A sunlight profile contains both physics and a clock
A vertical profile of light in the ocean looks like an uncomplicated plot: deeper sensors usually receive less sunlight. The interpretation becomes difficult when the instrument moves through a changing sky, when two different definitions of light are mixed, or when an exponential model is fitted beyond its useful range. Photosynthetically available radiation, commonly PAR, is usually considered over roughly 400–700 nanometres. An underwater PAR sensor often measures photosynthetic photon flux density, expressed as micromoles of photons per square metre per second. The attenuation coefficient Kd(PAR) describes the apparent reduction of downwelling light with depth. These are not interchangeable quantities: one is a light flux and the other an inverse depth. The one-percent euphotic depth is a particular threshold derived from a specified reference light field. This article explains the quantities, calculations, controls and limitations with illustrative examples, not fabricated surveys. The goal is to help a reader recognize which number the ocean observation actually supports.
What PAR means in practical units
PAR refers to a spectral range of solar radiation relevant to photosynthesis, not to an automatic measurement of primary production. Quantum sensors usually report photon numbers because photosynthetic reactions respond to photons; a radiometric instrument may instead measure energy in watts per square metre. A photon at the blue end carries more energy than one at the red end, so converting between energy and photon flux requires spectral information or an explicitly justified approximation. A reported 100 µmol photons m⁻² s⁻¹ cannot be renamed 100 W m⁻². Nor should a daily integral in mol photons m⁻² day⁻¹ be presented as an instantaneous measurement. Identify the sensor’s spectral response and calibration basis, not just the label PAR. A numerical result only becomes meaningful when the quantity, units, measurement interval and location are clear.
Understand downwelling irradiance
The symbol Ed describes downwelling irradiance, light arriving from the downward hemisphere at a horizontal surface. Its definition includes angular weighting and can be spectrally resolved at wavelength λ or integrated over a band. A flat light collector responds differently from a narrow field-of-view radiance instrument. The first integrates incident directions; the second describes light from a selected direction per unit solid angle. Ocean optical datasets sometimes mix these words, particularly when a satellite retrieval is summarized for general readers. A PAR sensor reporting underwater irradiance does not directly measure water-leaving radiance, absorption coefficient or total light energy in the water column. The physical geometry matters because the light field varies with sun elevation, clouds, wave focusing and particles. Preserve the sensor’s viewing or collecting geometry and its cosine-response performance when explaining why two instruments disagree.
The simplest exponential model
For a layer where the attenuation coefficient can be approximated as constant, downwelling PAR can be represented by E(z) = E(0) exp(−Kd z), with depth z positive downward in metres and Kd in m⁻¹. The model predicts a ratio E(z)/E(0) rather than an absolute measurement for every possible ocean condition. E(0) must be defined consistently; it may mean just below the water surface, not the irradiance in air above it. The negative exponent signifies decreasing light with depth. If Kd = 0.10 m⁻¹, a layer ten metres thick transmits exp(−1) ≈ 36.8% of the reference signal under the model. This is an idealized calculation. Real profiles can have changing attenuation with depth, changing sunlight during the cast, wave effects and instrument shading, all of which demand inspection before interpreting fitted coefficients as environmental constants.
How to derive Kd between two depths
Taking natural logarithms of the exponential equation gives ln E(z) = ln E(0) − Kd z. For two depths z1 and z2, with z2 deeper and readings positive, Kd = ln[E(z1)/E(z2)]/(z2−z1). Suppose a synthetic cast reports 600 µmol photons m⁻² s⁻¹ at 2 m and 300 at 7 m. The depth difference is 5 m, the ratio is 2, and Kd = ln 2 / 5 ≈ 0.1386 m⁻¹. The absolute sunlight level cancels if both readings belong to the same steady reference conditions. But the cancellation is only valid when changing atmospheric illumination has not contaminated the profile. A time-separated cast under racing clouds needs an independent surface reference or a procedure that removes that temporal variation. The mathematical ratio cannot repair a badly timed measurement by itself.
