone along the middle of the pond at 5, 50, 100, 150, 200, 275, 350 and 500 m from the Pot covered area (sampling locations are shown in Fig. S1 in the Supplement). For each sample, the temperature, redox, pH, and dissolved oxygen were analysed immediately on site. Subsamples (850 mL each) were also taken at each location and analysed in the laboratory for other properties analysis.

The IDM technique is a classic micrometeorological method that calculates emissions from gas measurements taken in the free air. The micrometeorological methods are generally preferable to other approaches, as they are non-intrusive techniques that are suitable for long-term measurements. Consider a treatment pond that is emitting gas to the atmosphere at an unknown rate Q, which causes the average gas concentration (C) downwind of the pond to rise above the background value (Cb). In the IDM technique, the measurement (C-Cb) is used to determine Q with the aid of an atmospheric dispersion model (Windtrax). WindTrax is a software (http://www.thunderbeachscientific.com, last access: 12 December 2025) based on the backward Lagrangian stochastic dispersion model (bLs) for calculating gas emission rate from a source area (Flesch et al., 1995). WindTrax uses the bLs method, which is based on Monin-Obukhov Similarity Theory, and simulates the relationship between concentration (C-Cb) and emission rate Q, (C/Q)_sim. Because the Q vs. (C-Cb) relationship depends on wind conditions, one must also make wind measurements.

In this study, open-path CH₄ sensors were located to the south and north of the experimental pond. During northerly winds, this configuration allowed upwind (Cupwind) and downwind (Cdownwind) measurements from the experimental pond, but not the southern ponds. Following Flesch et al. (2004), the CH₄ flux QIDM is calculated using IDM method following the equation (Eq. 1): 1QIDM=(Cdownwind-Cupwind)/(C/Q)sim where QIDM is the CH₄ gas emission rate (µg m⁻² s⁻¹), and Cdownwind, Cupwind is the line-average gas concentrations (in ppm) measured by downwind and upwind laser sensor, respectively (Fig. 1). The value of (C/Q)_sim is the simulated ratio of line-average concentration and emission rate, calculated by WindTrax modelling based on the ambient temperature and pressure, wind statistics, and atmospheric turbulent parameters.

A three-dimensional (3-D) sonic anemometer (CSAT-3, Campbell Scientific, Logan, Utah, USA) coupled with a datalogger (CR23X, Campbell Scientific, Logan, Utah, USA) was located at the south of the Pond 1 at a height of 2.34 m above the ground (Fig. 1), to record wind statistics at a frequency of 10 Hz that are needed for IDM calculation. 15 min statistics includes the friction velocity (u∗, m s⁻¹), turbulent velocity (u, v, w, T), and its variance (u2, v2, w2, T2) and covariance (uT, uv, uw, vT, vw, wT) in three dimensions, as well as ambient temperature (°C), wind speed (m s⁻¹), and wind direction. Ambient pressure (mbar) and rainfall (mm) were obtained from Bureau of Meteorology.

The concentrations of CH₄ were averaged into 15 min intervals then merged with wind variables as inputs for the IDM flux calculation using SAS software (SAS 9.4, SAS Institute Inc. Cary, NC, USA). The atmospheric turbulent parameters including Obukhov stability length (L, m), surface roughness (z0, m), and turbulent velocity fluctuation (σu/u∗, σv/u∗, and σw/u∗) were also calculated for the IDM simulation.

Following Flesch et al. (2016), the data for CH₄ emission calculations using the IDM methods were not counted when:

atmospheric turbulent conditions were poor: u∗ < 0.1 m s⁻¹, |L| < 2 m, z0 > 0.05 m,

The influent in the eastern part of the pond is directly connected to the anaerobic pot, where dissolved methane is high. By contrast, the western part of the pond has lower dissolved methane. We expected this to create a gradient of emissions from the eastern to western part of the pond. Because of the size of the pond, a single laser was not sufficient to capture the full pond, and instead two lasers were used to measure emissions along the gradient. Formally, this is a type of stratified sampling that scales each laser's measurements (eastern pond area and western pond area) to the full-lagoon emissions. The classic way to do post-stratification when using stratified sampling is to weight the estimate by the area that each strata represents (i.e., take a weighted average). In the summer campaign, the pond was divided into two strata of similar sizes (∼ 50 %–50 %). The total average emission was thus (0.5 times flux1 + 0.5 times flux2) × total pond area. The weights were varied by ±5 % to test the sensitivity of the results to this assumption and to provide a range of emissions for the measurements. The total emission for the summer campaign thus ranged from a low estimate of (0.55 times flux1 + 0.45 times flux2) × total pond area to a high estimate of (0.45 times flux1 + 0.55 times flux2) × total pond area. In the winter campaign, the east laser was closer to the Pot end covered area. The pond was divided into two strata of varying sizes (∼ 70 % west–30 % east). The total average emission was thus (0.70 times flux1 + 0.30 times flux2) × total pond area. After propagating uncertainty in the weights (±5 %), the total emission for the winter campaign ranged from a low estimate of (0.75 times flux1 + 0.25 times flux2) × total pond area to a high estimate of (0.65 times flux1 + 0.35 times flux2) × total pond area.

