From gravity changes to trustworthy ocean-mass statements
A GRACE ocean bottom-pressure anomaly map can look like an ordinary map of rising and falling water, but it is a processed estimate of mass-related change, normally expressed as equivalent water thickness. To read it correctly, separate the satellite gravity observation from the inversion, the monthly reference period, the pressure-equivalent unit, and the physical meaning of ocean mass. This guide makes those transformations explicit with worked examples and independent checks.
Three tempting shortcuts are unsafe: interpreting positive EWT as an identical rise in local sea surface, treating every coastal pixel as directly observed ocean mass, and comparing a monthly gravity field with a daily tide gauge without reconciling time scale. The alternative is a reversible chain from observation through corrections and units to a bounded conclusion.
The NASA ocean bottom-pressure product descriptions report the use of gravity data and processing corrections including glacial isostatic adjustment, background atmosphere–ocean model fields, and low-degree gravity terms. The exact handling depends on the release. Each section below asks what can legitimately be inferred and what further evidence is required.
The result is a mass-equivalent anomaly, not a tide gauge
How to translate GRACE ocean bottom-pressure results easily starts by asking what the satellite system actually detects. Gravity satellites infer mass redistribution from changes in the Earth’s gravity field. Processing steps translate that inference into an equivalent-water-thickness anomaly over an ocean region. The resulting centimetres are useful precisely because they permit mass accounting. They are not automatically centimetres of sea-surface height at the location of a map pixel. This distinction anchors the interpretation of ocean mass change, atmospheric loading, steric expansion, and long-term sea-level budgets. The worked numbers below are teaching examples, not claims about particular observed months.
The three quantities to keep separate
First consider ocean bottom pressure: it reflects the weight of overlying water and atmosphere under the chosen convention. Second consider equivalent water thickness: a way of writing surface mass density as the thickness of a hypothetical water layer. Third consider sea-surface height: it responds both to mass and to seawater density changes. In introductory arithmetic with density 1000 kg/m³ and gravity 9.81 m/s², one centimetre EWT equals 10 kg/m² and about 98.1 Pa. A physical water column may have other densities and atmospheric effects, so this conversion is an interpretation scaffold, not a replacement for reading the actual product definition.
A complete end-to-end check
Suppose a monthly GRACE product provides +0.020 m EWT at one broad ocean location, relative to its declared time mean. Multiply by 1000 kg/m³ to obtain +20 kg/m² mass equivalent. Multiply again by 9.81 m/s² to obtain about +196 Pa of bottom-pressure equivalent. The sign says the inferred field is above its chosen mean, not that absolute sea level there is 20 mm higher. To compare it with a bottom-pressure recorder, account for tides, atmosphere and sampling. To compare it with altimetry, consider the steric component. To assess accuracy, inspect the product’s smoothing, leakage treatment, geophysical corrections and uncertainty.
A practical order for reading data files
Identify the dataset’s full release title and producer first. Then locate its time coverage, time averaging, variable attributes, EWT unit, anomaly reference period and coordinate grid. Read the product’s correction description: some NASA products state that a GAD background term has been added back, glacial isostatic adjustment has been applied, low-degree terms revised, and fields de-striped or smoothed. These are scientific decisions. Record them before plotting trends or differences. Finally decide whether your question concerns regional mass redistribution, global mean ocean mass, or closure with altimeter and Argo observations. A map alone does not guarantee that one product is suited to all three.
How careful language protects the science
A report should give a reader enough information to reconstruct its claim. Say ‘inferred monthly ocean bottom-pressure anomaly of +1 cm equivalent water thickness relative to the stated mean’ rather than ‘the sea rose one centimetre’. If comparing two months, use the same release, baseline and ocean mask. If discussing uncertainty, distinguish measurement noise, omitted data, coastal leakage, GIA model choices and systematic product differences. Never convert a processing correction into an alleged new observation. Scientific translation is successful when the reader understands both the useful finding and the boundaries of what the instrument and processing can tell us.
