Why radium means several different clocks
A marine radium quartet record contains four isotopes of one chemical element, yet their values cannot be read as though they were one interchangeable tracer. Radium-224, radium-223, radium-228 and radium-226 have very different radioactive half-lives, and their distributions reflect sources, transport, radioactive decay and other losses. Scientists use them to investigate exchange between continental margins and the ocean, water residence times and possible submarine groundwater discharge. Reading a figure responsibly means translating activities and isotope ratios back into a physical model whose inputs and limitations are visible.
Some water near a beach has more radium activity than water farther offshore. That alone does not prove a precise volume of fresh groundwater entering the sea. Radium can be released by reactions and exchange in saline sediment pore water, by recirculated seawater passing through coastal sediments, and through other boundary sources. Rivers and offshore sediment exchange also contribute in some settings. An anomaly is evidence of a process to investigate; it is not a unique source label. The key question is what sources, sinks and mixing mechanisms must be accounted for before dividing a radium flux by an endmember concentration.
The quartet is often a more useful teaching tool than one single tracer because it shows how time changes what a signal can mean. The approximate half-lives are 3.66 days for 224Ra, 11.4 days for 223Ra, 5.75 years for 228Ra and 1,600 years for 226Ra. Short-lived isotopes can lose substantial activity across days to weeks. The longer-lived members can preserve information over different transport scales. Those timescales do not magically identify the origin of a parcel. Their value comes from confronting a transport model with isotope-specific losses and known or constrained input ratios.
The first task is to read the data dictionary. Is the quantity reported in becquerels per cubic metre, millibecquerels per litre, atoms per unit mass, or another convention? Does the value mean activity at collection time or after a laboratory counting and decay correction? Is it dissolved radium, total radium, or an operationally defined filtered fraction? Did the sampled water pass through a manganese-coated fibre, and is sample volume recorded? A table lacking these details may be visually persuasive while remaining quantitatively unusable. A sound interpretation begins by resolving the quantities, not by choosing a favourite tracer equation.
Think of a coastal box as an accounting system. Radium enters from one or more water and sediment boundaries, may be supplied within the box, leaves through water exchange and sometimes removal processes, and disappears radioactively according to its decay constant. The abundance can increase even if discharge rate is constant, simply because the residence time or water exchange changes. Conversely discharge can increase without a proportional seawater activity increase when rapid flushing also increases. We must hold both sides of the budget in view. The observed activity is a state, not a direct flowmeter.
Activity, half-life and decay arithmetic
Radioactive activity is the number of nuclear decays per second. One becquerel means one decay per second, not one atom or one microgram. If concentration is expressed as Bq per cubic metre, multiplying by a water volume in cubic metres yields the activity inventory in Bq. A transport flux in Bq per day is a different quantity and requires a water-volume flow or another mass-balance term. Readers should mark the denominator and the time basis. Confusing activity concentration with flux is perhaps the most dangerous simple arithmetic mistake in radium budget summaries.
The radioactive decay constant is lambda = natural logarithm of two divided by half-life. If time and half-life both use days, lambda has units per day. For 224Ra, use 0.6931/3.66 days to get about 0.1894 per day. For 223Ra, use 0.6931/11.4 days to get about 0.0608 per day. These figures apply to radioactive disappearance and not to dilution or coastal removal. A field sample may lose activity through more than one route, so a rapid observed decrease cannot be translated directly into an age without examining mixing, source changes and water exchange.
Consider a hypothetical isolated water parcel containing 100 units of 224Ra activity at a chosen reference time. After 3.66 days with no inputs and only radioactive decay, 50 units remain. After 7.32 days, 25 units remain. The exponent form is A(t) = A(0) times exp(-lambda t). This example holds a fixed parcel in a closed setting. A real estuary is often an open, advective, diffusive system with spatially varying sources. Using the same exponential fit on an offshore transect without a transport model silently assumes away important physical processes.
The same elapsed time affects isotopes differently. After approximately eleven days, 224Ra has experienced about three half-lives and only around one-eighth of its original activity remains in a closed parcel. Over eleven days 223Ra experiences about one half-life, so roughly half remains. The difference may provide information about time since release if the original ratio and subsequent additions are constrained. But in a real water body the initial ratio may vary between pore waters and sediment environments. A ratio cannot function as a stopwatch until those boundary uncertainties are addressed.