The one-percent depth equation
If the same constant-Kd model applies from the below-surface reference to the threshold depth, the one-percent euphotic depth Zeu satisfies E(Zeu)/E(0) = 0.01. Therefore Zeu = ln(100)/Kd ≈ 4.60517/Kd. At Kd = 0.10 m⁻¹, the model yields Zeu ≈ 46.1 m. At Kd = 0.20 m⁻¹ it yields 23.0 m. Thus stronger attenuation produces a shallower one-percent threshold. This depth is a useful operational measure of the light environment; it is not a direct assay of active photosynthesis, an absolute lower boundary for life, or a fixed ecosystem response threshold. A project may define its euphotic zone differently or use measured threshold crossing rather than a fitted constant coefficient. Always report which definition was used and where the surface reference came from.
One percent of which surface signal?
The surface reference can refer to irradiance just above the sea surface, just below it, or a value inferred from a regression intercept. The sea-air boundary reflects some incident light, and the fraction entering water depends on illumination and surface state. A below-surface ratio is not automatically the same as one percent of an above-surface deck sensor. When a methods section says ‘one percent PAR depth’, inspect the definition. If the calculation used an in-water extrapolated E(0−), note that choice. If it used simultaneous deck light readings to normalize the cast, show the normalization explicitly. A one-percent threshold obtained from two different reference conventions can differ even when the subsurface optical conditions are identical. Precision of words around the zero-depth datum is therefore part of numerical accuracy, not a stylistic preference.
Kd(PAR) is not automatically Kd(490)
Satellite ocean-colour products often provide Kd(490), the diffuse attenuation coefficient of downwelling irradiance at approximately 490 nm. PAR covers a broad visible band and may be measured with a quantum sensor. The spectrum of sunlight changes with depth because different wavelengths are absorbed and scattered differently, so broadband Kd(PAR) and monochromatic Kd(490) are not generally equal. NASA’s ocean-colour products identify Kd(490) with units m⁻¹ and algorithms based on remotely sensed reflectance. A user who copies a 490-nm value into the PAR euphotic-depth equation without evaluating its applicability adds an unacknowledged modelling assumption. Remote estimates of light penetration can be valuable, but the algorithm and water type need to be specified. The right caption names wavelength or band, instrument or retrieval, and definition of attenuation.
Kd is an apparent, not an inherent, optical property
Absorption and scattering coefficients characterize material interactions with light; Kd depends additionally on the angular and spectral structure of the illumination field. It is called an apparent optical property because the sun’s position, cloud conditions and water’s optical characteristics together shape the downwelling irradiance profile. A change in Kd need not result from one constituent alone. ACDOM, phytoplankton pigments, suspended minerals and water itself can all contribute to absorption and scattering that influence PAR. A profile of Kd cannot uniquely decompose those constituents. If an editor writes that Kd = 0.20 proves a specific concentration of chlorophyll or sediment, the claim requires an independent model and validation. The honest result is an attenuation measure under stated conditions, followed by qualified hypotheses about its drivers.
The deck reference compensates for moving clouds
During a shipboard cast, a surface reference sensor can measure sunlight at the same time the submerged sensor descends. Normalizing underwater readings by the deck reference can reduce the effect of cloud-driven intensity changes, if the instruments are appropriately calibrated, time-aligned and their geometries understood. Without such a reference, a rapid fall in solar irradiance during descent may masquerade as stronger vertical attenuation. Suppose underwater light halves while the deck signal also halves over the same interval. A naive in-water fit exaggerates vertical loss; a correctly normalized ratio may show that much of the apparent decline came from the sky. The surface sensor must not be shaded by the ship, and time stamps must match. Normalization does not solve wave focusing, angular errors or depth-specific scattering, but it addresses one important and often overlooked source of bias.
Check the sensor’s cosine response
An ideal horizontal irradiance sensor responds to incoming rays according to the cosine of their incidence angle. Real collectors have imperfect angular response, which can be especially important near the surface when the light field contains strong directional sunlight and reflected skylight. Tilt changes which directions the sensor favours. Instruments mounted on a profiler or buoy should have orientation metadata and appropriate correction or quality screening where the method requires it. A reading collected while the sensor is angled toward the sun may not be comparable to one with the collector level. The effect can vary with depth as illumination becomes more diffuse. A smooth logarithmic profile does not prove that angular errors were absent. Instrument calibration must cover spectral and angular behaviour relevant to the measurement, not merely a single bench-top brightness point.