Thirty interpretation and verification lessons
1. What an orbiting gravity mission actually senses
GRACE and GRACE Follow-On infer changes in Earth’s gravitational field from exquisitely sensitive satellite tracking; they do not carry ordinary pressure gauges resting on the seabed. Ocean bottom-pressure anomaly is an interpreted mass-related field obtained after processing gravity information and applying models and corrections. Keep the unit on the page, write the reference period explicitly, and distinguish an inference from a directly measured sea level. These three annotations make the statement testable by a reader using the original product documentation.
A monthly field of +2 cm equivalent water thickness means a change in column-integrated mass represented as the weight of a water layer two centimetres thick, relative to its reference. It does not assert that the sea surface rose two centimetres at that spot. In this worked example the figures are invented to demonstrate a calculation or a diagnostic distinction; they are not a reported GRACE measurement. Preserve units, signs and denominators so the reasoning can be reproduced without access to an illustration.
Treat satellite-derived bottom pressure as an inference with a documented processing chain, not a photograph of seawater height. Translation must preserve the words anomaly, equivalent water thickness, resolution and reference period. Where the source does not disclose a needed parameter, mark the inference as unresolved rather than inventing a correction or claiming unwarranted precision. A stronger report can be written after the relevant source field is verified.
2. Water-equivalent thickness is a mass unit in disguise
Equivalent water thickness, often abbreviated EWT, translates surface mass per area into the thickness of a hypothetical water layer at a specified reference density. With 1000 kg/m³ water, one centimetre corresponds to 10 kg/m² of mass anomaly. To test the claim, reconstruct its arithmetic in base SI units before converting back to a convenient map label. An order-of-magnitude check catches many tenfold and hundredfold errors without specialist software.
A map pixel labelled +1.5 cm EWT represents about +15 kg/m² relative to its chosen mean. Multiply 0.015 m by 1000 kg/m³ to recover 15 kg/m². This is a conversion of units, not an independently measured sea-level change. In this worked example the figures are invented to demonstrate a calculation or a diagnostic distinction; they are not a reported GRACE measurement. Preserve units, signs and denominators so the reasoning can be reproduced without access to an illustration.
Do not call centimetres EWT centimetres of local sea-level rise unless the relevant assumptions and spatial context justify that reinterpretation. Write ‘centimetres of equivalent water thickness’ in full at first mention. Where the source does not disclose a needed parameter, mark the inference as unresolved rather than inventing a correction or claiming unwarranted precision. A stronger report can be written after the relevant source field is verified.
3. Converting EWT to approximate pressure
Hydrostatic pressure from a thin uniform layer is Δp = ρgΔh, where ρ is the reference density, g is local gravitational acceleration, and Δh is equivalent thickness. The same quantity is also approximately surface mass per area multiplied by g. The best explanation identifies which part is instrument evidence, which part is processing, and which part is interpretation. Where another product could answer a different question, name that limitation instead of silently conflating outcomes.
Take +0.020 m EWT, density 1000 kg/m³ and g = 9.81 m/s². Then Δp ≈ 196.2 pascals, or approximately 1.962 hectopascals. This result is a gravitational pressure equivalent, not a direct gauge reading. In this worked example the figures are invented to demonstrate a calculation or a diagnostic distinction; they are not a reported GRACE measurement. Preserve units, signs and denominators so the reasoning can be reproduced without access to an illustration.
Record the reference density and gravity assumption. The convenient formula is a physical interpretation aid; a published satellite product may have additional ocean-atmosphere terms and corrections that determine precisely what its displayed value represents. Where the source does not disclose a needed parameter, mark the inference as unresolved rather than inventing a correction or claiming unwarranted precision. A stronger report can be written after the relevant source field is verified.
4. Anomaly means a declared time reference
A GRACE ocean mass map ordinarily shows departures relative to a specified time-mean reference period. Positive and negative values say more or less mass-related signal than that reference; they do not display the full static mass of the ocean. These distinctions matter for comparisons across missions, regions and time. If a graph hides its mask, baseline or correction methods, it does not supply enough context for confident attribution or a precise long-term trend.
Suppose a grid cell is +1 cm in one month and −0.5 cm in another relative to a common baseline. The difference is 1.5 cm EWT, but neither monthly number gives the absolute depth, absolute ocean mass or mean bottom pressure. In this worked example the figures are invented to demonstrate a calculation or a diagnostic distinction; they are not a reported GRACE measurement. Preserve units, signs and denominators so the reasoning can be reproduced without access to an illustration.
