Soil low-field nuclear magnetic resonance, usually shortened to low-field NMR, can measure signals from hydrogen nuclei in water and examine how quickly those signals relax. If you need to understand soil NMR T2 distributions, water-signal peaks, CPMG echo trains, pore-size estimates or surface relaxivity, the most useful first question is what water was present in the sample and how the relaxation signal was measured. A short T2 value is not automatically the diameter of a small soil pore.
In many soils, proton relaxation depends on water interacting with mineral and organic surfaces, as well as bulk-fluid properties and magnetic field gradients. Researchers can represent a measured signal as a distribution of relaxation times and, with a suitable model, estimate how detectable water is distributed among environments. The result depends on water saturation, echo timing, inverse modelling and surface relaxivity, so a T2 graph should not be translated directly into an absolute pore-size chart without its assumptions.
This guide shows how to read low-field 1H NMR results for soil: sample water state, T1 and T2, CPMG measurement, exponential decay, inverse Laplace reconstruction, regularisation, signal amplitude, paramagnetic minerals, and the conditional relationship between T2 and pore geometry. Original numerical examples use declared assumptions and units, not invented laboratory observations. The aim is to make a specialist relaxometry paper understandable without confusing instrument signal with an independently measured hydraulic property.
Quick answer: what is a soil NMR T2 distribution?
A soil NMR T2 distribution summarises how much measured proton signal is associated with different transverse relaxation times after an applied pulse sequence. In a water-bearing sample, much of the proton signal may come from water. Under particular assumptions, faster relaxation can be associated with water exposed to a larger pore-surface area relative to pore volume, while longer T2 components can reflect water in larger or less surface-dominated environments.
The inference is conditional. Dissolved ions, iron-containing minerals, organic surfaces, echo spacing and saturation can affect relaxation. A distribution of fitted T2 values is a model-based description of signal decay, not a photograph of individual pores. The first check is whether the signal amplitude represents all the sample’s water or only the water detectable within the instrument’s acquisition window.
Hydrogen nuclei are the signal source, not the pore walls
Proton NMR detects magnetic responses associated with hydrogen nuclei under an applied magnetic field and suitable radiofrequency pulses. In soil samples, water is often the primary proton-bearing substance of interest, but organic matter and other hydrogen-containing components may contribute depending on the method. The instrument does not directly scan the shape of a void in the same way X-ray microtomography records reconstructed attenuation.
A small pore can influence relaxation through the water it holds and the surfaces that water encounters. The route from measured signal to pore geometry passes through assumptions about relaxation physics. A strong explanation begins with the molecule being measured and identifies the later inference separately.
Low-field describes an experimental regime
The term low-field indicates a magnetic field strength lower than those commonly used for high-resolution analytical NMR spectroscopy, though exact boundaries depend on application. Low-field instruments are useful for water studies because proton relaxation can be examined in laboratory cores or other suitable samples. Field strength affects sensitivity and how some relaxation mechanisms contribute.
Do not compare two T2 plots merely because both are called NMR. Confirm magnet field, frequency, sample dimensions, coil geometry, measurement temperature and pulse sequence. These details determine which water components are observable and whether the results are comparable. A universal conversion from T2 milliseconds to pore diameter is not supplied by the word low-field.
T1 and T2 describe different forms of relaxation
T1 is the longitudinal or spin-lattice relaxation time; T2 is the transverse or spin-spin relaxation time. They describe distinct components of how magnetisation evolves toward equilibrium after excitation. Soil studies may report either or both, and the underlying experiment must be identified before a figure is interpreted.
A T1 distribution should not be relabelled T2 simply because both use milliseconds. Their sensitivity to pore surfaces, molecular motion and field conditions can differ. Research on fast field-cycling NMR of natural soils examined T1 distributions, while many water-content studies emphasise T2. The measurement type belongs in the first sentence of any technical summary.
Why a CPMG echo train appears in soil studies
The Carr–Purcell–Meiboom–Gill sequence, often abbreviated CPMG, uses an initial excitation and repeated refocusing pulses to produce echoes whose amplitudes decay over time. That echo train can be fitted to estimate transverse relaxation components. The sequence helps manage some inhomogeneity effects but does not erase every magnetic-gradient or diffusion-related contribution.
