The result is a measured carbon or nitrogen mass divided by a volume
A bottle of seawater may look empty while carrying a complex mixture of plankton cells, detrital fragments and other particles. Scientists filter a measured volume, analyse material retained on the filter and report particulate organic carbon (POC) or particulate organic nitrogen (PON) as a concentration. The word particulate is operational: it refers to what the selected filter and sampling method retained. The reported concentration is not the total carbon or nitrogen in seawater, nor is it a direct estimate of carbon exported to the deep ocean. Dissolved material, inorganic particles and filtration artefacts complicate the picture.
A reader encountering POC = 80 µg C L⁻¹ and PON = 10 µg N L⁻¹ should first identify the sample volume, method, correction for filter blanks and whether the reported values are mass of element or mass of material. A number in micrograms of carbon is not the dry mass of the captured particles; carbon is only one constituent. The measurement may involve a combustion elemental analyser that quantifies carbon and nitrogen from a prepared filter, with calibration standards and a method-specific treatment of inorganic carbon. These steps belong in the meaning of the reported number.
The key question is whether the mass attributed to the seawater particles has been distinguished from mass introduced by the filter, processing and adsorbed dissolved organic matter. NASA SeaBASS guidance calls for POC and PON data to be accompanied by methodological documentation and applied adsorption blank corrections. That requirement reflects a genuine analytical challenge: a filter can retain organic material from dissolved seawater constituents even when particles are absent or reduced. Describing that contribution as a mere baseline inconvenience understates its potential effect on a low-POC sample.
Why the sampling filter defines the measurement
Glass-fibre filters and other filter materials have different retention characteristics, handling requirements and background carbon. A nominal pore size or retention designation is not a magical boundary separating every particle from every dissolved molecule. Aggregates, flexible cells and fragments interact with the filter matrix; the applied pressure, filtration volume and clogging can change retention. A sample filtered gently through a specified medium is not automatically comparable with one forced through a different membrane or operating pressure. The method must state the chosen filter type and preparation rather than only saying that seawater was filtered.
The volume actually processed matters as much as the filter type. If 500 mL was measured before filtration but only 420 mL passed through the filter because the operator stopped at clogging, dividing carbon mass by 0.500 L would underestimate the concentration. A field sheet should record the filtered volume, not simply the bottle capacity. When different samples require different volumes, comparisons remain possible after correct normalization, but blank and detection effects may not scale identically. Retain filter identity, sample identifier, volume and relevant handling flags together.
Precombustion or other filter preparation can reduce filter background, but the suitable procedure depends on the validated method and equipment. Filters can be contaminated by handling, storage containers, ambient dust or laboratory reagents. A procedural blank follows the relevant handling steps without the same particle load; a sample-specific adsorption blank tests a distinct mechanism. These controls are not interchangeable just because both produce a small carbon mass. A persuasive report explains which blank is intended to correct which contribution.
What an elemental analyser measures, and what it does not
A CHN or CN elemental analysis often uses controlled combustion and detection to quantify carbon and nitrogen in the prepared sample. Its calibration may use known standards and a method-specific blank procedure. The instrument determines elemental mass, not the species identity of each organism retained on the filter. Analysts may report a POC concentration after accounting for inorganic carbon by a validated acid-treatment or other procedure; the technique and potential loss or alteration of organic material need to be documented. Not all laboratories use the same preparation, and it is unsafe to assume a carbon value is organic merely because it came from a seawater filter.
When a filter retains carbonate-bearing organisms or mineral particles, untreated total particulate carbon can include particulate inorganic carbon. POC is the organic component, so a result needs an operational definition and a credible separation method. Similarly, measured nitrogen on a filter may include contributions that the protocol must assess before calling the result particulate organic nitrogen. If a method reports total particulate nitrogen instead, the title and column header should say that. Readers should not silently relabel TPN as PON because the quantities sound similar.
A laboratory mass in micrograms becomes a seawater concentration only after background and blank corrections and division by a documented volume. For example, a corrected carbon mass of 45 µg collected from 0.75 L equals 60 µg C L⁻¹. That arithmetic is transparent, but the corrected mass is the difficult analytical object: it needs traceable calibration, honest blanks and correct inorganic-carbon handling. A polished graph of final concentrations cannot reveal whether the mass was corrected appropriately unless the supporting metadata remain available.