Always state the baseline period when comparing maps. If two datasets subtract different time means, their colour zero lines can differ even when both reflect the same physical variability. Where the source does not disclose a needed parameter, mark the inference as unresolved rather than inventing a correction or claiming unwarranted precision. A stronger report can be written after the relevant source field is verified.
5. Monthly estimates do not resolve daily tides
The gravity-field solutions typically represent monthly or near-monthly sampling windows. A single map is not a direct daily tidal time series and should not be used to infer a particular storm-surge peak or tide at a harbour. A useful translation carries the original technical boundary into ordinary language. Resist a visually appealing stronger claim if the input does not establish the physical process, sampling period or uncertainty involved.
If a map for June shows +0.8 cm EWT, it is not evidence of an 8 mm increase at precisely noon on 14 June. A strong short event can be diluted in a monthly mean or imperfectly represented by processing. In this worked example the figures are invented to demonstrate a calculation or a diagnostic distinction; they are not a reported GRACE measurement. Preserve units, signs and denominators so the reasoning can be reproduced without access to an illustration.
Separate sampling interval, product centre date and temporal averaging. For minute-scale sea-level dynamics use tide gauges, pressure sensors or suitable high-frequency models, not a monthly gravity anomaly grid. Where the source does not disclose a needed parameter, mark the inference as unresolved rather than inventing a correction or claiming unwarranted precision. A stronger report can be written after the relevant source field is verified.
6. Pressure is not the same as sea-surface height
Ocean bottom pressure responds to the mass of the water column and atmospheric loading, while sea-surface height responds both to mass addition and to expansion or contraction of seawater as temperature and salinity change. Before drawing a conclusion, ask what change to the selected baseline, coastal mask or processing release would alter the result. The answer shows whether the observation is robust and which assumptions remain visible.
If the sea surface rises 10 mm because of thermal expansion with negligible added ocean mass, satellite altimetry can see higher surface height while the mass-equivalent term from GRACE remains near zero. This illustrates the steric contribution. In this worked example the figures are invented to demonstrate a calculation or a diagnostic distinction; they are not a reported GRACE measurement. Preserve units, signs and denominators so the reasoning can be reproduced without access to an illustration.
The sea-level budget is often written total height change equals mass-related change plus steric change, subject to conventions and residual terms. Never equate a GRACE anomaly to the total sea-level trend without those distinctions. Where the source does not disclose a needed parameter, mark the inference as unresolved rather than inventing a correction or claiming unwarranted precision. A stronger report can be written after the relevant source field is verified.
7. Steric height needs temperature and salinity
Steric change arises because the density of seawater changes with its thermal and haline state. Profiling temperature and salinity observations, such as from Argo, are used to estimate this contribution, with coverage and depth limitations. Keep the unit on the page, write the reference period explicitly, and distinguish an inference from a directly measured sea level. These three annotations make the statement testable by a reader using the original product documentation.
An example budget with altimetry +4 mm/year and steric +1.2 mm/year would leave an illustrative mass-related residual of +2.8 mm/year if the budget and reference frames are consistent. These are invented teaching numbers, not a claim about a specific year. In this worked example the figures are invented to demonstrate a calculation or a diagnostic distinction; they are not a reported GRACE measurement. Preserve units, signs and denominators so the reasoning can be reproduced without access to an illustration.
Before interpreting such a subtraction, confirm identical calendar periods, ocean masks, spatial averaging, vertical coverage and uncertainty. Simple arithmetic alone does not prove that all observing systems have closed the budget. Where the source does not disclose a needed parameter, mark the inference as unresolved rather than inventing a correction or claiming unwarranted precision. A stronger report can be written after the relevant source field is verified.
8. An atmosphere term can appear in ocean bottom pressure
Some ocean bottom-pressure products include contributions from atmospheric mass loading alongside ocean mass. The NASA Level-3 descriptions explicitly mention the integral effects of atmosphere and ocean processes. To test the claim, reconstruct its arithmetic in base SI units before converting back to a convenient map label. An order-of-magnitude check catches many tenfold and hundredfold errors without specialist software.
