The question behind a seawater pH number
Seawater pH is a statement about hydrogen-ion activity expressed using a specified chemical convention, not a direct measure of how many molecules of carbon dioxide are in a bottle. A numerical value such as 8.05 becomes useful only when the reader knows the pH scale, analytical method, temperature, salinity, sample identity and quality control. In marine carbon studies, an apparent difference of a few thousandths can be important, so seemingly modest omissions in metadata deserve attention before any trend is interpreted.
Spectrophotometric seawater pH uses the colour change of an added indicator to infer the acid–base condition of a water sample. Light at two wavelengths is compared; the resulting absorbance ratio is interpreted through a calibrated chemical relationship appropriate to the indicator, temperature and salinity. It is not equivalent to casually dipping a generic pH strip into seawater, and its sophisticated optics do not protect the result from poor sample handling or an incorrect reference scale.
This guide is a translation manual in the practical sense: it translates laboratory ratios, method notes and quality flags into defensible scientific statements. Every numerical illustration below is a constructed teaching problem rather than an observation from a named voyage. The aim is to move from a raw signal to a properly qualified result, through uncertainty, controls and independent checks, and finally to the kinds of claims that the measurement can and cannot support.
First principles: acid-base chemistry in salty water
The familiar expression pH = −log10 of hydrogen-ion concentration is only the beginning. Real seawater contains ions that interact, and different operational pH scales differ in which proton-bearing species are included. A logarithm also means that a difference of 0.10 pH unit corresponds to a hydrogen-ion ratio of about 1.26, not an additive 0.10 concentration. Before calculating, the reader should ask what quantity the published pH actually represents and which scale the calibration uses.
The total hydrogen-ion scale, often labelled pH_T, includes the contribution associated with bisulfate; the seawater scale also incorporates fluoride-related chemistry. A free scale treats only free hydrogen ion. These are connected through thermodynamic constants and the composition of seawater, rather than through one universally valid numerical offset. Historical archives also include NBS-like operational scales used with glass electrodes. Values on unlike scales must not be placed on the same plot as though their labels were decorative.
Carbon dioxide entering water can alter the balance among dissolved CO2, bicarbonate and carbonate. Total alkalinity describes the water’s capacity to accept protons under a defined titration convention, while dissolved inorganic carbon counts relevant inorganic carbon species. pH is affected by the relative balance and environmental conditions; it is neither identical to alkalinity nor identical to dissolved inorganic carbon. Pairing complementary carbonate parameters permits further calculation, but the outputs inherit each input’s uncertainty and equilibrium assumptions.
The indicator is a chemical participant, not an invisible label
Meta-cresol purple is frequently used because it exists in protonated and deprotonated forms with different visible absorption. At the relevant seawater pH, the colour ratio reflects how much indicator occupies each form. The measurement rests on an equilibrium relationship and wavelength-dependent molar absorptivities, not merely on whether the sample looks purple or yellow to a person. A report must identify the indicator and the relevant calibration formulation because changing the formulation can change the calculated pH.
Adding indicator is itself a perturbation to the sample. The dye solution has acid-base properties, and even a small injected volume can shift the original balance. This matters especially in lower-buffer-capacity estuarine or meltwater samples where an adjustment established for open-ocean seawater may not transfer. Readers should look for dye-concentration trials, batch-specific correction, purification information and whether the computation explicitly accounts for the disturbance caused by the indicator.
The words purified dye carry analytical significance. Dye impurities can alter optical ratios and lead to systematic bias across samples. A high-precision instrument can repeat the same biased result indefinitely, so reproducibility is not proof of accuracy. First-party NOAA-hosted research on purified meta-cresol purple documents that dye preparation and quality assurance matter. When two surveys disagree, it is sensible to compare dye batches and calculation methods before concluding that the ocean changed.
What the spectrophotometer measures
A spectrophotometer sends light through a cell and measures transmitted intensity relative to a reference. Absorbance A is commonly represented as −log10(transmitted intensity divided by incident intensity), so absorbance depends on optical path length, absorber concentration and instrument background. In indicator work, suitable absorbances near wavelengths associated with the acid and base forms are corrected for baseline and combined as a ratio. The ratio removes some dependence on absolute indicator amount but does not magically eliminate every instrumental bias.