Wave focusing creates real fluctuations
Sunlight refracts through the moving sea surface, forming bright and dim patches in shallow water. An underwater sensor may experience large short-term changes while remaining at the same depth. Those oscillations are part of the light field, not automatically instrument noise. Averaging over suitable time and space can estimate a representative level, but the averaging interval must be chosen for the study. A very short average may preserve surface caustic variability; an excessively long average during changing weather may mix different conditions. Keep the raw record, document the aggregation and specify whether wave-induced spikes were included. In rough seas, it can be useful to estimate attenuation from a depth range where wave focusing is less dominant. The correct choice depends on water clarity and profile coverage. Treating every bright pulse as a measurement error loses potentially important information about near-surface exposure.
Shadows from the platform are not part of the water
A ship hull, rosette frame, winch wire, diver or autonomous vehicle can cast shadows on an optical detector. A shadowed reading may suggest an abrupt increase in attenuation that is entirely due to instrument placement. Check deployment side relative to the sun, the rig’s geometry, heading, wave conditions and whether the sensor’s field of view is obstructed. If a profile has a narrow dark interval coincident with a change in platform orientation, investigate shading before fitting Kd. Many field protocols aim to deploy optical sensors away from the vessel’s shadow. A data processor should preserve the screening reason, not simply erase the anomalous segment. The public figure can report that certain depths were excluded for platform shading. This gives readers a useful boundary around the inference rather than a deceptively flawless curve.
Dark offsets matter at depth
As PAR declines with depth, the sensor signal may approach its electronic dark offset or quantification limit. Subtracting a small uncertain offset from a small observed signal can produce large relative error, or negative corrected readings. Taking logarithms of those estimates is invalid when they are nonpositive. A deep profile that flattens into an apparent light floor may reflect instrument noise rather than a real constant sunlight intensity. Measure and account for dark response following the instrument’s procedure, and set an explicit quality threshold for the usable depth range. Do not force the logarithmic regression through deep measurements below that threshold. The final reported Zeu should indicate whether it was observed within the reliable range or extrapolated from shallower data. A claim of extreme light penetration is weak if it depends on readings indistinguishable from dark noise.
An approximate fit must disclose its depth interval
The water column is often layered. Surface phytoplankton, a clear middle layer and a turbid bottom nepheloid layer may have different attenuation coefficients. A regression across all depths can describe an average slope without representing any particular layer well. For an idealized two-layer example, let Kd = 0.08 m⁻¹ from 0–20 m and 0.18 m⁻¹ below 20 m. The top layer transmits exp(−1.6) ≈ 20.2% of the surface PAR. To reach one percent, the lower layer must reduce that remaining fraction by a factor of about 20.2, requiring roughly ln(20.2)/0.18 ≈ 16.7 additional metres. The resulting threshold is near 36.7 m, not the 57.6 m predicted by applying the surface coefficient alone. State layer boundaries and the logic used to integrate the light profile.
Why the regression intercept needs scrutiny
A log-linear fit ln E = b − Kd z provides an estimated intercept exp(b). It can be used to extrapolate the light field to zero depth, but this estimate inherits all limitations of the fitted range. If the profile starts at 10 m after an unknown surface layer, extrapolating the deep-water slope to the surface can misrepresent true E(0−). A statistically excellent fit to the deep points does not make the extrapolation physically correct. Compare the intercept with available near-surface observations and the deck reference, while acknowledging the sea-air interface. A descriptive plot should show the fitted interval, residuals and the position of the one-percent threshold. This lets the reader determine whether the estimate was interpolated within data coverage or extrapolated beyond reliable observations. The difference is fundamental to the claimed confidence of Zeu.