For 228Ra and 226Ra, a few days produce very little radioactive decay relative to their long half-lives. That does not make them inert or spatially uniform; their sources, sediment exchange, particle interactions and water circulation still shape the observations. When interpreting a joint isotope figure, avoid calling high 226Ra old water or high 224Ra young groundwater as universal rules. Activity depends on supply and residence time together. The model must specify why a particular source carries a certain activity and how it might have changed before reaching the measurement point.
How samples become activities
Sampling can begin by collecting a known water volume and extracting dissolved radium onto manganese-oxide impregnated fibre under a defined protocol. Short-lived isotope measurements are often obtained by radium delayed coincidence counting, known as RaDeCC. The field and laboratory record should include volume, flow through the extraction medium, sampling date, separation or extraction time, counting time, blank, background, counting efficiency and any procedural recovery checks. A later table value is the end of that chain, not its starting point. Every missing link expands uncertainty or makes comparisons ambiguous.
Count data are not activity data until converted with the relevant calibration and background corrections. Imagine a hypothetical counting channel registering 500 event counts over a stated duration, with 80 expected background counts in that interval. The background-subtracted total is 420 counts. An instrument-specific efficiency and the volume actually processed are then needed to estimate activity concentration. Dividing 420 by sample litres immediately, without accounting for count duration, detection efficiency and timing, would not produce a defensible Bq/L. The dimensions reveal that something is missing.
Decay correction must refer to an explicit time. A bottle sampled Monday but counted Thursday has experienced several days of 224Ra decay; its activity at counting time differs from activity at collection even without laboratory losses. If an instrument or laboratory reports collection-time-corrected activity, applying the same decay correction again creates a systematic overestimate. A metadata field indicating reference time is therefore as important as a unit label. For very short-lived species, ambiguous timestamps can dominate the interpretation of a coast-to-offshore gradient.
Uncertainty depends on counts, backgrounds, efficiency, blanks, sample volumes, extraction recovery and decay correction. At low counts, simple symmetric normal approximations may be inadequate, and detection limits need to be handled as part of the stated method. A below-detection observation is not physical zero. Do not insert zero into a regression or divide by an unquantified isotope activity when computing a ratio. It is often better to report an upper limit or use a statistical method designed for censored data rather than forcing an arbitrary tiny numerical substitute.
The USGS has published coastal field programmes measuring 223Ra and 224Ra from manganese-oxide fibre samples with RaDeCC instruments, and the GEOTRACES programme documents the broader quartet in ocean boundary studies. These sources show that high-quality work requires explicit sampling, extraction, timing and interpretation metadata. They do not imply that one method’s efficiency coefficients can be copied across detectors. Instrument calibrations and blank treatment belong to the measurement configuration that produced the data.
Set up a transparent coastal isotope budget
Start an analytical budget by drawing the boundaries of the water body. Define its surface area, mean depth and volume, the exchange pathways, and the radium activity units. Specify whether the calculation represents a short survey snapshot, a tidal mean or a longer steady-state estimate. Even the apparently simple word box is an assumption about mixing: if nearshore groundwater enters a poorly mixed bottom layer, one homogeneous volume may obscure gradients that are crucial to interpretation. Draw the physically meaningful compartments before choosing an algebraic form.
For an instructive one-box model, let the excess radium inventory be I in becquerels, a source flux F in becquerels per day, radioactive loss lambda I in becquerels per day, and net advective/export loss E in becquerels per day. If the inventory is steady and other sources or sinks are absent or separately quantified, F = lambda I + E. This identity is the skeleton of an interpretation, not a full field method. Additional sediment desorption, river inflow, diffusion and exchange pathways may need their own terms. State which ones have been evaluated or omitted.
Suppose a hypothetical lagoon contains two million cubic metres of water, with measured 224Ra excess above a properly estimated background of 0.5 Bq per cubic metre. The excess inventory is 1,000,000 Bq. With lambda 0.1894 per day, radioactive decay removes about 189,400 Bq/day under a homogeneous model. A calculation that divides 0.5 Bq/m³ by 0.1894/day produces a quantity in Bq·day/m³ and cannot represent groundwater discharge. The volume is essential for inventory; the time dimension enters through the loss rate. This units test is a first diagnostic.