Many published workflows use an absorbance near 434 nanometres and another near 578 nanometres for meta-cresol purple, with a baseline correction around a nonabsorbing wavelength. These are method-specific, not a universal wavelength pair for all possible dyes. The reader should find the recorded wavelengths and distinguish raw absorbances from corrected values. If an instrument changes wavelength calibration or stray light varies, the two channels may shift differently and distort their ratio.
Imagine two samples yielding corrected absorbances A578 = 0.420 and A434 = 0.300. Their dimensionless ratio is 1.400. That arithmetic is straightforward: 0.420/0.300 = 1.400. It does not by itself equal pH 1.400, nor can one insert it into any random published equation. The correct next step is to use a validated indicator relationship with appropriate temperature, salinity and scale; the worked ratio can be checked without fabricating the full chemical calibration.
A clean sample starts before an optical reading
Collecting seawater into a clean, appropriately flushed cell is part of the measurement, not merely preparation. Air bubbles scatter light, incomplete rinsing can contaminate the next sample, and gas exchange can alter the dissolved CO2 condition. Sampling from a Niskin bottle, through tubing, adds its own possible contact surfaces and delay. The collection record should identify bottle, station, pressure or depth, sampling sequence and any departures from standard handling.
Temperature change creates two distinct questions. First, the sample’s chemical equilibrium depends on its actual temperature at measurement. Second, the scientific question may concern pH at in-situ temperature rather than a controlled laboratory temperature. A dataset stating pH_T(25 °C) means a value defined at that reference temperature, not necessarily the pH in the ocean at the sampling point. Converting between temperatures requires a consistent carbonate-system model and appropriate input parameters.
Seal and temperature-equilibrate samples as the method specifies, avoiding bubbles and unnecessary warming. Record the thermometer and bath calibration rather than relying on the instrument’s nominal setpoint. Even a careful optical reading cannot correct an unidentified sample mix-up. Bottle labels, repeated field blanks where meaningful, duplicates and chain-of-custody information allow an anomalous value to be traced to a handling event rather than interpreted as a mysterious chemical feature.
From absorbance ratio to calculated pH
A useful way to understand the calculation is to begin with the indicator acid equilibrium: its protonated and deprotonated forms coexist at a ratio controlled by hydrogen-ion condition. Their absorbances at chosen wavelengths are mixtures of contributions from both forms. Solving the resulting equations produces an expression containing an indicator dissociation parameter and a logarithm of a function of the measured ratio. The constants embody laboratory calibration under specified conditions.
One NOAA data description uses an illustrative relationship pH = pK_ind + log10[(R−0.0069)/(2.222−0.133R)] for a specific workflow, where R is a baseline-corrected wavelength absorbance ratio. Those coefficients should be treated as belonging to that stated method, not a formula to paste indiscriminately into every modern dataset. A different indicator batch, temperature treatment, wavelength arrangement or calibration can require different coefficients and pK_ind. The guiding question is whether the equation matches the data provenance.
For R = 1.400 in that stated illustrative equation, its dimensionless logarithm argument is (1.400−0.0069)/(2.222−0.133×1.400), or about 0.684. The logarithmic contribution is about −0.165. The full pH still depends on pK_ind evaluated using its specified temperature and salinity conditions. This is an example of a scientifically honest stopping point: a reader can verify the optical mathematics without inventing a missing dissociation value.
A ratio calculation can fail while looking plausible
Consider a spreadsheet in which one row contains A578 = 0.480 and A434 = 0.320, so R should equal 1.500. A transposed formula A434/A578 gives 0.667. Both ratios look numerically plausible and neither causes an obvious spreadsheet error. The repair is to write the wavelength identifiers directly into the calculation heading and verify a row by hand. A quick comparison against a known reference can catch a whole column of inverted ratios.
A second subtle mistake is treating a baseline absorbance as if it had already been removed. Suppose A578 is 0.470, A434 is 0.320 and the common baseline is 0.020. The uncorrected ratio is 1.469, while (0.470−0.020)/(0.320−0.020) is 1.500. The reader cannot silently assume the difference is harmless because pH involves a logarithm of a nonlinear ratio transformation. Record precisely whether archival columns are raw or corrected.
Another failure is to substitute absorbance values in transmittance units. A transmission fraction of 0.50 is not an absorbance of 0.50; it corresponds to A = −log10(0.50), about 0.301. Treating the original transmission value as absorbance can distort the wavelength ratio. Keep distinct field labels for intensity, transmittance, percent transmission, absorbance and baseline-corrected absorbance. Most data-quality trouble starts when two similar-looking numbers silently acquire the same name.