Calculate attenuation from a complete example
Imagine a synthetic clear-water cast with simultaneous, normalized PAR of 800 at 1 m, 400 at 11 m and 200 µmol photons m⁻² s⁻¹ at 21 m. Each ten-metre interval halves, so Kd = ln 2/10 = 0.06931 m⁻¹ under the homogeneous model. If the fit is extrapolated to z = 0, the below-surface reference is E0 ≈ 800 exp(0.06931) = 857.4. Zeu ≈ 4.60517/0.06931 = 66.44 m. However, the measurements only extend to 21 m, so the one-percent depth is an extrapolation far below sampled depths. The arithmetic is clear but the environmental confidence is limited without evidence that attenuation stays similar. A responsible result states the model, range and extrapolated status instead of calling 66.44 m a directly observed boundary.
Calculate a turbid-water example
Consider readings of 600 at 0.5 m and 60 µmol photons m⁻² s⁻¹ at 8.5 m under stable illumination. Light falls by a factor of ten across 8 m, giving Kd = ln 10/8 ≈ 0.2878 m⁻¹. Under the constant-layer model, the one-percent depth is 4.6052/0.2878 ≈ 16.0 m from the below-surface reference, assuming that reference and attenuation hold. The 8 m interval itself only establishes a tenfold loss, not the complete one-percent threshold. If the bottom lies at 12 m, the predicted threshold is below the seabed, and no 16-m-water-column euphotic boundary physically occurs there. For a shallow coastal study, report that the modelled one-percent depth exceeds water depth rather than plotting a threshold beneath the sediment. Environmental context must be applied after the calculation.
Why one-percent depth is not a fixed photosynthesis limit
Plants and phytoplankton have different photosynthetic responses to light, and the compensation irradiance at which photosynthetic production balances respiration is not universally one percent of surface PAR. Species, temperature, nutrients, acclimation and mixing affect growth. The euphotic threshold is therefore an operational optical convention, not a biological switch. Some organisms may photosynthesize at lower percentages, while others have conditions that prevent significant net growth at one percent. A report can say that the one-percent light level occurred at a given depth. It should not claim that all photosynthesis ends exactly there. If an ecological conclusion concerns net primary production, it needs appropriate physiological or production measurements and a model of exposure through time. This separation of optical and biological definitions prevents readers from mistaking a convenient depth index for a universal boundary of marine life.
Spectral colour shifts with depth
Broadband PAR integrates light over many wavelengths, but the underwater spectrum changes with depth as red and other strongly absorbed wavelengths disappear more rapidly than some blue-green bands. This means a single Kd(PAR) may not remain constant even in water with relatively uniform constituent concentrations. The photon-weighting convention of the sensor further matters. Two instruments with different spectral responses may disagree when the spectrum shifts, although both are marketed as PAR sensors. For precise comparative studies, examine spectral radiometry or sensor calibration against the expected underwater spectrum. If the scientific question concerns a particular pigment or photobiological response, broadband PAR may be insufficient. A useful explanation says that PAR is a band-integrated photon measure and identifies when that simplification is adequate for the desired inference.
Depth, pressure and vertical coordinates
Most ocean profiles record pressure, but their axes are often labelled depth. Converting pressure in decibars to depth in metres depends on latitude, gravity and water density through appropriate oceanographic relationships. The rough near-surface similarity between one decibar and one metre is not an exact identity at all depths and locations. For high-precision attenuation calculations, use a documented pressure-to-depth conversion and maintain the sign convention. Some datasets define positive z upward or negative depth below surface; plugging negative coordinates into an equation designed for positive-down depth can reverse the fitted slope. A curve that appears to increase toward the seabed may simply have a sign or axis error. Retain geolocation and the coordinate convention, and verify that two successive depths correspond to the intended vertical distance. Even elegant optical calculations fail when their geometry is mislabelled.
Calculate a depth-unit error
An analyst sees two points separated by 500 cm and treats the difference as 500 m. If their light ratio is 2, the incorrect Kd is ln 2/500 ≈ 0.001386 m⁻¹. The correct separation is five metres, yielding Kd = ln 2/5 ≈ 0.1386 m⁻¹, one hundred times larger. The implied modelled one-percent depth shifts from roughly 3322 m to 33.2 m. A simple unit error becomes a spectacular environmental claim. Before trusting published light-penetration numbers, inspect whether distance, metres and centimetres have been converted consistently. Check whether a sensor’s output uses pressure units instead of physical metres. A ten-minute audit of one pair of records can catch an error that no amount of editorial polishing will fix.