Continue the hypothetical lagoon example. If all excess 224Ra came from a measured groundwater endmember with activity 40 Bq/m³ and if export and other sources were genuinely negligible, the decay-replacement water volume would be 189,400 Bq/day divided by 40 Bq/m³, or about 4,735 m³/day. The result is illustrative and deliberately incomplete. Real nearshore mixing and radium recycling can dominate; interpreting this 4,735 m³/day as a field estimate would be invalid unless the excluded terms were supported as negligible. A good worked example includes the statement of what it does not prove.
If the same model required an additional 100,000 Bq/day exported through coastal exchange, its total inferred source would become 289,400 Bq/day. Using the same 40 Bq/m³ endmember gives about 7,235 m³/day rather than 4,735. The enormous sensitivity to an omitted exchange term is the lesson. Two researchers who use the same isotope measurements but different boundary flux assumptions can reach different discharge estimates without arithmetic mistakes. Compare their assumptions and residence-time evidence first. Do not call one result more exact simply because it is expressed with more significant digits.
Endmember uncertainty can be equally large. Holding the hypothetical 289,400 Bq/day source fixed, an endmember activity of 25 Bq/m³ gives about 11,576 m³/day, while 50 Bq/m³ gives about 5,788 m³/day. The water flux changes inversely with the selected activity. Selecting only the highest-radium well produces a low flow estimate, not necessarily the representative coastal flux. Sample multiple pore-water environments, tidal states, salinity mixing zones and feasible contributing pathways. A groundwater endmember is an empirical distribution to characterise, not a convenient single number to choose after viewing the desired answer.
What isotope ratios can and cannot date
Ratios help remove a shared dilution factor only under specific conditions. If two isotopes were supplied together with a known starting activity ratio, had no additional inputs, and experienced the same non-radioactive dilution, their ratio changes because their decay constants differ. For 224Ra divided by 223Ra, R(t) = R(0) exp[-(lambda224 – lambda223)t]. Rearranging gives time = ln[R(0)/R(t)] divided by the difference in decay constants. This is an apparent elapsed time under a model, not the birthday of a groundwater molecule or an unconditional seawater age.
Use a deliberately hypothetical initial 224Ra/223Ra activity ratio of 4 and an observed ratio of 2. The decay constants are about 0.1894/day and 0.0608/day, so their difference is 0.1286/day. The log ratio is ln(4/2) = 0.6931. Dividing gives approximately 5.39 days. The arithmetic is valid for a closed parcel with the stated initial ratio, common physical losses and no new supply. A spatial transect near a beach may violate all three assumptions. Never report 5.39 days as a field-validated residence time merely because the logarithm was evaluated correctly.
Now imagine the starting ratio was not known to be 4; plausible source waters range from 3 to 5. With the same observed ratio of 2, the illustrative apparent times range from ln(3/2)/0.1286, about 3.15 days, to ln(5/2)/0.1286, about 7.13 days. This spread arises from the source assumption before considering analytical counting uncertainty or mixing. A single numerical age can hide an uncertain reference value. Present a range or probability distribution conditioned on the measured endmember evidence, and distinguish it from the uncertainty of a detector’s counting statistics.
Short-lived 224Ra may be continuously released from benthic sediments as water moves offshore. Such a source can maintain activity and change the isotope ratio, making a simple decay clock appear younger than it would in a source-free parcel. Conversely selective losses or mixing with low-activity offshore water can alter a ratio in ways that depend on both isotopes’ boundary patterns. The correct inference may require a transport model, radium sources and hydrodynamic observations. The ratio is most informative when competing processes have been constrained independently.
An age estimate is sometimes best described as an effective transport or apparent mixing time. That phrasing is scientifically useful because it avoids claiming that every water molecule has the same path length or travel time. In a dispersive coastal water body, parcels may enter at different locations and have a distribution of residence times. Ratios integrate that mixture through nonlinear functions. An average isotope ratio does not necessarily equal the ratio expected from the mean age. Simulations or sensitivity analyses can show whether a one-age approximation materially biases a conclusion.
Different isotope combinations have different windows of sensitivity. Short-lived pairs are informative at coastal timescales, while 228Ra gradients can trace broader shelf exchange and longer transport. The exceptionally long half-life of 226Ra makes radioactive loss small over ordinary coastal sampling periods. That does not make 226Ra a correction factor that can be subtracted without justification. Each measured isotope has distinct source and removal dynamics. When an interpretation uses a pair, report why that pair was chosen and which processes are expected to cancel or remain in the ratio.