Why dissolved organic matter can bias a particle measurement
Filtered seawater contains dissolved organic carbon and nitrogen. Some dissolved organic compounds can adsorb onto a filter, so the collected filter carries material that was not a pre-existing suspended particle. The effect may depend on concentration, water type, filter matrix, contact time and the volume passed. If a filter-based particle method ignores adsorption, its apparent POC or PON can be biased upward, particularly where the true particulate mass is small. A field blank made with ultrapure water may fail to reproduce this effect because the dissolved organic material in the seawater is different.
An adsorption blank attempts to estimate the contribution from dissolved material under a method that reduces or excludes particulate retention while preserving a relevant exposure of the filter to the dissolved sample. The exact configuration must be validated and described; there is no single universal blank volume or correction. A major NASA technical report on ocean POC highlights bias from dissolved organic carbon adsorption and the importance of matched field measurements for satellite product validation. The reader should therefore expect the blank procedure in the report, not assume that subtraction of a generic factory number solves the problem.
Even a well-designed blank carries uncertainty. Subtracting 15 µg of blank carbon from a filter containing 25 µg measured carbon leaves 10 µg attributed to particles, but a small shift in the blank estimate produces a large proportional change in the result. If the blank is poorly constrained, a positive corrected number can still be unreliable. The right response is to report flags or uncertainty and improve the method, not manufacture confidence by suppressing the blank record.
Worked example: four masses and one correct denominator
Suppose a seawater filter contains 68 µg carbon by the validated organic-carbon method. The procedural filter blank contributes 8 µg, while a validated adsorption blank for the sample conditions contributes a further 10 µg of a distinct, non-overlapping background. The corrected particle-associated carbon mass is 68 − 8 − 10 = 50 µg C. If exactly 0.50 L of seawater passed through the filter, the POC concentration is 50/0.50 = 100 µg C L⁻¹. These numbers are constructed for teaching; they do not assert that every POC protocol subtracts two independent blank masses. In practice a combined blank may already include the procedural contribution, and subtracting it twice would bias the result downward.
For a second sample, assume measured mass is 50 µg, the applicable combined blank is 10 µg and filtered volume is 0.80 L. The corrected concentration is (50 − 10)/0.80 = 50 µg C L⁻¹. The first sample therefore has twice the corrected mass concentration even though its measured gross filter carbon exceeds the second by only eighteen micrograms. The filter blank and denominator explain why gross filter mass is a poor comparison by itself. Reporting “first filter higher” without volume normalization does not establish a more carbon-rich water sample.
A useful worksheet keeps separate columns for sample volume, raw carbon mass, each defined blank, corrected carbon mass and final concentration. It should also show the analytical detection criterion and the data flag. If a final concentration is calculated by software, a reviewer should still be able to reproduce at least one example with a simple calculator. Transparency about the arithmetic is especially valuable when the blank correction is a large share of the gross mass.
Convert mass to amount carefully
Scientific reports may express POC as µg C L⁻¹, mg C m⁻³, µmol C L⁻¹ or mmol C m⁻³. These units are not all interchangeable as bare numbers. Because 1 µg per litre equals 1 mg per cubic metre, 100 µg C L⁻¹ is numerically 100 mg C m⁻³. To convert a mass of carbon to chemical amount, divide micrograms by approximately 12.01 µg per µmol. Thus 100 µg C L⁻¹ corresponds to about 8.33 µmol C L⁻¹. The units make the denominator explicit: the value describes amount of carbon per litre of sampled water, not carbon atoms counted directly.
Nitrogen requires its own molar mass. A concentration of 14 µg N L⁻¹ is approximately 1.00 µmol N L⁻¹ because the atomic molar mass is about 14.01 g mol⁻¹. Dividing the carbon mass value by fourteen when it represents carbon, or using a molecular weight for nitrogen gas N₂ instead of elemental nitrogen mass, leads to errors. The analyser’s output and column label should specify elemental mass. A good conversion worksheet records the elemental symbol and the molar mass used so a non-specialist can audit the change.
Volume conversion matters across sampling systems. A 1.2-L filtered volume is 0.0012 m³; a carbon mass of 120 µg collected from it is 100 µg L⁻¹ or 100 mg m⁻³. If an analyst enters 1.2 into a spreadsheet cell labelled millilitres, the result will be off by a factor of one thousand. Similar errors can occur when a software export stores filtered volume in millilitres but its documentation uses litres. Never infer the unit from how plausible the final environmental concentration appears.