The difference between instantaneous and daily light
An instantaneous PAR measurement has units such as µmol photons m⁻² s⁻¹. The daily light integral is the time integral, commonly expressed in mol photons m⁻² day⁻¹. In a deliberately simple exercise, a constant flux of 100 µmol photons m⁻² s⁻¹ lasting exactly two hours contributes 100 × 7200 / 10⁶ = 0.72 mol photons m⁻². The real sun does not supply a constant two-hour flux, so a daily integral generally needs a time series or an appropriate integration model. A profile collected at noon does not tell you the daily light dose at every depth. Cloud variability, sunrise and sunset, tides and water motion matter. When comparing ecological exposure, distinguish a brief light measurement from the integrated exposure experienced by moving plankton. The numbers answer different scientific questions.
Photons reaching a depth are not photons used biologically
A small photonic flux at depth may still influence photosynthesis, photochemistry or behaviour, but measured downwelling PAR does not reveal how many photons were absorbed by a particular organism. Particle absorption, cell orientation, pigment packages and physiological responses intervene. To calculate primary production, additional observations or modelling are required. A dataset that pairs oxygen production with light can support a response curve under specified conditions, but the PAR sensor alone does not measure oxygen generation. Avoid language such as ‘the instrument measured photosynthesis of 40 units’ when the instrument measured photons. A clear article gives the optical variable first, then specifies which assumptions or independent assays transform it into a biological estimate. This distinction keeps ocean optics useful without overselling its inferential reach.
A Secchi disk is not a quantum sensor
Secchi depth is a visual transparency measure based on the disappearance and reappearance of a disk under specified observation conditions. It is inexpensive and useful for monitoring, but it is not directly identical to Zeu, Kd(PAR) or Kd(490). Empirical relationships between Secchi depth and attenuation coefficients vary with water type, illumination, observer, disk method and optical constituents. A project can derive a local calibration, but applying one universal multiplier across clear oceans and highly coloured estuaries is unsafe. If a field log contains only Secchi depth, preserve that variable. Do not relabel the observation as a direct measurement of one-percent depth. When comparing with PAR sensors, report the empirical relationship and uncertainty. A good scientific translation tells the reader whether a number was observed, fitted or inferred from a proxy.
Satellite light-penetration maps have boundaries
NASA and NOAA distribute ocean-colour products that estimate diffuse attenuation from spectral reflectances, often at 490 nm. These can reveal regional patterns over times and places where in-water sensors are unavailable. But clouds, aerosol correction, glint, shallow bottoms, strongly coloured water and algorithm assumptions can affect retrievals. A satellite estimate at a pixel is also spatially averaged and may not match a single buoy or small coastal inlet. Product metadata specify band, processing level and algorithm version. A reader who sees a caption ‘satellite measured 30 m euphotic depth’ should ask whether that depth was retrieved directly, derived from Kd, or generated through an additional relationship. The key is not to reject remote sensing. It is to expose its chain of measurement, modelling and quality control so that comparisons with field profiles are defensible.
An example where light is nonexponential
Imagine normalized PAR of 1000 units at 0 m, 700 at 5 m, 420 at 10 m, and 100 at 15 m. The attenuation coefficient from 0–5 m is ln(1000/700)/5 ≈ 0.0713 m⁻¹. From 10–15 m it is ln(420/100)/5 ≈ 0.2870 m⁻¹. A single Kd for the entire 15 m span would be ln 10/15 ≈ 0.1535 m⁻¹, which hides the much sharper decline in the deeper interval. This might reflect a real optical layer, time-varying sky or measurement bias. Examine the full cast, deck normalization and complementary water properties before attributing the curvature. If a one-percent depth is needed, integrate a suitable layer-specific model or measure the crossing directly. A straight fitted line is convenient but not compulsory when the data reveal distinct regimes.