Precision, accuracy and the role of reference materials
Repeated pH measurements can be very close together while all sharing the same systematic offset. Precision is about scatter under a specified repeatability condition; accuracy is about closeness to a trusted value or a defensible reference method. For three replicates 8.041, 8.043 and 8.042 on the same scale and temperature, the mean is 8.042 and their small spread does not establish whether the whole set is unbiased. The reference method and validation record still matter.
A quality plan combines routine cell cleaning, wavelength verification, temperature checks, indicator controls, reference or intercomparison samples, and duplicate analyses at sensible intervals. For a cruise, it also benefits from crossovers with other cruises and consistency checks against measured carbonate chemistry. Each check answers a different question. A control dye assesses one part of the optical method; a paired total alkalinity measurement offers chemical context but is not an independent magical truth.
Before interpreting a shift from 8.052 to 8.047, ask whether the quoted uncertainty reflects only within-run repeatability or a broader combined uncertainty with calibration, indicator batch, handling, temperature and sampling included. Quoting three decimals is not evidence of three-decimal accuracy. Report the quality-control protocol, the measurement scale, the temperature reference and the most defensible uncertainty alongside any conclusion about time or space.
Cross-cruise comparisons and reference-scale translation
Comparing two ocean expeditions requires more than converting their dates into the same time zone. Check whether one archive reports total-scale pH at 25 °C and another reports in-situ seawater-scale pH. Even where both use spectrophotometry, they might employ different indicator preparations or calibration equations. A difference in plotted curves may originate in processing choices rather than environmental chemistry. The safer analysis starts by harmonising definitions and identifying irreducible uncertainties.
NOAA crossover analyses explicitly discuss pH measurements from different campaigns and the relationship between historical spectrophotometric and potentiometric methods. These reports are valuable because they expose the tension between method precision and comparability across laboratories. They do not license an arbitrary global adjustment to every dataset. The reader should use the particular program’s metadata and method revisions, then describe any applied correction as a correction with a source and an uncertainty.
A harmonisation log can contain columns for cruise, station, sample, original variable name, pH scale, reference temperature, analytical method, dye lot, equation version, processing revision, quality flag and final plotted value. Where an entry is unknown, mark it unknown. Inferring a missing scale from the number’s size is dangerous because different chemical conditions overlap. A scientifically useful table exposes assumptions instead of hiding them in a single clean-looking graph.
How much change does a pH difference represent?
A decline in pH does not translate directly into a percentage decline in alkalinity, carbonate concentration or organism growth. Because pH is logarithmic, a 0.10 decrease means roughly 10^0.10, about 1.259 times the hydrogen-ion quantity within a consistent scale and convention. That is about a 26 percent increase in the corresponding hydrogen-ion measure, not a 10 percent change. The exact ecological consequences require carbonate speciation, biological context and local exposure history.
If pH changes from 8.10 to 8.00 at a comparable scale and temperature, the arithmetic is clear but the cause is not uniquely identified. Dissolved CO2 uptake, mixing with a different water mass, respiration, freshwater inflow and temperature-dependent equilibrium can all influence observed pH. Examine dissolved inorganic carbon, total alkalinity, salinity, temperature, oxygen and sampling depth to distinguish potential pathways. A pH trend is evidence about acid-base state, not a single-process fingerprint.
At the same time, a flat pH series is not proof that the ocean carbonate system is unchanged. Total alkalinity and dissolved inorganic carbon can vary together in ways that partially counterbalance pH. Some organisms respond to carbonate saturation state or other conditions, not simply the number on a pH axis. A responsible translation moves beyond the memorable phrase ‘ocean acidification’ to the actual chemical variables, their scales and the uncertainty of the inference.
Uncertainty as a measurement budget
List uncertainty contributors at the stage where they enter: wavelength-dependent absorbance noise, optical cell condition, baseline subtraction, indicator purity, dye addition, bath temperature, salinity, calibration parameters and sample handling. Some produce random scatter; others create shared offsets between samples. Combining all terms as if they were independent random errors can understate systematic effects. A strong report separates observed replicate precision from between-batch and method uncertainties wherever the available evidence permits.