If a monthly OBP field rises during a persistent regional high-pressure pattern, part of the signal can reflect atmospheric mass rather than an influx of seawater alone. Reading it as pure ocean water mass may be misleading. In this worked example the figures are invented to demonstrate a calculation or a diagnostic distinction; they are not a reported GRACE measurement. Preserve units, signs and denominators so the reasoning can be reproduced without access to an illustration.
Consult the exact product definition, especially the treatment of de-aliasing fields and atmospheric signals. Translate ocean bottom pressure anomaly and ocean-mass anomaly carefully; they can be related but are not universally interchangeable labels. Where the source does not disclose a needed parameter, mark the inference as unresolved rather than inventing a correction or claiming unwarranted precision. A stronger report can be written after the relevant source field is verified.
9. What GAD means in a processing chain
GRACE Level-2 processing uses background atmosphere-ocean model products to handle rapidly varying contributions. The GAD field represents a specific modeled ocean/atmosphere term, and several Level-3 ocean products describe GAD as having been added back. The best explanation identifies which part is instrument evidence, which part is processing, and which part is interpretation. Where another product could answer a different question, name that limitation instead of silently conflating outcomes.
Two maps called ‘ocean bottom pressure’ can differ because one incorporates the restored background component while the other retains a different reduction convention. Their difference should not automatically be interpreted as a new ocean physical process. In this worked example the figures are invented to demonstrate a calculation or a diagnostic distinction; they are not a reported GRACE measurement. Preserve units, signs and denominators so the reasoning can be reproduced without access to an illustration.
Locate the processing note for the exact release. Keep ‘added back’ and ‘removed’ as literal processing operations, and verify sign convention before subtracting maps. Where the source does not disclose a needed parameter, mark the inference as unresolved rather than inventing a correction or claiming unwarranted precision. A stronger report can be written after the relevant source field is verified.
10. Land-to-ocean signal leakage
GRACE does not resolve every sharp coastline. Gravitational signals from large land ice and groundwater mass changes can leak into nearby ocean pixels after spatial smoothing, especially where coastlines border rapidly changing land mass. These distinctions matter for comparisons across missions, regions and time. If a graph hides its mask, baseline or correction methods, it does not supply enough context for confident attribution or a precise long-term trend.
Consider an ocean grid cell adjacent to a region losing ice mass. The map may display a trend partially shaped by the neighbouring land signal rather than solely by the ocean directly above that cell. In this worked example the figures are invented to demonstrate a calculation or a diagnostic distinction; they are not a reported GRACE measurement. Preserve units, signs and denominators so the reasoning can be reproduced without access to an illustration.
Treat coastal estimates as potentially contaminated unless the product and study apply appropriate leakage corrections. Prefer broad regional analysis with documented uncertainty to unsupported small-bay conclusions. Where the source does not disclose a needed parameter, mark the inference as unresolved rather than inventing a correction or claiming unwarranted precision. A stronger report can be written after the relevant source field is verified.
11. Smoothing suppresses small spatial structure
Level-3 mass fields are commonly filtered to reduce noisy correlated patterns. Smoothing and de-striping make coherent large-scale patterns easier to see, but they also blur or attenuate genuine local features. A useful translation carries the original technical boundary into ordinary language. Resist a visually appealing stronger claim if the input does not establish the physical process, sampling period or uncertainty involved.
A narrow current-system mass feature can appear lower in amplitude and broader after smoothing. Comparing the colour of a single cell with a precise in-situ instrument may create a false mismatch caused by scale. In this worked example the figures are invented to demonstrate a calculation or a diagnostic distinction; they are not a reported GRACE measurement. Preserve units, signs and denominators so the reasoning can be reproduced without access to an illustration.
Read grid spacing separately from effective spatial resolution. A product sampled on a small grid can still embody information averaged across far larger scales. Where the source does not disclose a needed parameter, mark the inference as unresolved rather than inventing a correction or claiming unwarranted precision. A stronger report can be written after the relevant source field is verified.
12. Mascon and spherical-harmonic solutions differ
GRACE gravity inversions can be distributed as spherical-harmonic solutions or as mass concentration, or mascon, products. Their constraints, regularisation, coastal handling and correction choices may lead to different regional estimates. Before drawing a conclusion, ask what change to the selected baseline, coastal mask or processing release would alter the result. The answer shows whether the observation is robust and which assumptions remain visible.