Spatial sampling and hydrodynamics
A coast-to-offshore decline in radium activity might reflect dilution, radioactive loss, particle interactions, changing endmember input or the geometry of circulation. Plot activity against actual position, water depth, salinity and sampling time rather than distance alone. A transect collected over six hours can cross different tidal stages. If the stations are arranged on a map as though observed simultaneously, a moving feature may look like a stationary spatial gradient. Consider the survey design before fitting an exponential line to distance. The scale of spatial separation matters relative to movement during sampling.
Nearshore radium is often sensitive to coastal sediment exchange. A sandy beach can draw seawater into its sediments during one part of the tide and return modified water later, enriching it in radium without providing an equivalent volume of new terrestrial freshwater. The term submarine groundwater discharge includes both fresh terrestrial groundwater and recirculated saline pore water in many studies. If the management question is freshwater supply or nutrient delivery, separating those components matters. Salinity and independent tracers can help but may not fully identify each pathway in a complex mixing zone.
A finite-volume balance becomes better grounded when paired with water exchange measurements. Current meters, residence-time estimates, salinity budgets and tidal exchange observations help constrain the export term. However, using a poorly validated hydrodynamic model can simply move uncertainty from the radium calculation into an invisible parameter. Report how exchange was measured or estimated, whether boundary flows are net or gross, and whether the averaging interval matches the isotope sampling. A daily rate inferred from one low-tide survey should not quietly be converted into a yearly mean.
Imagine an ocean margin where radium activity is high around a submarine spring but low at the beach. A simplistic shore-distance model could infer offshore transport of beach water, yet the actual source may lie offshore or in a buried permeable layer. Sample across the likely source locations and use sediment, hydrogeological and bathymetric context. The map should allow the source geometry to challenge the theory. Never automatically interpret every high-activity location as adjacent to the terrestrial coastline. Tracer distributions describe where water has been influenced by sources, not necessarily where it originally entered.
Repeated sampling across spring and neap tides can reveal that apparent endmember activity varies with circulation through sediments. A spring-tide increase in radium at the coast could reflect more pore-water flushing, stronger mixing or altered water residence time, and several processes can operate together. When comparing surveys, normalise to actual quantities with consistent methods and preserve the time of each sample. A high radium activity measured after stronger tides is not sufficient evidence for a higher freshwater recharge rate. Physical context prevents a chemical anomaly from being read too literally.
Do not confuse isotope flux with nutrient flux
Radium-derived water discharge is sometimes combined with a groundwater nutrient concentration to estimate nutrient loading. The multiplication is straightforward only if the concentrations represent the same water source and the inferred discharge includes the same physical component. If 500 m³/day of groundwater carries nitrate at 2 mg/L, the corresponding nitrate mass would be 500,000 L/day times 2 mg/L, or 1,000,000 mg/day, equal to 1 kg/day. This is a hypothetical arithmetic exercise, not an observed coastal nutrient budget. Mixing, reaction and spatial heterogeneity can change actual delivery.
Suppose the radium signal is primarily from salty recirculated seawater that passes through nutrient-rich sediments. If a researcher multiplies the total recirculating discharge by a freshwater well nitrate concentration, the calculation pairs mismatched pathways. A sound source model may require separate fresh and saline exchange volumes, each with its own nutrient endmember and reaction history. The useful reporting question is: which water entered the radium budget, and which water supplied the chemical concentration? Only a matched pair creates a meaningful product.
Nutrients can be transformed within sediments and the coastal water column. Nitrate may be consumed through denitrification or produced through nitrification; phosphate and metals may be released or retained as redox conditions change. Radium is a tracer of physical or geochemical exchange, not a chemically conservative proxy for every nutrient. A stable radium signal does not imply a constant nutrient concentration. To make a causal nutrient claim, measure the nutrient independently, document sample timing and consider reactions during the residence period between source and seawater collection.
Units must survive every conversion. A concentration in micromoles per litre times discharge in cubic metres per day yields micromoles times one thousand litres per cubic metre per day; it is not directly kilograms per year. Convert to moles, multiply by molar mass and scale the time interval only after defining whether the concentration represents the species or an elemental basis. For nitrate reported as nitrogen, the mass conversion uses nitrogen’s molar mass; for nitrate ion mass, it uses the nitrate ion’s molar mass. The symbol and reporting convention matter as much as the number.