Understand the C:N ratio before interpreting it
A POC:PON ratio may be presented as mass C:N or molar C:N. Those differ because carbon and nitrogen atoms have different masses. Suppose POC is 120 µg C L⁻¹ and PON is 14 µg N L⁻¹. The mass ratio is 120/14 ≈ 8.57 by mass. On a molar basis, carbon is about 120/12.01 ≈ 9.99 µmol L⁻¹ and nitrogen is 14/14.01 ≈ 1.00 µmol L⁻¹, so molar C:N is approximately 10.0. An article that quotes “C:N = 8.57” without telling the reader whether it is mass or molar is incomplete.
A commonly invoked reference for marine organic matter is the Redfield-like carbon-to-nitrogen atomic ratio around 106:16, or 6.625 molar. It is a conceptual benchmark, not a constant binding every marine particle, ocean region or plankton group. Higher measured C:N can reflect several processes, including detrital composition, nitrogen limitation or mixing with carbon-rich material, but none follows uniquely from the ratio. Analytical blanks, inorganic carbon, nitrogen detection limits and differing particle retention can all distort it. The ratio supports hypotheses that should be evaluated with environmental and methodological context.
If PON is near its detection limit, the ratio becomes unstable. For instance, dividing a well-measured carbon concentration by a nitrogen result that could plausibly range between 0.5 and 1.5 µmol N L⁻¹ yields a wide C:N interval. Reporting a ratio to three decimal places would pretend the denominator is precise when it is not. In some cases it is safer to report a qualified lower bound, an interval or no ratio, depending on validated reporting rules. A zero or negative blank-corrected nitrogen estimate must never be silently used as a positive denominator.
A filtering blank is not the same as a true sample
Consider a field duplicate where two independent water aliquots pass through separate filters. The difference between them may reflect real sample heterogeneity, filter handling and analytical variation. A duplicate made by splitting one filter after collection tests something else and may not represent full filtration variability. A procedural blank that follows handling without seawater particles detects laboratory and material background; an adsorption blank with particle-reduced seawater probes retention of dissolved substances. Using the same word “blank” for all three hides the reason each exists.
Blanks should be prepared and analysed at appropriate frequency, and their variability should be measured rather than assumed. A single small blank value collected weeks earlier may fail to represent a different filter lot or a new water mass rich in dissolved organics. That does not prove every sample needs a unique blank; it means the method should justify how blank samples represent analytical conditions. Laboratories may use batch-specific or sample-specific correction designs. Readers need sufficient documentation to see whether the correction was actually suitable for the conditions sampled.
Carryover and contamination during filter processing can be subtle. Forceps, storage foil, airborne dust and handling gloves can introduce carbon or nitrogen. Acid fumes used for inorganic-carbon treatment may require careful handling protocols and can change how certain materials are retained; the laboratory’s validated procedure matters. A blank may reveal a problem but cannot reconstruct the true sample if contamination varied unpredictably. Quality flags are therefore a meaningful part of results, not a sign that the laboratory did not know how to calculate.
How a high-POC sample can still be analytically problematic
In turbid water, a large particle load can clog a filter quickly. Filtration may stop before the intended volume passes, and particles may accumulate unevenly. A large gross carbon mass does not automatically mean the measurement is more reliable: overloading can create combustion or calibration range issues, while filter retention and heterogeneous deposition may affect reproducibility. A validated method specifies acceptable loading and how to handle samples that exceed it. The sensible response may be to filter a smaller measured volume, not force a clogged filter until pressure changes the sample.
Dilution or subsampling is another choice with consequences. If 100 mL of a thoroughly mixed original sample is diluted to 200 mL and the whole dilution is filtered, the final organic carbon mass still came from 100 mL original water. Dividing by 200 mL would understate the original concentration by half. Any correction for a procedural blank introduced by the added diluent must be applied under the validated method. This example illustrates why water-handling steps matter even when the final CHN result is a precise mass.
At the opposite extreme, oligotrophic clear water may require larger sampled volumes to collect enough particulate material for a robust result. But increasing volume can also increase dissolved organic matter adsorption to the filter. There is no universally optimal volume independent of water type and method. The sampling design should balance particle load, blank contribution, filter retention and instrumental range. Reporting the actual volume alongside every measurement is therefore essential to making an intercomparison meaningful.
