Seawater Dissolved Methane, Headspace GC and Air–Sea Flux

The first distinction: dissolved methane is not a rising bubble

Methane measurements in seawater often appear in a table as nanomoles per litre, but their meaning depends on the phase being sampled. Dissolved CH₄ is distributed among water molecules; free gas occupies bubbles or pockets that behave differently. A sample collected through a bubble plume may contain both phases unless the sampling design separates them. A headspace gas chromatograph may be used to reconstruct dissolved methane, yet a very large peak does not by itself prove that the original water held that entire quantity in solution. The analyst must identify exactly what was captured.

Methane is a greenhouse gas, a reduced carbon compound and an important product or substrate in microbial carbon cycling. It may originate from microbial or thermogenic processes and can be consumed by methane-oxidising organisms before reaching the surface. Transport, mixing, bubble escape and seawater solubility determine what remains in the water column. A measured concentration therefore cannot independently identify the original source, its rate of generation or the eventual atmospheric emission. Those conclusions require additional evidence and clearly defined boundaries.

The analytical challenge is especially sharp because methane is only sparingly soluble in water at ordinary surface conditions. Small uncontrolled gas exchanges can distort low-concentration measurements, while high-concentration samples risk detector saturation or carryover. This guide follows a defensible chain from sampling through headspace equilibration, calibration, concentration calculation, source interpretation and flux estimation. Its numerical examples are hypothetical teaching cases, not observations from any named cruise.

Separate four methane quantities before doing arithmetic

The first quantity is water-phase concentration, commonly nmol CH₄ per litre or kilogram. The second is gas-phase mixing ratio, often parts per million by volume or micromoles per mole, after a known gas sample reaches the instrument. The third is saturation relative to atmospheric equilibrium, a dimensionless ratio usually expressed as a percentage. The fourth is flux, an amount crossing a unit area per unit time. A fifth possible output is an oxidation or production rate, which must carry a time denominator and a measured or modelled process definition.

These quantities are connected but not interchangeable. A methane concentration of 15 nmol/L in a one-litre sample describes fifteen nanomoles in that litre, assuming uniform dissolved concentration. It does not establish that fifteen nanomoles reach the atmosphere each day. An air–sea flux requires a surface gradient and gas transfer model. Likewise, a rate of microbial methane oxidation cannot be obtained by simply multiplying concentration by a guessed constant. The rate must be justified by experimental data or an explicitly stated model.

The term ‘methane saturation’ also needs a reference state. A value of 200 percent means the water concentration is twice the declared equilibrium concentration, not that two hundred percent of a finite capacity has been filled. Ocean water is not a container with a fixed methane capacity comparable to the size of a fuel tank. The equilibrium reference changes with physical conditions and atmospheric composition. This is why good explanations retain the actual ratio rather than replacing it with the everyday idea of ‘full’.

The sampling journey introduces its own risks

Water collected from a rosette bottle, surface intake or specialised sampler may already have been affected by pressure change or turbulence. Methane-rich water can exsolve gas if pressure falls or conditions change. Sample bottles should follow a validated filling and sealing procedure suited to the concentration range and intended measurement. Record sample identity, depth, time, temperature, salinity, vessel type, filling method and any headspace created intentionally or accidentally. These fields are part of the measurement, not administrative decoration.

An accidental air pocket can remove methane from solution as the system equilibrates, changing the original dissolved concentration before controlled analysis. Conversely, a bottle that intercepts a methane bubble can overrepresent the dissolved pool if the bubble later dissolves or is included in an extraction calculation. A clear sample label should therefore distinguish dissolved water analysis from free-gas, sediment-headspace and bubble-capture methods. They are not alternative names for the same measurement.

Storage and transport matter as well. Bottle leaks, changing temperature, stopper diffusion and continuing microbial activity may affect methane over time. A fast assay is useful only if the entire preceding chain is controlled. A method validated for immediate processing should not be assumed valid for weeks of untested storage. Where preservation is used, its compatibility with the chosen chemical assay and intended microbiological interpretation needs validation. Recording the elapsed time to analysis makes later uncertainty assessment possible.