Read the echo spacing and number of recorded echoes. If very rapid relaxation occurs before the first measurable echo, some water-associated signal may be absent. If the sequence ends before long relaxation components decay sufficiently, estimates of those components may be uncertain. Acquisition design shapes the data long before a fitting algorithm generates the familiar distribution curve.
Exponential decay is the basic mathematical ingredient
For an ideal single T2 component, measured signal amplitude can be expressed as M(t) = M0 × exp(−t/T2), where M0 is the extrapolated initial amplitude and t is elapsed time. At t = T2, the ideal component retains approximately 36.8% of its initial signal. This equation represents a simplified relaxation behaviour, not a pore-size equation.
If M0 were 1,000 arbitrary instrument units and T2 were 100 milliseconds, the ideal signal at 100 milliseconds would be about 368 units. At 200 milliseconds it would be about 135 units. The arithmetic shows how a relaxation time controls decay. It does not determine a pore radius or water volume until further measurement and calibration conditions are specified.
Soil water rarely has one relaxation environment
Natural soils contain mineral surfaces, organic matter, aggregates and voids with variable size, chemistry and water connectivity. Water in different environments can contribute different relaxation components. A measured echo train can therefore resemble a sum or continuum of exponential decays rather than one perfect exponential with one T2.
The analyst may describe a distribution of relaxation times rather than assigning a single mean to all water. However, distributions recovered from noisy finite echo data are not unique without mathematical constraints. A smooth curve should not be mistaken for direct detection of thousands of independent pore classes. It is a useful representation constructed from the signal and a defined inversion procedure.
What inverse Laplace transformation does
A multiexponential decay can be modelled as an integral or discrete sum of exponentials weighted by component amplitudes. Recovering the weights from recorded echoes is an inverse problem often described as inverse Laplace transformation. The problem is sensitive to noise and finite measurement duration: different detailed distributions may fit a signal almost equally well.
This explains why two software packages can return different peak widths or small shoulders from the same echo train. Their choices of T2 grid, noise treatment, positivity constraint and regularisation may differ. An appropriate report includes the inversion approach and checks that identified features are robust rather than artefacts of overfitting.
Regularisation is a modelling choice, not a biological measurement
Regularisation discourages extremely unstable or jagged solutions in an ill-posed inverse problem. Strong regularisation may merge neighbouring components into one broad peak, while weak regularisation may allow noisy spikes. Neither appearance necessarily means soil water itself has changed. The shape of the reconstructed T2 distribution depends in part on how the inversion was constrained.
When a paper compares peak counts across treatments, ask whether fitting settings and signal-to-noise ratios were comparable. An apparent new peak that disappears under a small plausible change in regularisation should not be confidently labelled a new pore class. Sensitivity analysis makes the result more trustworthy than selecting the most visually impressive curve.
Echo spacing creates an early-time blind spot
Very short T2 components can decay before the acquisition sequence samples them reliably. An instrument with relatively long echo spacing may therefore underestimate water in environments with rapid transverse relaxation. This is particularly relevant in soils containing paramagnetic phases, strong local magnetic gradients or water closely associated with mineral surfaces.
If an author calls the integrated distribution area total water content, check whether the method captures rapidly relaxing protons. A good comparison may use independent gravimetric water measurement to assess the missing fraction. A small area under the curve could mean less water, but it could also mean some water became harder to detect under the acquisition conditions.
Temperature and fluid chemistry affect relaxation
Molecular mobility changes with temperature, and dissolved substances can alter relaxation properties. This is important when comparing soils with different salt concentrations or measurements made at different temperatures. A shift toward shorter T2 does not necessarily indicate pore contraction; changes in fluid properties or surface interactions may contribute.
Look for environmental control during measurement, especially in time-series experiments involving wetting, drying, freezing or changing solute concentrations. If temperature changed, a researcher should identify its potential effects before assigning all variation to geometric restructuring. The core lesson is that relaxation responds to a chemical and physical environment, not just an empty space.
Paramagnetic iron and manganese can shorten T2
Soil minerals containing iron or manganese in suitable chemical forms can influence proton relaxation through local magnetic interactions. Paramagnetic surfaces can accelerate relaxation and change the fitted distribution even if two soils have similar pore geometry. Thus mineral composition can confound simple attempts to infer pore size from T2 alone.
A comparative study of NMR relaxometry and hydraulic measurements discusses the influence of paramagnetic constituents and internal gradients in natural soils. This evidence does not mean every iron-rich soil yields useless NMR data. It means pore-size inference must account for relaxation mechanisms and calibration rather than assuming the same conversion across different mineral assemblages.