If one mascon solution shows a broad +1.1 cm EWT anomaly and a filtered spherical-harmonic solution gives +0.7 cm, the difference can stem partly from inversion and filtering rather than an observational contradiction. In this worked example the figures are invented to demonstrate a calculation or a diagnostic distinction; they are not a reported GRACE measurement. Preserve units, signs and denominators so the reasoning can be reproduced without access to an illustration.
Specify processing centre, solution family and version in any comparative statement. Never combine maps as though their grids and filters were interchangeable without an explicit harmonisation step. Where the source does not disclose a needed parameter, mark the inference as unresolved rather than inventing a correction or claiming unwarranted precision. A stronger report can be written after the relevant source field is verified.
13. Geocentre and low-degree coefficients matter
GRACE gravity fields require careful handling of the Earth’s centre-of-mass motion and low-degree spherical-harmonic terms. Public processing notes describe geocentre and C20/C30 adjustments in some products. Keep the unit on the page, write the reference period explicitly, and distinguish an inference from a directly measured sea level. These three annotations make the statement testable by a reader using the original product documentation.
A small systematic change in an adopted low-degree coefficient can modify broad-scale ocean mass patterns. Such a correction can shift a trend estimate without any new ocean observation having occurred. In this worked example the figures are invented to demonstrate a calculation or a diagnostic distinction; they are not a reported GRACE measurement. Preserve units, signs and denominators so the reasoning can be reproduced without access to an illustration.
Record whether a product incorporated replacements for degree-one, C20 and C30 terms. A revised processing note is not simply administrative: it may change scientifically important broad-scale estimates. Where the source does not disclose a needed parameter, mark the inference as unresolved rather than inventing a correction or claiming unwarranted precision. A stronger report can be written after the relevant source field is verified.
14. Glacial isostatic adjustment is a solid-Earth correction
Postglacial rebound changes gravity and the Earth’s shape even where modern water mass is not changing. A glacial isostatic adjustment, or GIA, model can be used to separate this long-term solid-Earth contribution from contemporary mass anomalies. To test the claim, reconstruct its arithmetic in base SI units before converting back to a convenient map label. An order-of-magnitude check catches many tenfold and hundredfold errors without specialist software.
If a trend estimate changes after applying a different GIA model, the remaining difference is an uncertainty in correction assumptions rather than evidence that the ocean instantly gained or lost that amount of water. In this worked example the figures are invented to demonstrate a calculation or a diagnostic distinction; they are not a reported GRACE measurement. Preserve units, signs and denominators so the reasoning can be reproduced without access to an illustration.
When reporting long-term trends, retain the GIA model name or citation and discuss model uncertainty. Do not describe GIA corrections as measured seawater temperature changes. Where the source does not disclose a needed parameter, mark the inference as unresolved rather than inventing a correction or claiming unwarranted precision. A stronger report can be written after the relevant source field is verified.
15. Annual cycles can obscure long-term changes
Ocean mass-related signals can have seasonal cycles caused by water exchange and circulation alongside slower trends. Estimating a long-term slope from two arbitrary months can confuse a seasonal difference with a trend. The best explanation identifies which part is instrument evidence, which part is processing, and which part is interpretation. Where another product could answer a different question, name that limitation instead of silently conflating outcomes.
Suppose December is +2 cm EWT and June is −1 cm in a repeating seasonal pattern. The six-month difference does not establish a continuing rate of +6 cm per year. Multiple years and appropriate seasonal modelling are necessary. In this worked example the figures are invented to demonstrate a calculation or a diagnostic distinction; they are not a reported GRACE measurement. Preserve units, signs and denominators so the reasoning can be reproduced without access to an illustration.
Display monthly observations alongside a fitted seasonal cycle and trend when appropriate. State the fitting period, missing-month treatment and uncertainty instead of multiplying a short difference into an annual claim. Where the source does not disclose a needed parameter, mark the inference as unresolved rather than inventing a correction or claiming unwarranted precision. A stronger report can be written after the relevant source field is verified.