How headspace equilibration recovers a dissolved inventory

One common method establishes a gas phase of known volume and composition in a sealed sample vessel. After controlled equilibration, the methane in that gas phase is measured by gas chromatography. Some methane remains dissolved in the water, so the analyst uses an appropriate partition relation to reconstruct the original amount. An approach that measures only the amount in headspace without accounting for residual dissolved methane can undercount the starting inventory. A different mistake is to treat headspace methane mixing ratio directly as nmol/L of original water.

A conceptual mass balance says the amount initially dissolved in water plus any methane in added gas equals methane finally in the gas plus methane finally remaining in water, provided no leaks or chemical transformations intervened. The gas and water terms each require their own volume and concentration. Real procedures also account for gas pressure, temperature, background methane, sample dilution and exact solubility conventions. The equation is a framework for understanding the analysis rather than a replacement for a validated laboratory standard operating procedure.

Henry’s law coefficients can be expressed in multiple reciprocal conventions, such as dissolved concentration divided by partial pressure or the reverse. Even a correct coefficient gives an incorrect answer when inserted into an equation designed for its reciprocal. A reader should always write the coefficient units and verify that the resulting concentration carries mol per litre of water, not mol per litre of gas. The factor connecting the phases depends on temperature and salinity, and should not be copied from a freshwater example into seawater without adjustment.

Worked example: methane moves into the analytical headspace

Use a hypothetical vessel containing sixty millilitres of seawater and ten millilitres of initially methane-free gas. After equilibration, suppose the gas-phase methane concentration is 300 nmol per litre of gas. To make the arithmetic transparent, define the liquid-to-gas concentration ratio under the example’s conditions as 0.04. The residual liquid concentration is 0.04 times 300, or 12 nmol per litre of water. This deliberately supplied ratio is not a universal value for methane solubility.

The amount remaining in water is 0.060 L times 12 nmol/L, or 0.72 nmol. The amount in gas is 0.010 L times 300 nmol/L, or 3.00 nmol. Their sum is 3.72 nmol. Divide by the initial sixty millilitres of water to reconstruct 62 nmol/L, assuming the initial gas had no methane and the closed system preserved all analyte. The outcome demonstrates why the instrument’s gas concentration and the original seawater concentration are different quantities.

Now suppose the analytical gas initially contained 0.12 nmol methane. Then the starting water inventory was 3.72 minus 0.12, or 3.60 nmol, corresponding to 60 nmol/L. This two-nanomole-per-litre difference may be meaningful for some comparisons. It also shows why headspace blanks and incoming gas certification are necessary. The example assumes idealised conditions; an actual method must use its validated volume and solubility corrections, temperature control and recovery checks.

Gas chromatography detects methane after physical separation

Methane can be separated on an appropriate gas chromatography column and measured using a flame ionisation detector in many standard configurations. The instrument sees an injected gaseous sample, not the ocean water itself. Calibration links detector response to gas concentration or mole fraction. The response may depend on injection loop volume, pressure, carrier gas conditions and instrument settings. A clear chromatographic peak should be treated as part of the evidence, not as independent confirmation that sampling and extraction were unbiased.

Blanks and reference gases perform different jobs. An instrument blank tests residual methane or contamination in the injection pathway. A headspace-process blank can reveal contamination from vessels or added gases. A continuing check standard tests detector response after the initial calibration. A matrix spike tests recovery across the method under defined conditions. A single excellent calibration line does not exclude later carryover or loss of volatile methane from sample bottles.

High methane samples may challenge the working range and contaminate the next analysis if equipment retains gas. Running a clean blank after a strong sample helps reveal carryover. At the other extreme, low methane samples approach the validated detection or quantification threshold. The correct response to an uncertain peak is a qualified report or reanalysis, not adding decimals until the result appears reliable. Good laboratory records preserve peak area, standards, calculations and quality flags separately.