How to deal with surface saturation
Very bright sunlight can push a detector beyond its valid response range. A saturated surface PAR reading may appear capped at a maximum value, reducing the apparent contrast between shallow and deep observations and biasing fitted attenuation. Some sensors use adjustable gains or logarithmic responses, which make instrument-specific calibration essential. An analyst should inspect raw output status and exposure settings rather than assume the digital maximum is a real light level. If the top readings are invalid, fitting deeper data may still estimate a layer-specific Kd, but extrapolating to the surface becomes more uncertain. A report should state what portion of the profile was excluded, why and how that affected the one-percent threshold. Numerical algorithms cannot recover information that the sensor never recorded because of saturation.
Choosing a regression method
Ordinary least-squares regression of ln E on depth is common for estimating a representative Kd in a chosen interval. It assumes an error structure that may not hold when noise is additive in untransformed irradiance rather than logarithmic space. Near the noise floor, the log transform can magnify fluctuations. A weighted fit or nonlinear fit of E(z) may be preferable for certain instruments and data regimes. The correct choice should be guided by measurement uncertainty and residual checks, not by which method produces the most visually pleasing curve. State the depth interval, number of observations, handling of temporal illumination, and whether correlations introduced by smoothing were considered. A high R² does not prove the constant-Kd model applies below observed depths. Report both the fitted coefficient and evidence for its adequacy.
How to propagate uncertainty into Zeu
For a constant-Kd model, Zeu = 4.60517/Kd. When the uncertainty in Kd is small relative to its value, approximate standard uncertainty is u_Z ≈ (4.60517/Kd²) u_K. In an illustrative case Kd = 0.12 ± 0.01 m⁻¹, the central Zeu is 38.38 m and its linearized standard uncertainty is about 3.20 m. This excludes model inadequacy, surface-reference error and uncertainty in using a broadband approximation. Reporting 38.375 metres to three decimal places would imply an unjustified precision. A clear result might say about 38 ± 3 m under the fitted constant-attenuation model, specifying the uncertainty interpretation. If Kd is near zero or has a large relative uncertainty, this linear approximation becomes unreliable and more careful interval estimation is needed.
Uncertainty can be correlated across depths
A deck sensor used to normalize every underwater reading introduces a common reference uncertainty. Likewise, the same sensor calibration factor applies to all points in one cast. Errors are therefore not always independent across depth. In a ratio of two in-water readings, a common multiplicative calibration error may largely cancel; an offset error may not. This means one should not simply add every quoted percentage error as though each depth were a separate experiment. Assess which uncertainties change relative shape and which move the entire profile together. The Kd estimate depends strongly on relative depth variation, while the absolute flux and its daily integral depend on calibration scale. Explicit uncertainty sources help avoid both unjustified pessimism and overconfidence. A transparent report can show a measurement uncertainty budget and a separate model sensitivity analysis rather than collapse everything into one opaque error bar.
What a negative fitted Kd suggests
If downwelling PAR appears to increase with depth over an interval, the simple regression may produce a negative Kd. This does not prove that seawater has negative attenuation. Near-surface focusing, moving clouds, instrument tilt, shading, stratified illumination, clock mismatches or noise can produce such a pattern. Occasionally a localized light maximum is physically real because of surface lensing, but it should not be interpreted as uniform exponential gain. Check the raw time series and the simultaneous surface reference. If the interval is too short or noisy to estimate attenuation reliably, mark the result unresolved. Forcing the slope positive conceals the underlying diagnostic. A usable measurement method specifies conditions under which Kd is estimated and how intervals failing those conditions are treated. Physical intuition is a check, not permission to manipulate the data.
Comparing two stations on the same day
Suppose station A has Kd(PAR) = 0.08 ± 0.01 m⁻¹ and station B has 0.11 ± 0.01 under independent, compatible standard uncertainties. The difference is 0.03 and the standard uncertainty of the difference is approximately 0.0141 m⁻¹. That is about 2.12 combined standard uncertainties, suggesting a meaningful difference under the stated statistical assumptions, but not establishing the cause. Stations may differ in pigments, CDOM, particles, surface illumination or measurement geometry. If the estimates were calculated using shared calibration data, their covariance matters. A defensible comparison names the observational contrast and investigates causal explanations with other measurements. The phrase ‘water B was three-hundredths murkier’ is neither a unit-correct explanation nor an adequate ecological interpretation. Report the coefficient, its definition and what the differences imply about light loss over a specified distance.