16. A missing GRACE month is not a zero anomaly
Mission processing and orbit or instrument limitations can leave temporal gaps. A missing grid is unknown at that interval, not an observation of average mass conditions. These distinctions matter for comparisons across missions, regions and time. If a graph hides its mask, baseline or correction methods, it does not supply enough context for confident attribution or a precise long-term trend.
If the available sequence has January +0.5 cm and March +1.0 cm with February missing, a chart joining points may be visually convenient, but it cannot establish that February had +0.75 cm without interpolation assumptions. In this worked example the figures are invented to demonstrate a calculation or a diagnostic distinction; they are not a reported GRACE measurement. Preserve units, signs and denominators so the reasoning can be reproduced without access to an illustration.
Keep missing months as gaps in scientific tables or clearly mark estimates. Do not feed zeros into regression simply to produce a complete rectangular array. Where the source does not disclose a needed parameter, mark the inference as unresolved rather than inventing a correction or claiming unwarranted precision. A stronger report can be written after the relevant source field is verified.
17. An apparent trend needs an uncertainty model
Mass anomalies contain measurement, modelling, processing and sampling errors, which can be temporally correlated. A least-squares slope alone does not describe its credibility when seasonal or serial correlation remains. A useful translation carries the original technical boundary into ordinary language. Resist a visually appealing stronger claim if the input does not establish the physical process, sampling period or uncertainty involved.
A fitted rate of +2.0 ± 0.4 mm/year should preserve the definition of the uncertainty interval and the statistical method. Two data products with apparently different central rates may still overlap within uncertainties. In this worked example the figures are invented to demonstrate a calculation or a diagnostic distinction; they are not a reported GRACE measurement. Preserve units, signs and denominators so the reasoning can be reproduced without access to an illustration.
Distinguish formal regression error from broader systematic uncertainty such as GIA and coastal leakage. Never present a central slope with more decimal places than the error model can support. Where the source does not disclose a needed parameter, mark the inference as unresolved rather than inventing a correction or claiming unwarranted precision. A stronger report can be written after the relevant source field is verified.
18. Global ocean average requires careful masking
Computing a global-mean ocean mass anomaly requires an ocean mask and area-weighting decisions. Land contamination, ice-covered regions, marginal seas and coastal zones alter the balance. Before drawing a conclusion, ask what change to the selected baseline, coastal mask or processing release would alter the result. The answer shows whether the observation is robust and which assumptions remain visible.
If one grid covers five times as much ocean area as another, its contribution to the area mean is five times larger under simple area weighting. Averaging all displayed grid numbers without area weights misrepresents the basin average. In this worked example the figures are invented to demonstrate a calculation or a diagnostic distinction; they are not a reported GRACE measurement. Preserve units, signs and denominators so the reasoning can be reproduced without access to an illustration.
Report the exact ocean area used and how coastal and polar cells were treated. The denominator is physical ocean area, not the count of data-table rows. Where the source does not disclose a needed parameter, mark the inference as unresolved rather than inventing a correction or claiming unwarranted precision. A stronger report can be written after the relevant source field is verified.
19. Regional anomalies are not globally transferable
A positive anomaly in one basin can coexist with a negative anomaly in another because ocean mass is redistributed by circulation and atmospheric forcing. A coloured regional map is not necessarily the global ocean mass budget. Keep the unit on the page, write the reference period explicitly, and distinguish an inference from a directly measured sea level. These three annotations make the statement testable by a reader using the original product documentation.
A hypothetical +2 cm patch in the northern basin and −2 cm patch in the southern basin does not guarantee cancellation; their areas and true mass contributions must be weighted and compared. In this worked example the figures are invented to demonstrate a calculation or a diagnostic distinction; they are not a reported GRACE measurement. Preserve units, signs and denominators so the reasoning can be reproduced without access to an illustration.
Keep regional distribution, basin means and global total distinct. Translate ‘mass redistribution’ separately from ‘global mass gain’ so the physical mechanism is not changed. Where the source does not disclose a needed parameter, mark the inference as unresolved rather than inventing a correction or claiming unwarranted precision. A stronger report can be written after the relevant source field is verified.