Surface-to-volume ratio explains one common relation
In a fast-diffusion and surface-dominated regime, a simplified expression is 1/T2 ≈ ρ2 × S/V, where ρ2 is transverse surface relaxivity and S/V is pore surface area per pore water volume. Surface relaxivity has dimensions of distance per time, while S/V has inverse-distance units. Their product therefore has inverse-time units, consistent with 1/T2.
This relation can be useful for interpreting how confined water samples pore walls. It assumes relaxation is sufficiently averaged by diffusion and that other contributions, such as bulk-fluid relaxation and diffusion in gradients, are minor or separately handled. The symbol ≈ matters; natural soils do not automatically satisfy every simplifying condition.
Pore geometry changes the meaning of S/V
For an ideal long cylinder of radius r, ignoring its ends, the surface-to-volume ratio is approximately 2/r. For a sphere of radius r it is 3/r. Natural pores are often neither cylinders nor spheres, and their surfaces may be rough. The same measured S/V can correspond to different shapes and connection patterns.
A researcher using an idealised pore geometry is making a modelling choice. A cylinder-based conversion from T2 to radius does not give the same number as a sphere-based conversion under identical relaxivity. Reports should name the geometry and avoid presenting the result as a direct microscopic measurement of every irregular pore in the soil.
Worked example: T2 to a model-based radius
Assume a long cylindrical pore, fast diffusion, surface-dominated transverse relaxation and constant ρ2 = 10 micrometres per second. If T2 equals 0.10 seconds, the simplified relation r ≈ 2ρ2T2 gives r ≈ 2 × 10 × 0.10 = 2 micrometres. The corresponding diameter would be approximately 4 micrometres.
This is an illustrative conditional estimate, not a universal rule converting 100 milliseconds into four micrometres. The value changes if relaxivity differs, if bulk or magnetic-gradient terms matter, or if the geometry is not cylindrical. Every step should retain both the numerical unit and the model assumptions, because omitting either turns an estimation into an invented direct measurement.
A tenfold relaxivity change means a tenfold radius change
Keep T2 = 100 milliseconds and the same cylinder model, but increase assumed ρ2 from 10 to 100 micrometres per second. The inferred radius rises from 2 to 20 micrometres. The measured relaxation time did not change at all; only the conversion assumption did. This is why surface relaxivity can dominate pore-size uncertainty.
Research on quantitative pore-size determination using soil 1H NMR examines calibration of surface relaxivity and comparison with water-retention-derived pore distributions. That work illustrates why researchers do not simply print a universal T2-to-radius lookup table for all soils. A good article reports uncertainty in ρ2 and avoids claiming nanometre-level pore precision from a poorly constrained conversion.
Water-filled pores are the principal observable
A saturated soil sample can contain water in a broad range of pores, whereas an unsaturated sample contains water only in portions of its pore network. As soil drains, larger pores may empty while smaller spaces remain wet, changing the detectable proton populations and their relaxation times. The T2 distribution may therefore change even if the solid framework stays fixed.
A reader comparing scans at different matric potentials should ask whether the goal is pore geometry, water redistribution or both. Treating a drying-induced shift as proof that soil pores physically shrank would be unwarranted without additional structural evidence. For water dynamics, NMR offers real strengths precisely because it responds to water occupancy rather than only a static geometry.
A T2 peak is not a distinct family of open channels
A strong peak near 10 milliseconds may be consistent with abundant detected water in environments where relaxation is relatively rapid. It is not necessarily evidence for one narrow pore-radius class or one population of connected channels. Several different water environments can overlap in relaxation time because surface chemistry and pore geometry vary.
If a study assigns a peak to ‘bound water’, examine how bound was defined and calibrated. A dividing time borrowed from petroleum reservoir studies may not transfer reliably to a clay-rich soil with different relaxation properties. A defensible description states the measured component and method-specific interpretation instead of exporting an unverified industry threshold.
Signal amplitude is not always an absolute water volume
Within a well-calibrated system and under suitable acquisition conditions, detected proton signal can relate to water quantity. But coil sensitivity, sample positioning, instrument gain, incomplete relaxation recovery and missing short-T2 components can change measured amplitude. Normalised T2 distributions may show relative proportions even when they do not preserve total signal.