Uncertainty propagates through the nonlinear transformation from absorbance ratio to pH. A small change in the numerator and denominator may cause a larger or smaller final pH change depending on the ratio and its calibration equation. Numerically perturb each measured absorbance by a plausible analytical uncertainty while holding the other fixed; re-evaluate the validated expression to see sensitivity. For correlated baseline errors, perturb both channels together rather than pretending their errors are unrelated.
Quality flags are not optional decorations attached to otherwise perfect values. If a dataset labels samples as good, questionable, bad, not reported or duplicate mean, apply the actual archive’s definitions rather than assuming every project uses the same code set. A quality-controlled map should state which classes were included. Retaining questionable observations for a sensitivity analysis may be useful, but it should not silently redefine those values as fully acceptable.
What the method cannot decide by itself
Spectrophotometric pH does not directly determine sea–air CO2 flux. That calculation needs appropriate sea and air carbon dioxide conditions, gas transfer assumptions, temperature, solubility and temporal coverage. The companion article on air–sea CO2 flux follows a different measurement chain. A reader should not convert a pH plot into an atmospheric carbon uptake claim without independent parameters and a model.
Likewise, pH alone does not reveal the biological consequences of an exposure. Temperature extremes, oxygen status, food availability, species life stage and ecological interactions can determine outcomes. A short laboratory interval may not resemble chronic exposure in a variable estuary. Refrain from inventing a causal story when the dataset contains only a chemical snapshot. State what was measured and identify the additional experiments or paired observations required for the biological question.
The seawater alkalinity companion explains a complementary chemical quantity. Its titration signal is not a pH measurement, despite the involvement of acidity in the analytical procedure. Both are valuable when interpreted as distinct observables. A system-level reader preserves these distinctions before using software to calculate carbonate speciation or aragonite saturation state.
A compact interpretation workflow for real datasets
Start by copying the data dictionary rather than only the numbers. Check the pH field’s unitless notation, operational scale, reference temperature, sample collection method, analytical equipment and data-processing version. Next, verify whether absorbances or already calculated pH values are supplied. If raw absorbances are supplied, identify correction fields and the calibration equation. Record what is missing; the absence of metadata is an information limitation, not permission to infer convenient defaults.
Second, inspect a small set of records manually. Verify wavelengths, direction of division, baseline treatment, suspicious zeros, duplicate samples and repeated station identifiers. Plot pH against depth and temperature, but use colour or shape to disclose quality flags and method changes. Check reference values or duplicates, then examine whether discontinuities coincide with instrument servicing, dye lot changes or data processing updates. This is ordinary measurement hygiene, and it prevents a large amount of false interpretation.
Third, write a claim with explicit scope. A good form is: samples analyzed spectrophotometrically and reported on the total scale at 25 °C had a particular range under stated quality criteria; the pattern covaried with named environmental variables; method and sampling uncertainties limit conclusions about the cause. Avoid saying that a single cruise proves a planetary trend. The strength of the statement should match the number, distribution and comparability of actual samples.
Worked interpretation: ratios and a suspected batch offset
Suppose two technicians measure a stable check sample. Technician A records corrected absorbances 0.450 and 0.300 at the designated wavelengths, giving R = 1.500. Technician B records 0.444 and 0.300, giving R = 1.480. The ratios differ by 0.020. This difference alone does not tell us the pH offset because the ratio-to-pH mapping is nonlinear and method specific. It does tell us that optical channels or chemistry should be investigated before pooling results.
Now imagine the same technicians repeatedly obtain their respective ratios from several check cells, with very small scatter. The absence of scatter does not rescue agreement: one or both workflows can be stably biased. Investigate indicator lot, optical baseline, wavelength accuracy, cell cleanliness and dye addition volume. Compare against a shared independent reference procedure rather than ‘correcting’ one analyst to the other by assumption. Document the repair and preserve the original data if recalculation is performed.
For a teaching sensitivity check, hold A434 = 0.300 and compare A578 = 0.450 with 0.444. The change in the numerator is 0.006, about 1.33 percent of 0.450, while the ratio changes 1.33 percent from 1.500 to 1.480. If A434 changed too, the ratio behaviour could differ. This demonstrates why the raw-channel record is useful: a pH-only archive hides much of the evidence needed to separate a chemical signal from instrumental drift.
Worked interpretation: temperature and the same bottle
Imagine a bottle analyzed at 25 °C but collected at a colder depth. A reported pH_T(25 °C) of 8.050 is neither a claim that the sea was 25 °C nor an automatic in-situ pH of 8.050. Carbonate equilibria respond to temperature, and reporting conventions may intentionally standardise measurements for comparison. To compute the in-situ value, one needs additional carbonate parameters and a correctly configured equilibrium model, not an arbitrary fixed subtraction.