A common communication error calls a radionuclide an added pollutant simply because it is radioactive. The radium quartet discussed here is naturally occurring and is being used as a tracer of ocean exchange; interpreting the measurements as a human contamination emergency would require a different risk assessment and context. Likewise a high tracer activity does not directly establish unsafe drinking water at a beach or harmful marine food. Separate environmental tracer application from a regulatory exposure conclusion, and avoid unnecessary alarm while remaining precise about measured activity.
Uncertainty, blanks and analytical checks
Background subtraction is not interchangeable with declaring offshore activity zero. If the sampled coastal water contains a regionally distributed 228Ra background, subtracting a location-specific baseline may be required for an excess inventory. Yet an arbitrary offshore station might itself receive boundary inputs. Define the background region using hydrography and tracer structure and test how the result changes under alternative plausible baselines. When reported excess values are negative, do not automatically replace them with zero; examine whether they arise from uncertainty or an invalid assumed baseline.
Short-lived isotopes pose an especially severe timing problem. For 224Ra at a half-life near 3.66 days, one day represents a measurable decay fraction. If samples are counted at different delays after collection, their raw count totals cannot be compared before proper decay adjustment. Trace sample collection, Mn-fibre preparation, first count, possible second count and reported reference time. Two analysts can reproduce a calibration only if these timestamps and detector factors are available. An exact-looking activity with unknown timing is less useful than a flagged interval with transparent limitations.
Counting uncertainty is not identical to environmental variability. A detector might return a precise estimate for a single bottle that does not represent a tidally varying beach. The field programme’s spatial and temporal sampling design may dominate the uncertainty of a calculated source flux. Report both levels. A tight error bar from counting statistics should not be used as the uncertainty of an annual discharge estimate unless tide, season, endmembers and export have been assessed. Sampling more often can improve representativeness, but only if the new samples cover physically relevant conditions.
Consider a hypothetical isotope budget where F is 300,000 Bq/day with approximately 20 percent standard uncertainty and the selected endmember C is 40 Bq/m³ with 25 percent uncertainty. The ratio Q = F/C is 7,500 m³/day. Under a simplified independent-error linear propagation, its relative uncertainty is sqrt(0.20² + 0.25²), about 32 percent, or approximately 2,400 m³/day. This is not a rigorous uncertainty interval if F and C are correlated, asymmetric or poorly characterised. It reveals why reporting Q = 7,500.0 m³/day conveys unjustified precision.
Sensitivity is often more informative than a single error bar. Vary the offshore background, water volume, exchange time, sediment source term and pore-water endmember within plausible independently observed ranges. Show how each changes the derived discharge. Some results may be robust in direction but fragile in magnitude. That distinction is useful to coastal planning: the method might support the conclusion that sediment exchange matters while not supporting a narrow discharge value. Be explicit which conclusion survives the sensitivity tests instead of choosing one apparently optimal parameter combination.
Do not confuse absence of measured 224Ra with proof of absent groundwater. The isotope can decay, dilute below detection or fail to be recovered or counted well. A long offshore transport time may attenuate the short-lived signal while a longer-lived isotope remains. Source endmember activity may also be low in a particular hydrogeochemical setting. Report detection thresholds and the survey’s age window. A non-detection limits some models under stated assumptions; it does not categorically eliminate every exchange pathway. This is especially important when interpreting sparse offshore stations.
Worked transfer problems
Problem A: a sample has 224Ra activity 16 Bq/m³ at one time and, in a closed no-input parcel, is remeasured after one 3.66-day half-life. What remains? Half remains, so 8 Bq/m³. If the value is 10 instead, the excess could reflect additional source, different mixing, counting uncertainty or other processes. The measured difference alone cannot distinguish them. The key is to state the simplified parcel assumptions before accepting the halving result as a prediction, and not to treat that exercise as a natural coastal budget.
Problem B: a lagoon of 1.5 million m³ shows an average excess activity of 0.4 Bq/m³. What is the inventory? Multiply to get 600,000 Bq. What is the radioactive 224Ra loss using lambda 0.1894/day? Approximately 113,640 Bq/day. This is not a groundwater flux because export and other sources are unknown, and no representative groundwater endmember has yet been supplied. Identifying the missing information is part of the answer. A unit-consistent intermediate quantity is valuable even when it is not the final desired result.