When a graph’s area is set to 100%, the data cannot by themselves tell whether one soil contained twice as much water as another. Relative peak percentages belong to the detected signal denominator of each sample. If absolute water content matters, look for an independent calibration or paired mass measurement, and do not convert normalised areas directly into millilitres of water.
Worked example: relative component areas
Suppose a hypothetical normalised T2 distribution contains 35% of detectable signal in a short-relaxation component and 65% in a longer-relaxation component. These are proportions of the detected proton signal, not 35% and 65% of the soil’s physical pore volume. If a second sample produces much less total signal but is normalised the same way, the plots may look similar despite differing water quantities.
The correct sentence is ‘The short-T2 component represented 35% of the modelled detectable signal under the chosen acquisition and inversion.’ It is not ‘Thirty-five percent of all soil pores are microscopic.’ The latter adds claims about total porosity and pore dimensions that the measurement did not directly establish.
Water retention offers an independent calibration route
A soil water retention curve relates water content to matric potential and can inform which pore regions drain over a pressure range under model assumptions. Comparing relaxation distributions at selected moisture states with retention measurements can help estimate surface relaxivity and assess how T2 components relate to water-filled pore sizes. This is a cross-method calibration, not proof that the two observables are identical.
The soil water retention guide explains the hydraulic side. A 2009 study comparing soil NMR relaxation and retention data shows why soil-specific approaches can be more successful than assuming one relaxation-to-size factor. Readers should match saturation and sample treatment before using such comparisons.
NMR signal and imaged pores answer different questions
X-ray micro-CT reconstructs image contrast in a three-dimensional grid, while low-field NMR relaxation interrogates proton-bearing fluid environments through time-dependent magnetisation. A micro-CT map may reveal larger geometrical voids but miss unresolved fine pores. NMR can respond to water in smaller environments while retaining ambiguity about exact geometrical shape and surface chemistry.
The soil micro-CT pore segmentation guide explains voxel resolution, segmentation and connected networks. A combined study can be stronger than either method alone when sample treatment and moisture are comparable. It should not label NMR peaks and CT components as one-to-one pore objects without appropriate calibration.
What an inversion peak width can and cannot mean
A narrow T2 peak may indicate that the chosen inversion assigns much of the detected signal to a limited relaxation range. It need not establish a narrow geometric pore-size distribution. Finite echo sampling, regularisation and the assumed relaxation kernel affect how a real range of environments appears in a reconstructed spectrum. A broad peak can also reflect several processes that overlap in relaxation time.
When a paper interprets one peak as a physical pore class, look for validation using moisture steps, independent imaging or a controlled reference material. A sensitivity check across plausible inversion parameters is especially helpful. If the peak location remains stable but its width changes greatly with regularisation, the location may be more defensible than a claim about detailed pore-size breadth. Describe the robust result, not every artefact of the fitting output.
Why inverse reconstruction does not discover unlimited detail
An echo train contains a finite number of noisy observations. Dividing the inversion grid into hundreds of candidate relaxation times does not create hundreds of independent measurements. The apparent resolution of a plotted curve may be finer than the instrument’s ability to distinguish adjacent components. For two similar relaxation times, the fitting problem can be poorly conditioned.
An informative report describes the time window, noise level, chosen grid and resolution checks. If a claimed tiny secondary peak sits beside a broad main peak, ask whether it survives modest variations in the fitting settings. This is a general lesson in technical translation: numerical smoothness and finely spaced axes should not be confused with new experimental information.
Water saturation changes which pores contribute signal
At near saturation, water can occupy a broad range of connected and isolated spaces that are accessible under the sample preparation. As matric potential changes during drainage, some larger pores empty and their water signal diminishes. The relaxation distribution may consequently shift toward components associated with more strongly held or differently situated water.
The solid framework may remain substantially unchanged during this transition. Therefore, a change in T2 distribution does not uniquely indicate pore collapse or swelling. To support structural change, compare independent images, dimensions or retention behaviour. If the question is water availability, report the state of the sample and the amount of detectable water rather than forcing the observation into a purely geometric narrative.
Shrink–swell soils make the interpretation harder
Certain clay-bearing or organic-rich soils can change volume as water conditions change. In those systems, drainage may remove water and alter pore geometry simultaneously. A T2 shift can reflect both changes in water occupancy and changes in the physical framework. Different clays and organic matter surfaces can also change relaxivity and interactions at the interface.