Two bottle records labelled 8.050 can also reflect different conditions if one uses a total scale and another an NBS operational scale. The numerical equality is superficial. Before merging, identify the hydrogen-ion scale and conversion constants appropriate to the seawater composition and temperature. If the required information is absent, keep the measurements separate or explicitly label the comparison exploratory. A plot that looks more complete after an unjustified conversion can be less scientifically useful.
In a classroom exercise, ask whether temperature standardisation is always wrong. It is not: a consistent reference temperature can aid comparisons of measured chemistry, provided the definition is transparent. The error occurs when the standardised value is presented as the in-situ condition or compared with a differently defined series. The correct repair is a visible label such as ‘total-scale pH at 25 °C’ plus a method note about any additional calculated in-situ representation.
Documentation, communication and cross-language precision
Specialist translation is often a discipline of preserving definitions. Translate ‘total scale’, ‘seawater scale’, ‘in-situ temperature’, ‘absorbance ratio’, ‘indicator perturbation’ and ‘quality-control flag’ without collapsing them into ordinary language that implies interchangeability. In another language, an unqualified phrase equivalent to ‘acidity’ can refer to titratable acidity, proton activity or a casual taste description. Prefer a glossary that keeps the operational quantity and unit or scale attached.
For a teaching brief, one sentence can carry the entire interpretive chain: the pH value was calculated from corrected indicator absorbance ratios using a documented calibration and reported on the total hydrogen-ion scale at a stated temperature. A second sentence can describe the observed comparison and its uncertainty. This is more useful than several paragraphs declaring that the seawater is ‘healthy’ or ‘unhealthy’ without defined endpoints. Readers need verifiable measurement statements.
To develop the general habit of precise reading, use eduKate’s vocabulary learning resources to distinguish near-synonyms, inspect context and practise sentence-level meaning transfer. A technical term is not simply a difficult word to memorise: it is a compact contract about what has been measured. Change the contract and the apparent numerical claim changes with it, even when every digit is copied correctly.
Independent practice: scenario decisions and checked answers
Problem one: a table lists a total-scale pH measured at 25 °C and a second table lists in-situ seawater-scale pH. May the values be subtracted directly? Answer: no. The operational hydrogen-ion scales and temperatures differ, and the conversion requires suitable chemistry and constants. First confirm both labels and recover the method metadata. A direct subtraction would combine environmental differences with method differences, leaving the reader unable to assign meaning to the result.
Problem two: corrected absorbances are 0.390 at the numerator wavelength and 0.260 at the denominator wavelength. Compute R. Answer: 0.390/0.260 = 1.500, dimensionless. If a colleague reports R = 0.667, the wavelengths have likely been reversed. The repair is not to choose the number that gives a more attractive pH; it is to check the definition of R in the actual method and then recalculate consistently.
Problem three: a dye volume is doubled and the calculated pH changes steadily across repeated additions. Does doubling dye prove that the sample is contaminated? Answer: no. Indicator addition can disturb the sample acid-base balance, particularly in lower-buffer environments. Repeat the validated dye-perturbation check, examine indicator stock composition and calculate or experimentally determine the correction according to the method. The relationship is evidence that dye effects need assessment, not a diagnosis by itself.
Problem four: a dataset quotes 8.013, 8.014 and 8.013 and calls the result accurate to 0.001. Is that justified? Answer: those three figures suggest good repeatability under the tested conditions, but they do not establish accuracy. A systematic offset due to dye impurities or temperature could affect all three similarly. Ask for external comparison, reference checks and an uncertainty budget before making the stronger claim.
Problem five: samples collected from the same station are reported in reverse depth order, and pH seems to increase abruptly at the deepest bottle. What next? Answer: do not infer a unique physical process from one suspicious record. Verify bottle numbers, depth units, sample labels, quality flags, duplicate measurements and associated oxygen, salinity and carbon measurements. Only after the data chain is reconciled should a natural water-mass explanation be entertained.
Problem six: the ratio at one wavelength pair is correct but a colleague uses an equation published for a different indicator. Is arithmetic accuracy sufficient? Answer: no. The constants arise from the specific dye equilibrium and optical response. The resulting pH might be close by accident. Retrieve the cited calibration, match it to the indicator, temperature and salinity, and document why that equation is applicable before reporting the number.