Dissolved methane is not a direct measurement of its source

Methane may be generated by microorganisms under suitable anoxic conditions, by geological processes or through other pathways. It can enter water from sediments, seepage, groundwater, hydrothermal systems or biological activity within the water column. A methane-rich sample is compatible with several of these possibilities. It is a mistake to treat concentration as a fingerprint that uniquely identifies the source. Methane derived from different sources may mix, and subsequent oxidation can alter the measured pool before sampling.

Stable carbon and hydrogen isotopic compositions can add evidence about methane origin and transformation, but they also need careful interpretation. Methane oxidation may fractionate isotopes; mixing may move an observed value between endmembers. A single isotope ratio is not a complete source diagnosis. The strongest investigations combine gas abundance, accompanying hydrocarbons where applicable, isotopic evidence, hydrography, seep mapping and process measurements. Each observation narrows the range of plausible stories without making all alternatives disappear.

One common narrative error is to equate a near-bottom methane maximum with atmospheric emission. If methane is oxidised or remains isolated below a stratified layer, a large benthic source might not create an equally large surface flux. Conversely, a small concentration excess across a vast ocean area could contribute materially when integrated carefully. The source, standing inventory and boundary export must therefore be calculated or measured separately.

Oxidation turns a one-way story into a competition of processes

Methane-consuming microorganisms can reduce the dissolved pool through oxidation. Oxygenated water may support aerobic methane oxidation, while certain sediment environments permit anaerobic pathways using other electron acceptors. Their relative importance depends on environmental chemistry and microbial ecology. A decrease in methane over depth, distance or incubation time might reflect oxidation, but mixing, degassing and sampling losses can produce similar numerical trends. The observation needs appropriate controls before it becomes a biological rate.

Consider a hypothetical closed incubation whose methane concentration falls from 30 to 24 nM over three hours. The simple average net decline is two nM per hour. That is not automatically an oxidation rate, because headspace exchange, leaks and microbial production can also influence net concentration. A biological interpretation needs validated treatment controls, time series and consideration of whether the bottle conditions represent the ocean. A steady decline in a sealed bottle is a result of that incubation, not evidence that the open ocean everywhere behaves at the same pace.

In a stratified water column, oxygen, temperature, salinity and density profiles help determine where methane might be consumed or transported. A sharp methane decline near the oxycline is suggestive of a transformation zone, but cannot quantify its turnover without further work. The reader should distinguish a vertical gradient, which includes transport, from a temporal rate, which requires time information. This elementary distinction prevents many attractive but unsupported process diagrams.

From concentration excess to an illustrative surface flux

To estimate a net diffusive air–sea exchange, a simplified expression multiplies a gas transfer velocity by the difference between measured surface-water methane concentration and the concentration expected in equilibrium with the atmosphere. The transfer velocity must be appropriate for methane and the physical conditions. The equilibrium reference must use the correct atmospheric composition, temperature, salinity and pressure conventions. A sample drawn from far below the exchange boundary is not a surface flux observation by itself.

Imagine a surface concentration of 15 nM, an equilibrium concentration of 3 nM and an illustrative transfer velocity of 1.5 metres per day. The concentration excess is twelve nM, which is twelve micromoles per cubic metre. Multiply by 1.5 m/day to obtain eighteen micromoles per square metre per day, positive from sea to air under the declared sign convention. The calculation is dimensionally correct under the simple model, but the chosen k is hypothetical and must not be presented as a site-specific prediction.

The boundary flux is not necessarily the total methane release. Gas bubbles may bypass the dissolved pathway and contribute ebullition, which requires different observation or modelling methods. Even when a dissolved gradient is accurately measured, excluding bubble transport can omit an important component at some locations. Conversely, not every bubble emitted at the seabed survives intact to the surface; dissolution and microbial transformations intervene. A complete budget identifies which pathways were included and which remain outside the estimate.

The special problem of bubble-rich samples

An upward bubble plume can be visually dramatic, but a sample bottle drawn through it poses a difficult representativeness question. If bubbles enter and subsequently exchange methane with the surrounding water in the bottle, the measured concentration may no longer represent the dissolved concentration at the sampling location. A laboratory can accurately measure the material it receives while still answering a different question from the intended one. Sampling design must distinguish water-only collection, bubble capture and two-phase flow.