A responsible experiment records sample volume or deformation where it matters and avoids assuming a fixed pore model across all moisture states. A sample measured wet and then air-dried may not be structurally identical when rewetted. If a study claims irreversible structural change, it needs additional evidence beyond a single pair of relaxation spectra.
Internal magnetic gradients create extra transverse decay
Magnetic susceptibility differences between water and mineral grains can create gradients inside a soil sample. Water molecules diffusing through these gradients can contribute to the observed decay during a pulse sequence. Consequently a measured T2 may become shorter for reasons other than the surface-to-volume relation used to infer pore sizes.
Echo spacing is important because diffusion-related attenuation can depend on sequence timing. If two measurements use different echo spacing, their relaxation distributions may not be directly comparable without accounting for such effects. This is why a methods section should specify timing parameters rather than publishing only the final T2 histogram.
Surface relaxivity can vary within one soil
Even one soil sample may have mineral and organic surfaces with different capacities to promote water-proton relaxation. A constant surface relaxivity is a useful model approximation but can be inadequate when pore walls have heterogeneous chemistry. A pore dominated by iron-rich surfaces might relax differently from a pore of similar shape bounded by different materials.
This variation complicates attempts to assign a unique radius to each T2 value. When a study reports a pore-size distribution from relaxometry, determine how ρ2 was calibrated and whether its uncertainty was propagated. A physically attractive pore histogram is weaker when its central conversion parameter was assumed without suitable justification.
Bulk water relaxation adds another term
A more complete representation of transverse relaxation can include contributions from bulk-fluid relaxation, surface relaxation and diffusion through magnetic field gradients. Writing 1/T2 approximately equal to ρ2S/V is therefore a conditional simplification, not a complete law for every soil. If bulk relaxation is substantial, ignoring it can misattribute chemical effects to surface area.
A reader can test the logic conceptually. Suppose the same pore structure is filled with two fluids whose intrinsic relaxation differs. The measured decay may change even though the pore radius did not. This thought experiment shows why water chemistry and calibration belong in a pore-size inference. No numerical output can reveal a unique geometry from incomplete physical assumptions.
A second worked radius example with geometry
Keep an illustrative T2 of 0.10 seconds and surface relaxivity 10 micrometres per second. Under a long-cylinder model, radius equals approximately 2ρ2T2, giving 2 micrometres. Under a sphere model, the ideal S/V is 3/r, so radius is approximately 3ρ2T2, giving 3 micrometres. The inferred diameter changes from about four to six micrometres.
The NMR signal is identical in both calculations. Only the geometric model changed. Natural soils contain irregular and interconnected spaces, so neither idealisation should be presented as an individually observed cavity. The exercise teaches that a number needs a model label and uncertainty, especially when comparing pore sizes produced by different analytical methods.
A single universal T2 cut-off is risky
Some porous-media applications divide water into ‘bound’ and ‘free’ components using one numerical relaxation threshold. Such a threshold may have practical meaning under a particular calibration, mineral composition and saturation state. It cannot automatically be transferred to every clay-rich soil or organic layer, because relaxivity, internal gradients and the amount of undetected short-T2 water differ.
When a report adopts a cut-off, identify its original source and validation method. Prefer reporting a threshold as ‘the operational cut-off used in this analysis’ rather than as an intrinsic boundary between physically bound and freely flowing water. Real water mobility is a continuum influenced by pore connectivity, capillarity and chemical interactions.
Water mobility and hydraulic conductivity are not synonyms
NMR relaxation can provide insights into water environments and, with calibrated models, aspects of pore structure or mobility. Hydraulic conductivity, however, measures how readily water moves through connected pathways under a driving gradient. Two samples may have similar T2 distributions but different pore connectivity, throat restrictions or water-filled continuity, leading to different conductivities.
A hydraulic interpretation should therefore distinguish measured relaxation from a modelled property and independently tested flow. If a paper estimates permeability from a T2-derived pore-size distribution, read its equation and validation dataset. A high correlation within one set of samples does not establish a universal conversion across soils with different textures and mineralogy.
How NMR compares with mercury intrusion
Mercury intrusion porosimetry uses pressure-driven entry of a non-wetting liquid into accessible pores under idealised capillary assumptions. It gives information strongly influenced by entry throats, sample preparation and pressure history. NMR relaxometry instead responds to hydrogen-bearing fluid in pore environments and can investigate water at different moisture states under suitable conditions.