Problem seven: a report claims that a 0.1 pH decrease means ‘10% more acidic seawater’. Answer: a logarithmic pH difference of 0.1 corresponds to about 10^0.1 = 1.259 times the relevant hydrogen-ion measure, which is roughly 26% more under consistent scale definitions. Nevertheless, this calculation does not supply a percentage change in alkalinity or biological damage. Repair both the numerical interpretation and the ecological overclaim.
Problem eight: an old and a new survey show different average pH, but the old survey used glass electrodes and the new one used spectrophotometry. Answer: method effects must be checked before attributing the entire difference to the ocean. Compare the pH scales, reference temperatures, calibrations, station coverage, seasons and uncertainties. A defensible analysis may show a change, yet the difference in instruments remains part of the uncertainty argument.
Problem nine: a pressure sensor recorded a water depth but the data set offers only the pH number. Can pH be assigned to a surface or deep layer? Answer: only if sample identity and depth mapping are verified. Do not assume the row order equals vertical order or that all bottles represent the same time. The analytical chain needs the cruise’s sample table and station metadata. A correct chemical measurement attached to the wrong water mass makes an incorrect environmental story.
Problem ten: an environmental presentation includes a colourful pH contour map and asks how much CO2 entered the ocean last month. Answer: the contour alone is insufficient. Sea–air flux requires gas-gradient and transfer information, among other things, and monthly integration requires coverage across time and space. The pH map might inform carbonate conditions, but it should not be repurposed as a direct flux measurement. Identify the missing variables rather than inventing a conversion.
Problem eleven: two analysts obtain the same pH number but one gives no uncertainty or scale. Should the two results be called equivalent? Answer: no. Equality of rounded numbers does not establish that their chemical definitions or reliability match. Request the method, temperature, scale, indicator correction and sampling details. If these cannot be found, label the comparison as limited and make no artificial claim of high agreement.
Problem twelve: a teacher wants a one-sentence summary of a well-checked profile. Answer: ‘In quality-controlled bottle samples, total-scale spectrophotometric pH at the reported reference temperature varied across depth; the pattern should be interpreted alongside temperature, salinity and the carbonate-system measurements.’ This is deliberately bounded. It preserves what the method can establish without attributing an unmeasured process or biological consequence.
Frequently asked questions
Is seawater pH unitless? Conventionally yes, because it is a logarithmic measure of a defined hydrogen-ion quantity relative to a reference. But unitless does not mean definition-free. The pH scale, temperature and analytical method remain essential. A value with three decimal places and no other metadata is not enough for high-precision chemical comparisons.
Does a higher absorbance ratio always mean a higher pH? The response direction depends on the indicator, the wavelength arrangement and the calibrated equation. For a correctly defined m-cresol purple workflow, the ratio contains information about the acid and base populations. Never transfer a direction rule from one indicator or wavelength convention to another without checking.
Is a glass electrode useless in seawater? No. Potentiometric methods have legitimate uses, but their calibration, matrix and scale conventions need careful treatment when compared with spectrophotometric measurements. The question is whether the method and accuracy are suitable for the decision being made, not whether one instrument category can be dismissed universally.
Can pH prove ocean acidification by itself? A well-controlled time series of comparable pH measurements provides direct evidence about change in the measured acid-base condition. Establishing its cause, spatial extent and broader carbonate consequences benefits from additional observations, careful sampling and physical context. A single number is not a global trend.
Why is the third decimal often discussed? Marine carbon processes may involve small changes relative to natural variability, and well-designed spectrophotometric systems can resolve subtle differences. Yet decimal places must not be mistaken for trustworthy uncertainty. Quality control and traceability are as important as numerical resolution.
What should I read next? Start with the alkalinity measurement guide for the complementary titration quantity, then the air–sea exchange guide for flux. The questions are related, but their measurements, denominators and permissible conclusions are different.
Primary methodological sources
NOAA Ocean Acidification Program, seawater pH spectrophotometric methods and indicator studies: field photometer study; dye perturbation research.
NOAA NCEI original descriptions: spectrophotometric pH method and example equation; Atlantic pH crossover methodology.
NOAA publication archive: purified meta-cresol purple interlaboratory reproducibility. These are methodological sources, not evidence that the constructed exercises here were observed in the field.