There are four sources of relative uncertainty in our flux estimate: instrument precision (e1 = 2 %), inversion-dispersion model (e2 = 20 %) (Laubach and Kelliher, 2005), sampling uncertainty (i.e., the standard error, SE, calculated using the number of observations and standard deviation among the 15 min measurements, which contributes around e3 = 1.2 % to 4.0 %), and the pond area represented by each laser in our weighted average (e4 = 2.0 % error in total flux for the summer campaign, and e4 = 5.2 % for the winter campaign, coming from varying the weight by ±5 % around their centred value, see section 2.6, resulting in a range of total average emission). These four sources of uncertainty are added in quadrature (sqrt (e1∧2+e2∧2+e3∧2+e4∧2), where the ei's are the different error terms) to propagate the uncertainty and get our final estimates. Note that the total uncertainty is dominated by the inverse-dispersion model term (e2) as the terms are added in quadrature.

To convert from relative uncertainty to absolute uncertainty (1σ), the 1σ figure was calculated using the definition for the coefficient of variation (CV) to represent the relative uncertainty (CV = 1σ/mean). Solving for sigma gives: 1σ = CV × mean. For example, using a CV of 20.6 % and a mean flux of 386.5 µg m⁻² s⁻¹ for the averaged Pond 1 in this study, a 1σ value of 79.6 µg m⁻² s⁻¹ is obtained.

15 min CH₄ fluxes from Pond 1 varied spatially between the eastern and western pond areas (Fig. 2). Higher fluxes were observed at eastern pond area, ranged from ∼ 200 to over 1900 µg m⁻² s⁻¹, and lower fluxes ranged from ∼ 13 to over 570 µg m⁻² s⁻¹ were observed at western pond area. Furthermore, a clear 24 h diurnal pattern in the fluxes was observed at the eastern pond area, with lower emissions around midday and higher emissions at night-time or early morning. This diurnal pattern of maximum CH₄ emission at 08:00 am local time (LT) was also reported in Glaz et al. (2016). In contrast, no obvious diurnal pattern of the emissions was observed at the western measurement location.

Daily averaged CH₄ fluxes varied among the seasons (Fig. 2). In the summer season, the daily average of CH₄ flux was 252.0 (±9.4, n = 139) (±SE for sampling uncertainty, n, observation numbers) and 582.1 (±23.3, n = 185) µg m⁻² s⁻¹ from the western and eastern pond area over the 5-week measurement period, respectively. In the winter season, the daily averaged CH₄ flux was 134.7 (±2.3, n = 797) and 874.2 (±10.6, n = 692) µg m⁻² s⁻¹ from western and eastern pond area over the 7-week measurement period, respectively. The emissions at eastern pond area were ∼ 2–6 times higher than that at western pond area, during both winter and summer measurements. The much higher eastern flux measurement in the winter campaign was mainly due to the laser measurement location being much closer to the anaerobic Pot covered area than during the summer measurement campaign. In contrast, the laser's locations at the western pond area remained the same during both summer and winter campaigns and the pond measurements were higher in summer than that in winter by 46 %.

The measured winter flux at western pond area was much higher than the higher range of the reports, e.g., the anerobic ponds in Australia (7 ± 1 g m⁻² d⁻¹) (Hernandez-Paniagua et al., 2014), duckweed treatment ponds of 1276 ± 299 mg CH₄ m⁻² d⁻¹ in the US (Sims et al., 2013), organic matter enriched sludge treatment wetlands of 1900 mg CH₄ m⁻² d⁻¹ in Norway (Søvik and Kløve, 2007), and 5400 mg CH₄ m⁻² d⁻¹ in Spain (Uggetti et al., 2012). Different measurement techniques, wastewater composition, climate conditions and operation management could introduce variability of methane fluxes (Reinelt et al., 2017).