These are not duplicate pore-size instruments. One may assign an entry diameter to an ink-bottle pore’s narrow throat while another responds to water within its larger body. If two distributions disagree, check the definition of size before deciding the sample measurements are inconsistent. Differences may be informative about what the methods actually detect.
How to examine NMR measurements during drying
A useful drying series records water mass or volumetric content alongside each relaxation measurement. Plot changes in total detected signal and the T2 distribution as the sample loses water. Distinguish normalised component areas from absolute signal: when all curves sum to one, a growing short-T2 percentage can occur merely because longer-T2 water was lost faster.
Suppose initial detected signal is 1,000 units, with 300 in a short component and 700 in a long component. After drying, total signal becomes 400 units, with 240 short and 160 long. The short share rises from 30% to 60%, although its absolute signal falls from 300 to 240. The correct interpretation must keep both relative and absolute scales.
A time series needs temperature and field consistency
Repeated NMR measurements over days can track changing water distribution in a soil sample, but small differences in temperature, sample placement and hardware settings can affect the recorded decay. If water is added or removed between measurements, the process may also disturb the original structure. A change in the curve therefore requires a stable experimental basis.
The best time-course reports describe measurement repeatability and how instrument conditions were standardised. They may use gravimetric water checks and independently measured dimensions. Without such information, a tiny apparent shift can be overinterpreted. A reader should ask whether the reported difference exceeds ordinary instrumental and sample-handling variability.
What proton signal from organic matter might mean
Some soils contain substantial hydrogen-bearing organic matter in addition to water. Depending on the instrument, pulse timing and material mobility, contributions from organic phases may overlap with or be invisible in the detected window. A relaxation peak should not automatically be classified as water merely because the sample was moist.
The researcher should describe the intended signal and any measures used to evaluate non-water contributions. Comparing a carefully prepared wet and dry reference can sometimes inform interpretation, but drying may change both structure and material properties. A good method acknowledges those limitations instead of treating the dry soil as an inert empty background.
Calibration with independent water mass
Gravimetric water content is calculated from the difference between wet and dry mass divided by a defined dry-soil mass. An independently measured water mass can help assess whether integrated NMR signal follows total sample water across a set of conditions. A calibration may reveal that some rapidly relaxing water is not fully detected.
For an illustrative 100-gram dry soil sample whose wet mass is 125 grams, gravimetric water content is 25 grams divided by 100 grams, or 0.25 grams of water per gram of dry soil. That value is not automatically volumetric water content. Converting it to a volume fraction requires bulk density or an equivalent measured sample-volume relationship.
From mass water content to volumetric water content
Let gravimetric water content be 0.25 grams of water per gram of dry soil and dry bulk density be 1.20 grams of dry soil per cubic centimetre of bulk soil. Assuming water density near one gram per cubic centimetre, volumetric water content is approximately 0.25 × 1.20 = 0.30 cubic centimetres of water per cubic centimetre of soil.
The calculation identifies the reference volume that a water content uses. It does not come from NMR alone; it uses independently measured dry mass and bulk density. When reading an NMR calibration, verify whether the reported water content is gravimetric, volumetric or an instrument-normalised signal. Those categories cannot be silently interchanged.
The danger of extrapolating from sand to clay
A model calibrated in clean quartz sand may have comparatively uniform surface properties, whereas a clay-rich soil can contain paramagnetic minerals, complex bound-water environments and swelling surfaces. The same relaxation time may correspond to different characteristic pore environments. Applying one calibration across both without validation can produce biased pore-size estimates.
A comparison of natural-soil NMR and retention-derived pore distributions shows why sample-specific relaxivity can matter. The broader lesson is that calibration belongs to a defined material domain. When translating an empirical model’s performance, name the range of soil types it was tested on and avoid promising accuracy outside that range.
Sample volume and representativeness remain limitations
A small core may not capture rare root channels or cracks that control larger-scale drainage. NMR can sample a water-bearing volume, but its result represents the instrument’s sensitive region and the specimen’s conditions. A soil field is spatially heterogeneous, and repeated technical scans of one core cannot substitute for independent field samples.
To generalise, inspect sample collection, replication and whether soil layers or management units were represented. A precisely fitted relaxation distribution from one handful of soil may be excellent analytical work but still insufficient to characterise an entire landscape. Statistical inference is separate from instrumental measurement quality.