Uncertainty when subtraction dominates the result
Suppose a sample yields 30 µg C gross mass and an applicable combined blank of 20 µg, leaving 10 µg corrected POC. If independent standard uncertainties are 2 µg for the gross mass and 3 µg for the blank, the corrected mass standard uncertainty is approximately sqrt(2² + 3²) = 3.61 µg. That is about 36 percent of the ten-microgram net estimate, before filtration-volume or representativeness uncertainty. Quoting the final concentration without acknowledging this subtraction would mislead a reader about its precision. The numbers are illustrative, not a universal detection threshold.
Now suppose the volume is 0.50 L with a standard uncertainty of 0.01 L. The concentration is 20 µg C L⁻¹. A first-order independent relative uncertainty approximation combines 3.61/10 for net mass with 0.01/0.50 for volume, giving sqrt(0.361² + 0.020²) ≈ 0.362, or about 36 percent. The concentration standard uncertainty is therefore around 7.2 µg C L⁻¹ under the toy model. When blank and sample errors are correlated, a covariance term is needed. The calculation demonstrates why blank reproducibility is often more important than another decimal place in recorded water volume.
A negative corrected net mass can occur statistically when a small sample is overwhelmed by variable blank contributions. It does not imply negative organic particles in the ocean. The correct handling depends on laboratory detection and reporting rules. Some protocols flag a below-detection result, some report a qualified numerical estimate and its uncertainty; arbitrary clipping to zero can distort averages and ratios. A good data file preserves the gross mass and blank information so future users can understand how the reported result arose.
POC concentration is not export flux
An optical satellite product or a shipboard filter measures a property of water at a place and time. Carbon export flux asks how much particulate carbon crosses a surface per unit area per unit time, such as mg C m⁻² d⁻¹. The two quantities have different denominators. A bottle yielding 100 µg C L⁻¹ does not tell us how many milligrams crossed a depth horizon per day. Estimating flux requires additional processes, such as settling, sinking, mixing, particle turnover or a validated trap-based approach. A concentration profile may inform a flux model, but it is not itself a flux measurement.
A related marine sediment imaging article addresses a different archive of deposited material, while the optical backscatter guide explains a light-scattering proxy. Neither can directly replace filter-based elemental analysis. Optical models can estimate POC from spectral properties using calibration, but those products should be identified as algorithms rather than treated as independent filter measurements. The distinction matters especially when the same field samples are used to train and test a satellite model.
A useful sentence reads: “The sampled water contained a blank-corrected particulate organic carbon concentration of X µg C L⁻¹ under the stated filtration and elemental-analysis protocol.” That is narrower but far more accurate than saying “X micrograms of carbon sank” or “the plankton fixed X micrograms”. Scientific communication often improves by keeping the numerator and denominator visible in prose.
A complete paired POC and PON example
A teaching laboratory filters exactly 0.60 L of seawater and reports 92 µg carbon and 20 µg nitrogen in the processed sample filters. Its appropriate validated blank correction is 14 µg carbon and 5 µg nitrogen, respectively. The net masses are 78 µg C and 15 µg N. Dividing each by 0.60 L gives POC = 130 µg C L⁻¹ and PON = 25 µg N L⁻¹. Their mass ratio is 130/25 = 5.20. Converting to molar amounts gives POC ≈ 130/12.01 = 10.82 µmol C L⁻¹ and PON ≈ 25/14.01 = 1.784 µmol N L⁻¹. Thus the molar ratio is roughly 6.06.
The values are constructed and the shared filtration volume is part of the premise. If carbon and nitrogen were measured on separate aliquots with different filtered volumes, mass values would first need separate normalization. In an actual dataset, POC and PON blanks may be determined with different uncertainties; the ratio is not automatically more precise because both elemental masses came from one filter. The calculation is worth writing in full because readers often mistake the mass C:N of 5.20 for the molar C:N of 6.06. Both are legitimate quantities only when correctly labelled.
The example should not be recast as proof that the water’s particulate matter has “ideal” Redfield composition. The molar ratio is close to a familiar reference, but organic matter contains mixed organisms and detritus and can be influenced by blank correction or other chemical contributions. A hypothesis about nutrient limitation or ecological state requires additional evidence. The correct translation preserves the arithmetic and avoids treating a descriptive ratio as a causal conclusion.