Relating the new calculation to CDOM and backscatter
A PAR profile integrates the effects of multiple optical processes. One can examine CDOM absorption to understand one mechanism that removes light and particulate backscatter to understand one pathway that redirects light. Neither quantity is identical to Kd(PAR), which characterizes the resulting downwelling light field. Together they can support mechanistic analysis when measured consistently. The relationship should not be reduced to a naïve sum of aCDOM(440) and bbp(700) because the wavelengths, angular effects and definitions differ. An educational explanation should make the connection while preserving the distinct observables. The public reader gains a more useful picture by understanding how absorption, scattering and illumination interact than by memorizing three unrelated acronyms.
A practical error-repair conversation
A scientist hands an editor a chart titled ‘Light penetrated 100 metres because Kd was 0.05’. The editor checks the underlying records and finds that the reported coefficient was Kd(490), not broadband Kd(PAR), and that the statement about 100 m used Zeu = 4.605/K, which gives 92.1 m, not 100. The correct revision would identify the wavelength, computation and approximate depth, then qualify the substitution if it is being used as a proxy for euphotic depth. The correction is not a matter of pedantry; it changes both the measurement and the result. If the goal is biologically relevant light penetration, an underwater PAR profile or validated conversion may be more appropriate. This worked editorial decision illustrates how to turn a plausible but unsupported sentence into a traceable scientific explanation.
Field metadata that should survive translation
A reproducible profile might carry station, timestamp in UTC, latitude, longitude, water depth, pressure, depth_m, sensor model, serial number, PAR raw units, calibration date, deck irradiance, tilt, wavelength response, dark offset, processing steps, fit start and end depth, Kd(PAR), uncertainty, reference definition, Zeu method and quality flags. The purpose is not to demand an enormous table for every reader; it is to preserve the information needed to evaluate the number. A chart can simplify the presentation while its downloadable data carry full metadata. Unknown values should be marked unknown, not invented. Missing readings are not zero light. A quality code should identify why a segment was excluded, such as saturation, shading or sensor floor. Technical translation succeeds when shortening a report does not strip the evidence a future analyst needs.
Transfer problem: what if the sunlight changed?
At 5 m a sensor reads 400 µmol photons m⁻² s⁻¹ while the deck reads 800, and at 15 m the underwater sensor reads 150 while the deck reads 600. Without normalization, the ten-metre attenuation estimate is ln(400/150)/10 ≈ 0.0981 m⁻¹. After dividing by simultaneous deck readings, normalized values are 0.50 and 0.25, giving ln(0.50/0.25)/10 ≈ 0.0693 m⁻¹. The difference arises because available surface sunlight decreased during the cast. This example assumes deck normalization is valid for the sensor arrangements and conditions; such assumptions require checking in field practice. It demonstrates why a time-resolved reference can materially change the interpretation. The atmospheric light field is a moving input, not a hidden constant one can always ignore.
Transfer problem: light at a planned depth
Suppose a constant-Kd model has Kd = 0.15 m⁻¹ and corrected below-surface PAR E0 = 1200 µmol photons m⁻² s⁻¹. At 12 m the predicted signal is 1200 exp(−0.15×12) ≈ 198.4 µmol photons m⁻² s⁻¹. The fraction remaining is about 16.5%. The one-percent depth is approximately 30.7 m. These are two different answers: irradiance at one specified depth and a threshold depth. If water clarity changes at 20 m, the threshold calculation using one Kd may no longer be valid, even though the 12 m prediction remains within the modelled upper layer. Readers should be trained to state the depth interval of applicability whenever they use an exponential. A reported model prediction is not a substitute for a direct measurement at the target depth.
Transfer problem: layer boundaries
Consider a profile with Kd = 0.05 m⁻¹ in the upper ten metres and 0.20 m⁻¹ below. After the first ten metres, remaining PAR fraction is exp(−0.5) ≈ 0.6065. To reach 0.01, the lower layer needs attenuation ratio 0.01/0.6065 ≈ 0.01649, which takes ln(0.6065/0.01)/0.20 ≈ 20.53 m. Zeu is therefore about 30.53 m. Applying the upper-layer coefficient all the way down would predict 92.10 m and misrepresent the water column. A reader should see that layer-by-layer multiplication of transmission fractions is appropriate when the coefficient changes. The correct environmental explanation also asks why the change occurs, using independent profiles of particles, pigments, salinity or other properties rather than assuming a particular cause.