Pressure change further complicates the issue. A water sample taken at depth may experience depressurisation on recovery, allowing dissolved gases to exsolve if conditions permit. If bubbles form and are lost before analysis, reconstructed concentrations can be biased low. A method designed for ordinary surface seawater may not be suitable for strongly supersaturated deep seep samples. Choosing a pressurised sampler or specialised gas-tight method can be essential for questions where gas phase stability is central.

Compare two imaginary records: one describes methane in nmol per litre of filtered water, another describes millilitres of gas released per litre of recovered sediment slurry. These cannot be equated by a simple unit conversion because the physical fractions and collection histories differ. A good comparative review preserves both and explains their operational definitions. A source word such as ‘methane concentration’ should not conceal that the two studies did not measure the same reservoir.

Reference equilibrium, supersaturation and cautious vocabulary

Methane solubility changes with temperature and salinity. Atmospheric mole fraction is a gas-phase composition, not the water reference concentration until the appropriate solubility relation has been applied. The ratio Cwater/Cequilibrium, when defined consistently, gives saturation. If surface seawater contains eight nM against an equilibrium estimate of two nM, the ratio is four and saturation is four hundred percent. That is an excess concentration of six nM, not a claim that four times the measured methane will necessarily escape.

At the opposite extreme, if water concentration is below equilibrium, a simplified net diffusive flux can be directed from atmosphere into water. The sign of a plotted flux must be read alongside the convention used by its author. Some studies define atmosphere-to-sea uptake as positive, while others use upward emissions as positive. Translating a negative sign into the word ’emission’ without checking the definition can reverse the physical interpretation.

Even strong supersaturation should not be equated with source attribution. A surface excess can arise through vertical mixing, lateral transport or biological activity. It may also vary quickly after weather changes. When estimating emission over an area, researchers need adequate coverage across space and time, not only one extreme station value. The correct statement is usually that measured surface water exceeded a specified equilibrium reference under the sampled conditions, followed by a modelled exchange estimate with declared uncertainty.

Calibration, blanks and low-level methane

At low dissolved concentrations, the background of the gas used for extraction matters. A nominally clean gas may contain trace methane, and laboratory air can influence samples through leaks or deliberate equilibration. Blanks must be defined so that readers know whether they test the gas supply, vessel preparation, chromatograph pathway or full field procedure. A process blank is useful because it experiences more of the analytical chain than a detector-only blank. Neither automatically replaces field quality controls.

For a hypothetical sample with a water-equivalent uncorrected value of 5.2 nM and a validated process blank of 0.4 nM, subtraction yields 4.8 nM. But if the blank drifts between 0.1 and 1.0 nM during the batch, merely selecting the smallest value is not honest. The variation could materially influence low-level results. A proper quality review examines the time sequence of blanks, standards, sample order, and possible carryover from methane-rich samples.

Detection and quantification limits depend on the method, volume and instrumentation. A value below a validated limit does not prove that methane is absent; it sets a constraint on what can be reported. If a method reports less than one nM, converting that censoring statement into exactly zero can bias regional statistics and flux estimates. Keep the less-than sign, unit, method and threshold visible. The same care applies to negative blank-corrected estimates, which should be handled through the declared reporting procedure rather than interpreted as physically negative methane.

Why a laboratory result should carry its uncertainty history

Consider three corrected dissolved methane results from replicate aliquots: 9.6, 10.0 and 10.4 nM. Their arithmetic mean is 10.0 nM and their sample standard deviation is 0.4 nM. That describes the spread of those measurements. It does not include any shared loss before aliquoting, the correctness of the headspace partition relation or the representativeness of the original water bottle. Reporting only ‘10.0 ± 0.4 nM’ without identifying what the error represents can create unjustified confidence.

A method comparison may find that a new headspace procedure agrees with a reference method over a defined concentration interval but not outside it. Such validation should not be extrapolated to highly methane-rich seep samples without testing. Similarly, a calibration line with a strong fit does not prove zero bias if its standards were all prepared from a mislabelled stock gas. Accuracy requires independent checks, while precision describes repeatability under stated conditions.