What an NMR figure should disclose
A readable figure identifies T2 in milliseconds or seconds, the scale of the horizontal axis, whether amplitude is absolute or normalised, and the inversion method. It should report the water state and acquisition parameters, particularly echo spacing. If curves are offset vertically for visual clarity, the offsets must be explained.
The caption should not simply call the x-axis pore size unless an explicit validated T2-to-size conversion has been applied. When pore sizes are shown, list surface relaxivity, assumed geometry and calibration range. Good captions keep the measurement and inference visible, allowing readers to understand how much the figure demonstrates without searching for hidden assumptions.
Common reading mistakes and targeted repairs
One error is treating the shortest detected T2 as the smallest physical pore. Repair it by checking the acquisition window, relaxivity and geometry. Another is interpreting a longer peak as universally more mobile water. Repair it by confirming the soil-specific calibration and saturation state. A third is equating a normalised peak area with absolute water volume. Repair it using total-signal data and independent water calibration.
A fourth error is calling every T2 shift a structural change. Repair it by considering moisture occupancy, fluid chemistry and magnetic gradients. A fifth is treating inverse-model peaks as directly observed pore classes. Repair it by checking regularisation sensitivity. Finally, do not equate a derived pore-size distribution with a measured hydraulic conductivity.
Independent exercise: diagnose the false pore claim
A hypothetical study reports T2 = 0.10 seconds in two soils and assigns both the same pore diameter. Soil A has a calibrated surface relaxivity of 10 micrometres per second; soil B has 30 micrometres per second. Under the same cylindrical surface-dominated model, A’s inferred diameter is four micrometres while B’s is twelve. The identical T2 values do not imply identical pore radii when the conversion parameter differs.
Now suppose soil B also has strong internal magnetic gradients. The simplified conversion may not be valid at all. The correct response is to assess the physical model before trusting the numbers. This is a transfer skill applicable to many inverse measurement problems: an observable becomes a material property only through warranted assumptions.
Independent exercise: keep the denominator
A paper shows a normalised short-T2 peak growing from 30% to 60% as soil dries. Can the reader say the quantity of short-T2 water doubled? No. The share may rise because longer-T2 signal diminished. A valid comparison needs unnormalised signal or independent calibration of total detectable water.
If total signal declined from 1,000 units to 400, the first short component was 300 units, while the second at 60% was 240 units. Its absolute signal decreased despite its growing share. The worked solution reveals why reporting percentages alone can reverse an interpretation of water amount.
Frequently asked questions about soil NMR
Is NMR a photograph of pores? No. Does a short T2 always mean a small pore? Not without method-specific assumptions. Can water content change T2 while geometry remains similar? Yes. Does a bigger normalised peak mean more water mass? Not necessarily. Is surface relaxivity identical in all minerals? No.
Can inverse Laplace fitting invent apparent tiny peaks? Noise and regularisation can change the result. Is a pore-size estimate the same as measured hydraulic conductivity? No. Can paramagnetic iron influence relaxation? Yes. Does a T1 experiment measure exactly the same property as T2? No. Can one core describe a whole field without sampling evidence? No. The recurring answer is to identify the observable before choosing the ecological claim.
Primary research and connected reading
The 2018 low-field borehole NMR study explains relaxation physics, water state and pore-environment caveats. Research on quantitative soil 1H NMR pore-size distributions and comparison with water-retention measurements shows why calibration matters. A comparative NMR and soil-hydraulic study addresses complex relaxation mechanisms in real porous media.
For neighbouring interpretations, read soil X-ray micro-CT pore connectivity, soil water retention and field capacity and cross-linguistic vocabulary transfer. The first concerns imaged structure, the second hydraulic state and the third precise language when one technical word changes meaning.
Final principle: follow water signal through the model
Soil NMR relaxometry offers a powerful view of how proton-bearing water behaves in a complex porous material. Its measured echo decay is transformed into relaxation distributions and, in some applications, modelled pore-size estimates. Every transformation adds conditions that can be tested and explained.
The reliable reader starts with the water state and pulse sequence, inspects detectable signal and inversion choices, then examines surface relaxivity and independent calibration before making a statement about soil pores. That sequence allows researchers and learners to appreciate the measurement without treating a fitted peak as an actual image of a pore or an unverified promise of hydraulic conductivity.