20. Temporal filtering changes the question
A seasonal cycle removed for trend analysis, a 12-month running mean, and an unfiltered monthly record are different representations of the same observations. Filtering can shift peaks and obscure short-lived variability. To test the claim, reconstruct its arithmetic in base SI units before converting back to a convenient map label. An order-of-magnitude check catches many tenfold and hundredfold errors without specialist software.
If a monthly anomaly jumps temporarily by 3 cm and returns to zero, a 12-month average will show a much smaller, broader deviation. It would be mistaken to conclude that the instrument did not observe a sharp monthly episode. In this worked example the figures are invented to demonstrate a calculation or a diagnostic distinction; they are not a reported GRACE measurement. Preserve units, signs and denominators so the reasoning can be reproduced without access to an illustration.
Provide both filter description and time scale. Ask whether the audience needs monthly dynamics, multi-year variability or a fitted trend before choosing the appropriate graph. Where the source does not disclose a needed parameter, mark the inference as unresolved rather than inventing a correction or claiming unwarranted precision. A stronger report can be written after the relevant source field is verified.
21. Reference density is a convention, not a chemical assay
Equivalent water thickness uses a standard density for translating mass per area into a length. It does not measure the instantaneous temperature-dependent density of the real seawater column. The best explanation identifies which part is instrument evidence, which part is processing, and which part is interpretation. Where another product could answer a different question, name that limitation instead of silently conflating outcomes.
Ten kg/m² of added mass corresponds to 1 cm EWT at 1000 kg/m³, but the thickness of an actual parcel of warm or salty seawater with different density would not be exactly the same. In this worked example the figures are invented to demonstrate a calculation or a diagnostic distinction; they are not a reported GRACE measurement. Preserve units, signs and denominators so the reasoning can be reproduced without access to an illustration.
Translate EWT as an accounting equivalent and retain the density convention. When physical sea-surface height is needed, use the correct seawater density and distinguish steric volume effects. Where the source does not disclose a needed parameter, mark the inference as unresolved rather than inventing a correction or claiming unwarranted precision. A stronger report can be written after the relevant source field is verified.
22. Mapping pressure to centimetres can invite false precision
The conversion of approximately 100 pascals to about one centimetre of water is convenient at ordinary gravity and reference density. It is an approximation whose constants must match the problem being solved. These distinctions matter for comparisons across missions, regions and time. If a graph hides its mask, baseline or correction methods, it does not supply enough context for confident attribution or a precise long-term trend.
With 50 Pa pressure change and ρ=1000 kg/m³, g=9.81 m/s², the equivalent thickness is 50/9810 m, about 0.00510 m or 0.51 cm. Rounding it to 5.1 cm is a tenfold unit error. In this worked example the figures are invented to demonstrate a calculation or a diagnostic distinction; they are not a reported GRACE measurement. Preserve units, signs and denominators so the reasoning can be reproduced without access to an illustration.
Write every conversion step, including Pa, kg/m², m and cm. Check order of magnitude independently before interpreting a pressure map in a report for non-specialists. Where the source does not disclose a needed parameter, mark the inference as unresolved rather than inventing a correction or claiming unwarranted precision. A stronger report can be written after the relevant source field is verified.
23. Ocean-bottom pressure gauges provide a different scale of evidence
In-situ bottom-pressure recorders observe local pressure at a particular site and high sampling frequency. They are important cross-checks for local dynamics but do not share satellite gravity’s native spatial footprint. A useful translation carries the original technical boundary into ordinary language. Resist a visually appealing stronger claim if the input does not establish the physical process, sampling period or uncertainty involved.
An instrument on the seabed can resolve tides and pressure pulses that a monthly GRACE solution cannot separate. Conversely, an individual gauge says little by itself about the integrated mass over a whole ocean basin. In this worked example the figures are invented to demonstrate a calculation or a diagnostic distinction; they are not a reported GRACE measurement. Preserve units, signs and denominators so the reasoning can be reproduced without access to an illustration.
Compare compatible time averages after accounting for tides, atmosphere, instrument drift and spatial scale. Two pressure products are not automatically duplicates merely because their axis labels match. Where the source does not disclose a needed parameter, mark the inference as unresolved rather than inventing a correction or claiming unwarranted precision. A stronger report can be written after the relevant source field is verified.