Some uncertainty is shared across samples. If every result uses one solubility relation with an incorrect convention, averaging many samples will not remove the error. Other components, such as small independent detector noise, may shrink in an average. A careful interpretation identifies calibration, field handling, extraction, conversion and spatial sampling uncertainties rather than combining them into one unexplained percentage.

A methane profile is a map of water, not a clock

A set of bottles sampled at increasing depth can show methane concentrations rising toward the seabed. The pattern may be consistent with a benthic methane supply, but it is not a direct measurement of the emission rate from sediment. The shape also depends on mixing, stratification, transport and biological oxidation. Water from a different location may enter the section and carry a pre-existing methane signal. A useful interpretation joins the methane profile to temperature, salinity, oxygen, density and relevant physical measurements.

An apparent sharp gradient can be exaggerated by interpolation when only a few depths were sampled. A smooth colour contour drawn through widely spaced bottles does not mean methane was measured continuously throughout that band. Show sample positions or state the interpolation assumptions. Check whether all stations were sampled on the same day, because a survey that moves through a changing ocean can blend spatial and temporal effects.

Depth and pressure are also different coordinates. Hydrographic measurements often use pressure in decibars, even when a general description says ‘depth’. Preserve the original quantity in analytical files and specify any approximation used for interpretation. A measured methane difference across two pressure surfaces cannot be turned into a vertical molecular flux without an additional transport model or diffusion relation. A concentration gradient is an input to such calculations, not the flux itself.

Calculating a diffusive gradient is not the same as calculating total release

A vertical diffusive flux through water or sediment depends on the concentration gradient and an appropriate diffusivity, often with porosity or tortuosity adjustments in sediments. The sign convention and coordinate direction must be stated. The expression differs from the air–sea transfer formula because it describes transport through a medium rather than exchange across the ocean surface. An analyst who substitutes an atmospheric transfer velocity for a sediment diffusion coefficient has mixed two physical models.

Suppose a purely hypothetical concentration changes by twenty micromoles per cubic metre over a vertical distance of two metres. The gradient magnitude is ten micromoles per cubic metre per metre, or ten micromoles per cubic metre-squared. To obtain a diffusive flux, multiply by a diffusivity in square metres per second and apply the declared direction. No numeric flux follows until the coefficient is supplied. This problem illustrates why an attractive slope alone cannot tell us how many methane molecules cross a boundary per day.

Even a sound diffusion estimate may not include advection or bubbles. Methane-rich porewater can move through sediments by pressure-driven flow, and gas bubbles can move rapidly in ways not represented by molecular diffusion. A published ‘methane flux’ must identify which pathways are counted. A useful budget may need separate dissolved diffusive, advective and free-gas components, each with uncertainty and a physical boundary.

Transfer exercise: compare methane units without losing the phase

Exercise one gives a dissolved concentration of 20 nmol/L and a water sample volume of 25 mL. The inventory is twenty times 0.025, or 0.50 nmol. Exercise two gives 10.25 nmol/L and density of 1.025 kg/L; the corresponding mass-based concentration is 10 nmol/kg. These operations are ordinary dimensional analysis, but they protect interpretation because 20 nmol/L is not the same as twenty nanomoles in an arbitrary sample bottle.

Exercise three gives a headspace concentration of 300 nmol per litre of gas in a ten-millilitre gas space. The amount in that space is 3 nmol. It is not possible to infer the original water concentration without the water volume, dissolved residual, initial gas methane and relevant partition relation. If a report supplies only the headspace signal, the missing variables are a limitation, not permission to assume they are zero.

Exercise four presents two surface waters, one at 15 nM with equilibrium at three nM and another at eight nM with equilibrium at two nM. Their saturation ratios are 500 percent and 400 percent respectively. Their excess concentrations are twelve and six nM. Under equal illustrative gas transfer velocities, the first has twice the simple diffusive flux magnitude. Under different winds or exchange parameters, the relative fluxes could change. A saturation percentage is not itself a flux comparison.