Transfer problem: a below-floor instrument
A profile reports 0.9, 0.5, 0.2 and −0.1 µmol photons m⁻² s⁻¹ at great depths after dark correction. These values are close to the sensor’s declared precision and one is negative. It would be incorrect to fit their natural logarithms or claim that negative light was observed. The analyst should determine the reliable quantification region and estimate Kd using valid positive readings above the floor, if enough data are available. A one-percent depth beyond that range is an extrapolation with uncertainty rather than a measured crossing. If the threshold cannot be resolved, report a bound or an unresolved result. An honest limitation is more valuable than a dramatic but physically impossible number generated by a spreadsheet that ignores measurement constraints.
Independent questions and answers
Question: Does Kd(PAR) have the same units as PAR? Answer: no; Kd is in m⁻¹, whereas a common PAR photon-flux unit is µmol photons m⁻² s⁻¹. Question: If Kd doubles under the same homogeneous model, what happens to one-percent depth? Answer: it halves. Question: Does Kd(490) equal Kd(PAR)? Answer: not generally. Question: Is Zeu necessarily measured? Answer: it may be estimated or extrapolated depending on the method. Question: Can a Secchi disk directly measure photon flux? Answer: no. Question: Does one percent mark the absolute biological limit of photosynthesis? Answer: no. Question: Why use simultaneous deck irradiance? Answer: to help remove atmospheric intensity changes during a profile. Question: What makes a deep negative PAR value? Answer: usually a correction or noise limitation requiring evaluation, not physically negative sunlight.
How to write a trustworthy result
A useful sentence might read: ‘A quality-controlled underwater PAR profile gave a fitted attenuation coefficient of approximately 0.12 m⁻¹ over 2–25 m; under a constant-coefficient extrapolation from a defined below-surface reference, the one-percent depth is about 38 m, with uncertainty and fit limitations described separately.’ This conveys measurement, calculation and assumption. A weaker sentence would say ‘the ocean stops producing oxygen at 38 metres because the satellite measured a Kd of 0.12’, which substitutes a biological cause for an optical convention and may confuse satellite and field measurements. Use accessible language, but do not replace qualified physical quantities with dramatic conclusions. The best summary tells the reader what was observed, which equation was used, and what independent evidence would be needed to extend the inference.
Concluding method-to-meaning workflow
Start with instrument units and spectral response; identify whether PAR is instantaneous or integrated, and whether irradiance refers to a deck, below-surface or modelled reference. Synchronize the depth and sunlight records, inspect dark offsets, saturation, tilt, platform shading and wave effects. Fit attenuation only over a defensible depth interval, inspect residuals and consider stratification. Derive a one-percent depth with the correct formula and mark it as observed, interpolated or extrapolated. Report confidence from instrument uncertainty and model assumptions separately. Compare satellite Kd products only after identifying wavelength and retrieval algorithm. If the intended conclusion concerns ecology, introduce independent biological evidence. Accurate translation here means converting a light trace into a transparent physical account, not turning a single exponent into an entire marine ecosystem story. Readers who follow this sequence can diagnose common calculation and interpretation errors before they propagate into decisions.
Connected reading routes
The associated seawater CDOM absorption guide explains one mechanism contributing to light loss; the ocean optical backscatter guide explains how particles redirect light. Neither should be silently renamed Kd(PAR). For broader language study, How English Works and the eduKate vocabulary library can help a reader practise distinctions between direct observations, derived estimates and hypotheses. These are contextual routes; methods for the PAR calculation still come from primary ocean optics protocols and carefully documented instrument procedures. Combining exact terminology with transparent arithmetic is the most reliable way to explain sunlight penetration across different water bodies.
Primary sources and methodological references
NASA PACE diffuse attenuation coefficient Kd overview
NASA Aqua MODIS 490 nm diffuse attenuation product
NOAA Light Attenuation Kd(PAR) dataset and definitions