Problem C: an idealised 224Ra/223Ra source ratio is 3.6, the observed ratio is 1.8, and the decay-constant difference is 0.1286/day. The apparent elapsed time in the closed no-input model is ln(2)/0.1286, approximately 5.39 days. A river or benthic source adding 224Ra during transport would invalidate this clean result. Explain why observing a ratio is not enough: you must establish the starting ratio, relative dilution behaviour and ongoing supply. When those are uncertain, a conditional apparent age or range is more accurate than an exact birthday.
Problem D: an activity flux of 240,000 Bq/day is attributed entirely to an endmember with 30 Bq/m³. Divide to obtain 8,000 m³/day. If the endmember could actually range from 20 to 40 Bq/m³, the same flux implies 12,000 to 6,000 m³/day respectively. If 60,000 Bq/day of that flux belongs to an independently measured sediment source, subtract it first and then redo the water-flux calculation. The correct sequence is source partition, matching endmember, and division. Conflating all radium supply with one groundwater class exaggerates interpretive certainty.
Problem E: a paper reports discharge in cubic metres per day and nitrate as 50 micromoles per litre. Can you multiply and report tonnes of nitrogen per year immediately? No. Identify whether the nitrate value is nitrate ion or nitrogen-equivalent, convert cubic metres to litres, micromoles to moles, multiply by the matching molar mass, and handle the time unit. Also check whether the nutrient source matches the radium-derived water flux and whether nitrate is transformed between sediment and sea. A perfect conversion cannot rescue a mismatched physical model.
Problem F: a station has radium below a counting detection threshold after a long delay. Does this mean the station had zero radium activity at sampling? No. Consider detection limits, radioactive decay between sampling and counting, extraction recovery, blank and detector efficiency. The correct translation is a detection-qualified result at a stated reference time. If sample handling makes the upper limit unreliable, flag that too. Never silently insert zero into a source ratio or logarithmic age calculation. A careful non-detection can still constrain a transport model when its measurement context is retained.
Frequently asked questions
Can radium trace only fresh groundwater? No. Radium-rich saline pore water and recirculated seawater can contribute substantially, along with other sources. A study seeking fresh groundwater must establish a way to partition its sources using salinity, independent tracers, hydrogeology or a suitable joint model. The presence of radium is not a direct salinometer. The word discharge should be accompanied by a definition of which water pathways are included.
Why measure four isotopes if one already shows a gradient? Different half-lives and source behaviour supply partly independent constraints on transport and coastal exchange. They also expose contradictions in an oversimplified model. A model that explains 224Ra but fails to explain 228Ra may have incorrect timing, boundary sources or mixing assumptions. The quartet is useful because the comparison challenges the interpretation, not because four numbers automatically guarantee a unique answer.
Does a high 224Ra/223Ra ratio mean young seawater? It may be consistent with relatively recent source influence under suitable starting ratios and restricted post-release processes. Without knowing the source ratio and additional inputs, the interpretation is ambiguous. Report it as an isotope ratio first, then present any age model with assumptions and sensitivity. The scientific value lies in a testable inference, not a shorthand label.
Is a becquerel a mass unit? No. One Bq is one radioactive decay per second. Activity concentrations such as Bq/m³ describe decays per second per water volume. Converting between activity and atom count requires the isotope’s decay constant and conversion units. A radium activity flux in Bq/day is not directly a groundwater volume flux without a matching activity concentration in Bq/m³.
Can a single coastal sample produce an annual SGD rate? It can contribute to an estimate, but seasonal, tidal and hydrodynamic variability generally must be evaluated. Annualising an instant or one tidal phase requires assumptions that should be explicit. Repeated sampling across relevant physical conditions can reveal whether a steady-state approximation is reasonable. The answer depends on the study design, not on the convenience of dividing by a year.
What is the most useful first check when two studies disagree? Compare their system boundaries, isotope reference times, background activities, endmembers, exchange models and definition of SGD. Only after verifying those should one compare numerical rates. Both studies may be internally consistent while estimating different portions of an exchange system. Technical translation means preserving that distinction and explaining it in language a non-specialist can follow.
Primary sources and related eduKateSG reading
Methods: GEOTRACES radium factsheet; USGS 223Ra/224Ra field data and RaDeCC method; USGS coastal groundwater tracer study; BCO-DMO global 226Ra/228Ra data; US EPA radionuclide half-lives. Worked numerical cases are hypothetical and used solely to teach methods.
Related eduKateSG science: Groundwater CFC/SF6 apparent ages; Radon-222 groundwater mass balance; Ocean pH sensor quality control. Each tracer needs its own appropriate physical and chemical model.