Transfer exercise: recognise what a time series proves

Imagine a bottle whose methane concentration falls from 30 to 24 nM during three hours. The net slope is minus two nM per hour. The statement ‘methane oxidation occurred at two nM per hour’ is stronger than the data warrant because the bottle may have leaked, methane may have partitioned into an unaccounted gas volume, and biological production may have occurred simultaneously. A valid oxidation-rate claim requires appropriate controls and a design that isolates the process being estimated.

Now imagine a second incubation in which a killed control shows almost the same decline. That result challenges the biological interpretation, but killing or preservation may also change the sample chemistry. A more convincing design compares validated controls, time-zero measurements, multiple timepoints and replicate bottles under carefully documented conditions. Rates should be tested for linearity instead of assumed constant across an arbitrary interval.

The lesson transfers beyond methane. A concentration change is an observed net outcome. A mechanism is an explanation requiring separate evidence. When writing for non-specialists, the sentence ‘methane decreased in the incubation’ is a valid observation, while ‘bacteria consumed all the missing methane’ needs demonstrated attribution. Precision in language makes the research more useful, not less interesting.

Building a reusable methane data record

A strong record keeps original chromatographic response, gas standards, sample and headspace volumes, temperatures, pressures, calculated gas concentration, phase partition relation and final water-equivalent methane concentration as separate variables. Also retain bottle, station, coordinates, UTC sampling time, depth or pressure, salinity, dissolved oxygen and relevant context. Without these fields, future researchers may be unable to reproduce the result or determine whether a later recalculation changed the observation or merely the conversion.

Quality indicators should identify contamination, unusual bubbles, storage anomalies, failed calibration, concentrations above the validated range and values below quantification. An unexplained numeric quality code may be useful to software but opaque to readers. Publish the coding dictionary. A dataset that preserves raw values and processing history makes genuine corrections possible without silently replacing the original evidence.

A cross-language data dictionary should distinguish dissolved methane, free gas, headspace methane, methane mole fraction, supersaturation, oxidation rate and air–sea flux. Keep CH₄ notation, inequality signs and unit denominators intact. Translating ‘methane-rich’ as ‘high methane emissions’ changes the claim from concentration to boundary transfer. Likewise, an ‘elevated near-bottom concentration’ does not identify a seafloor emission rate unless additional measurements justify that inference.

A whole-cruise comparison needs more than two attractive averages

Suppose one cruise reports a mean methane concentration of five nM and a later cruise reports seven. The observed difference between the reported means is two nM, or forty percent relative to five. But the environmental interpretation depends on how the stations were chosen, seasons, depths, laboratory methods and treatment of nondetects. If one expedition sampled more coastal seeps or used a different extraction protocol, the means need not describe a temporal rise in a comparable population.

An equally important issue is heterogeneity. A few very high samples can dominate an arithmetic mean, while median values and maps reveal a different picture. Presenting sample count, range, distribution and sampling locations is more informative than one unqualified ‘ocean average’. If a broad area is represented by one station while a small inlet has twenty, an unweighted sample mean may overrepresent the inlet in an area-wide interpretation.

To repair an overconfident trend, define the population being compared, align sampling strata where possible, review analytical intercomparability and report uncertainty from both measurements and spatial coverage. If the evidence cannot establish a trend, state the two observed summaries without forcing a causal conclusion. That is a better result than either pretending the numbers are identical or claiming all differences reflect environmental change.

Frequently asked questions about dissolved seawater methane

Does a methane-rich bottom water sample mean methane is reaching the atmosphere? Not necessarily. It may be oxidised, transported laterally or retained below a stratified layer. Does a high headspace peak prove the water originally contained a high dissolved concentration? Only when the original phase inventory and extraction conditions are known. Can bubbles simply be counted as dissolved methane? No. Bubble transport and dissolved concentration are distinct and may require different collection designs.

Does supersaturation measure the percentage of methane that will escape? No. It is a ratio of measured concentration to atmospheric equilibrium concentration. Does a lower concentration after incubation prove microbial oxidation? No. Process attribution needs controls. Can a single δ13C or δ2H value uniquely identify a source? No; source mixing and isotopic fractionation can complicate the pattern. Is an accurate calibration sufficient? Not if sampling and extraction were biased.