A two-column quality review: evidence and decision
A useful audit separates the observation from the decision it supports. Observation: a reference filter shows unexpected extra carbon after handling. Decision: review contamination and batch-specific blanks; do not silently subtract a guessed correction. Observation: filtered volume differs from the intended bottle volume. Decision: normalize by the recorded volume actually processed. Observation: the CHN result exceeds the instrument’s validated calibration range. Decision: follow the laboratory’s approved reanalysis or dilution procedure rather than extrapolate beyond the range. Observation: a carbonate-rich sample appears high in total particulate carbon. Decision: check how inorganic carbon was distinguished from organic carbon.
Another set of observations concerns time and storage. If filters from one cruise were frozen promptly while another set spent hours at warmer temperature, comparability needs evaluation. If a precombusted filter lot changes, the blank history should be revisited. If replicates diverge after handling, investigate sample heterogeneity and process controls before assigning the difference to seawater. None of these checks guarantees perfect evidence. They do something more practical: reveal which decisions are supported, which measurements require caution and which records should remain uncombined.
There is also a risk of double counting blank corrections when two stages of processing each apply an offset. For example, a data-export program may already provide blank-corrected organic mass, while an analyst’s spreadsheet subtracts the same blank again. The resulting concentration could be much too low or even negative. A good workflow states whether each column is raw, procedural-blank corrected, adsorption-blank corrected or final, and shows transformations in the order actually performed.
Practice questions with worked answers
Practice one: A corrected POC mass of 36 µg is collected from 300 mL of seawater. The concentration is 36/0.300 = 120 µg C L⁻¹. Expressed as mg C m⁻³ it is also 120 because 1 µg L⁻¹ equals 1 mg m⁻³. Expressed as µmol C L⁻¹ it is approximately 120/12.01 = 9.99. If the analyst divides by 300 without converting millilitres to litres, the final number becomes a thousand times too small. This problem tests the denominator rather than the carbon chemistry.
Practice two: Gross carbon is 21 µg, a matched combined blank is 25 µg and filtered volume is 0.50 L. The arithmetic net is −4 µg, or −8 µg C L⁻¹. That is not a physically negative amount of particulate organic carbon. Investigate blank applicability, analytical variation and the laboratory’s detection/flagging conventions. Do not report a confidently negative ecological concentration; also do not conceal the diagnostic by changing the value to zero without recording that decision. The raw and corrected numbers should remain traceable.
Practice three: POC is 72 µg C L⁻¹ and PON is 12 µg N L⁻¹. Mass C:N is 72/12 = 6.0. Molar C:N is (72/12.01)/(12/14.01), about 7.0. The molar ratio is not equal to the mass ratio and neither has concentration units because numerator and denominator units cancel after the appropriate conversion. A report that says only “C:N = 6” is ambiguous even though an arithmetic operation was correct.
Practice four: A filter with 48 µg corrected carbon was collected from 0.80 L. A second with 36 µg was collected from 0.40 L. The concentrations are 60 and 90 µg C L⁻¹, respectively. The second water sample has the larger concentration despite carrying a smaller collected total mass on its filter. This is a direct demonstration that filter load and environmental concentration are different quantities. Any comparison must also consider whether both samples passed compatible blank and calibration quality checks.
Practice five: A study reports two POC measurements with similar values but very different adsorption blanks relative to gross filter masses. Which one is necessarily better? Neither can be judged from the ratio alone. Examine uncertainty of the blank, filter and CHN method, volume, sample matrix, calibration and replication. A small blank is helpful but does not guarantee correct organic-carbon separation; a larger well-characterized blank may still permit a usable estimate. Quality is demonstrated by evidence, not by the most reassuring-looking correction.
Detection, censoring and the temptation to keep only positive results
When the true particle signal is near the method’s blank, measured net masses may scatter on both sides of zero. If an analyst retains only positive corrected values, the reported average can become biased upward because negative noise realizations are systematically discarded. A proper reporting strategy follows a defensible detection-limit and censoring protocol, sometimes using qualified values or statistical methods suitable for censored observations. There is no universal numerical detection limit that can be borrowed from another laboratory’s instrument without checking its blanks and method performance.
The same concern affects C:N ratios. If PON is frequently below detection while POC remains quantifiable, calculating ratios only for high-PON samples may bias the apparent distribution. It may be more useful to report available POC and PON separately with flags and avoid unsupported ratios for censored denominators. A reader should be able to distinguish “not detected above this method limit” from “physically absent”. The former is an analytical statement; the latter is a much stronger claim that the experiment often cannot support.
Field design matters before the analyser is switched on
Sampling location, depth, timing, replication and filter volume should be designed around the question being asked. A single bottle from a bloom edge cannot establish a basin-wide carbon concentration. If the aim is a seasonal comparison, matching depths and accounting for water-mass structure matter. If the aim is particle composition, independent pigment, microscopic, isotopic or other biogeochemical analyses may be needed. Good analytical chemistry can measure a carefully chosen sample accurately while still failing to represent the region a headline describes.
Depth integration is also different from point concentration. To estimate an inventory in mg C m⁻², integrate concentration in mg C m⁻³ over depth in metres. For a hypothetical uniform 50 mg C m⁻³ across 20 m, the inventory is 1,000 mg C m⁻². That still does not give a rate of sinking or production. Real profiles may have strong gradients, so use observed depth intervals with a justified numerical integration. The unit on the final inventory shows that a length has been integrated, while the absence of a time denominator shows it is not a flux.
Frequently asked questions
Is POC the same as dissolved organic carbon? No. The operational separation depends on filtration, and a blank is needed partly because dissolved organics may adsorb. Is particulate carbon automatically organic? No; carbonate and other inorganic carbon can be present and require a validated method of distinction. Does a CHN number identify a species? No; it estimates elemental mass in the prepared material. Does a filter retaining more carbon prove the water has higher POC concentration? Not unless the filtered volumes and corrections are considered.
What does a high C:N mean? It describes elemental proportions under the stated convention; it does not uniquely diagnose nutrient limitation, species or decay state. Why are blanks so important? Their signal can be substantial relative to low particle masses, and subtraction affects both the estimate and its uncertainty. Why retain a negative corrected value? Because it may reveal a limit or bias in the method; it should be handled by explicit reporting rules. Can POC be estimated from satellite ocean colour? Yes, using models and validation, but a remote estimate is not a direct filter mass.
The most reliable short description says which water was filtered, what mass was measured, how organic carbon or nitrogen was distinguished, what blank was applied, and what actual water volume formed the denominator. If the text goes further toward a statement about ecosystem processes, it needs independent support. This separation between observation, correction and ecological inference is the central skill for translating laboratory results into useful knowledge.
An independent audit of a small-particle-carbon result
Imagine an external reviewer receives a results table with values described as POC but no blank correction column. The first question is whether the quoted masses are already corrected and, if so, how. The next is whether the filtration protocol exposed filters to dissolved organic matter in a way that was represented by the chosen blank. Then the reviewer checks actual filtered volumes, calibration range, filter preparation, inorganic-carbon removal and sample handling. Those requests are not administrative obstacles. They determine whether two laboratories’ POC numbers represent comparable fractions of seawater carbon. Without them, a pooled map or global algorithm may combine systematic differences as if they were natural variability.
A useful repair is to release the raw filter mass and blank information alongside the corrected concentration, subject to the data programme’s conventions. The analyst can recompute each conversion, flag samples with a low net-to-blank ratio and compare duplicate filters. Reprocessing should be versioned, not silently substituted for earlier measurements. If correction cannot be justified retrospectively, retaining a qualified observation is better than inventing a reference value. Transparent uncertainty allows future researchers to decide whether a measurement is appropriate for trend analysis or satellite validation.
For an illustrative decision, compare two results: 70 ± 5 and 73 ± 6 µg C L⁻¹, where the uncertainties are independent standard uncertainties under a stated model. Their difference is 3 µg C L⁻¹, and its standard uncertainty is sqrt(5² + 6²), about 7.8 µg C L⁻¹. The difference is smaller than the uncertainty in that difference, so it should not be described as a confirmed increase from these measurements alone. A trend claim would need more observations and an analysis appropriate to their error structure. This final check connects laboratory rigor with responsible prose: the denominator, correction and uncertainty all belong to the meaning of the number.
Primary sources and further reading
NASA SeaBASS POC and PON submission requirements, including adsorption blank corrections; NASA Technical Reports Server biases in in-situ particulate organic carbon measurements; and NASA ocean-colour data-product context. Numerical case studies are original teaching examples; they are not sourced field measurements.