Why report methane in nmol/kg as well as nmol/L? Because mass and volume denominators reflect different data conventions and seawater density. Why are headspace temperature and gas composition required? They control partitioning and background corrections. Why distinguish a diffusive flux from ebullition? Because gas follows different transport pathways, and a budget omitting one may underestimate total release in some settings. Clear answers depend on reading the method, not just the headline number.

A defensible final interpretation, with boundaries intact

A high-quality methane report identifies what phase was sampled, documents collection and extraction, uses calibrated analytical measurements and presents a clear calculation from detector response to dissolved concentration. It defines any equilibrium reference and exchange model separately. The conclusion can then describe where methane was enriched, how it related to water-mass structure and what independent evidence suggests about processes. It should not let an illustrative flux become a claim of observed atmospheric emission.

Readers should look for explicit uncertainty at each stage. A precise chromatogram can coexist with uncertain field preservation; an accurate dissolved measurement can coexist with a poorly constrained gas transfer coefficient. Sensitivity tests help show whether a claim survives plausible alternatives. If an apparent signal disappears when reasonable corrections are applied, that finding is valuable and should not be hidden by choosing the most dramatic calculation.

The complete skill is to preserve the chain of evidence: sample, phase, detector, calibration, amount, concentration, possible mechanism and boundary transfer. Once those steps remain distinguishable, even technically demanding ocean methane studies become readable without being oversimplified. The strongest scientific translation makes it clear what is known, how it was inferred and what still requires another measurement.

Five advanced audits for tricky marine methane datasets

Advanced audit one tests the effect of a shared blank. A sample reports a headspace-derived 6.2 nM and the laboratory’s process blank is 0.8 nM, making its valid corrected concentration 5.4 nM. A second sample is 5.8 nM before the same correction, giving 5.0 nM. The difference between corrected samples is 0.4 nM, exactly the difference between the original values when the same blank applies to both. This does not mean the blank is unimportant: the absolute concentrations and their uncertainty still depend on it, and a varying blank would change the situation. The exercise illustrates why common and sample-specific errors must be treated separately.

Advanced audit two asks what happens when the analyst changes units. A result of 12 nmol methane per litre corresponds to 12 µmol per cubic metre. If the gas transfer coefficient is 1.5 metres per day and the equilibrium reference is three nM, the gradient is nine µmol per cubic metre and the simple upward flux is 13.5 µmol per square metre per day. Writing 13.5 nmol per square metre per day would be a thousand-fold unit error. The cancellation of metres in the calculation makes the error visible before it reaches a figure caption.

Advanced audit three concerns a profile collected across a hydrographic front. Two stations have equal methane concentrations, but their temperature and salinity differ. Atmospheric equilibrium methane need not be equal in the two water masses. Therefore their saturation ratios may differ even though the reported concentrations match. An analyst should not assign the same reference concentration to both merely because it simplifies a regional map. The environmental conditions and solubility formulation determine the appropriate comparison, and uncertainty in those inputs should accompany the result.

Advanced audit four examines isotopic wording. A sample has a measured carbon isotope composition and a high methane concentration. A draft states that this proves methane originated from one named microbial pathway. The isotope value may constrain source hypotheses but can also reflect mixing, oxidation and fractionation. A stronger report combines isotope data with gas ratios, transport evidence and appropriate process studies. The repair is not to hide the isotope observation; it is to retain the measurement while making the causal inference conditional on the rest of the evidence.

Advanced audit five asks whether a laboratory check can replace field representativeness. Ten repeat analyses of one sealed bottle agree closely, but the bottle was collected beside a visible gas plume without a two-phase sampling plan. Reproducibility of the bottle’s contents does not prove that those contents represent dissolved methane in the surrounding water. The appropriate response is to document the plume context, separate dissolved and free-gas sampling where possible, and avoid using the bottle as an unqualified background value. Replicate precision and environmental representativeness are different virtues.