24. Sea-level closure is a relationship among three observing systems
Satellite altimetry measures sea-surface height relative to a geodetic reference. Argo profiles constrain steric height through seawater density. GRACE provides a mass-related contribution. Their budget comparison tests consistency of independent methods. Before drawing a conclusion, ask what change to the selected baseline, coastal mask or processing release would alter the result. The answer shows whether the observation is robust and which assumptions remain visible.
For an illustrative year, altimetry +5 mm/year, Argo steric +1 mm/year and GRACE mass +3.5 mm/year leave +0.5 mm/year as a residual. That residual may reflect coverage, bias and processing, not an instantly identifiable physical fourth mechanism. In this worked example the figures are invented to demonstrate a calculation or a diagnostic distinction; they are not a reported GRACE measurement. Preserve units, signs and denominators so the reasoning can be reproduced without access to an illustration.
Do not force exact arithmetic closure by silently changing dates or masks. Document the residual, formal uncertainty and known measurement differences before stating what the observations support. Where the source does not disclose a needed parameter, mark the inference as unresolved rather than inventing a correction or claiming unwarranted precision. A stronger report can be written after the relevant source field is verified.
25. A negative EWT anomaly is not ‘negative water’
Negative anomaly on the map means the inferred mass-equivalent field is below its chosen reference level. It does not mean physical ocean depth is negative or that water is absent. Keep the unit on the page, write the reference period explicitly, and distinguish an inference from a directly measured sea level. These three annotations make the statement testable by a reader using the original product documentation.
A cell at −2 cm EWT compared with its time-mean mass column still contains an immense amount of seawater. The signed value concerns a small variation about a baseline, not the total column. In this worked example the figures are invented to demonstrate a calculation or a diagnostic distinction; they are not a reported GRACE measurement. Preserve units, signs and denominators so the reasoning can be reproduced without access to an illustration.
Use ‘below the reference mean by two centimetres water equivalent’ rather than ‘minus two centimetres of ocean’. This is a language repair that preserves the measurement without distorting it. Where the source does not disclose a needed parameter, mark the inference as unresolved rather than inventing a correction or claiming unwarranted precision. A stronger report can be written after the relevant source field is verified.
26. A map update can change old numbers
Reprocessing can revise correction models, gravity solutions, masks, smoothing and calibration assumptions. A newer release may give different estimates for a month observed years earlier. To test the claim, reconstruct its arithmetic in base SI units before converting back to a convenient map label. An order-of-magnitude check catches many tenfold and hundredfold errors without specialist software.
If the first data release showed +0.9 cm and a later release +1.1 cm for the same historical month, that does not imply a new event occurred in the past. The product version changed. In this worked example the figures are invented to demonstrate a calculation or a diagnostic distinction; they are not a reported GRACE measurement. Preserve units, signs and denominators so the reasoning can be reproduced without access to an illustration.
Cite release number, access date and variable name. Keep an archived computation record so an earlier estimate can be reproduced and transparently compared with later improvements. Where the source does not disclose a needed parameter, mark the inference as unresolved rather than inventing a correction or claiming unwarranted precision. A stronger report can be written after the relevant source field is verified.
27. The global mean and a regional optimized product are not always the same
Some GRACE ocean mass representations are optimized for regional bottom-pressure patterns and should not be assumed to preserve the global mean mass signal without the appropriate treatment. Dataset notes may explicitly warn about this. The best explanation identifies which part is instrument evidence, which part is processing, and which part is interpretation. Where another product could answer a different question, name that limitation instead of silently conflating outcomes.
Imagine a processed map designed to remove broad background components before highlighting regional differences. Taking a raw grid mean as global ocean mass trend would then answer the wrong physical question. In this worked example the figures are invented to demonstrate a calculation or a diagnostic distinction; they are not a reported GRACE measurement. Preserve units, signs and denominators so the reasoning can be reproduced without access to an illustration.
Read whether the product is meant for regional anomalies, global mean ocean mass or both. Prefer the product’s own documentation over the intuitive interpretation of its colour scale. Where the source does not disclose a needed parameter, mark the inference as unresolved rather than inventing a correction or claiming unwarranted precision. A stronger report can be written after the relevant source field is verified.
28. Comparing GRACE and GRACE Follow-On requires a continuity plan
The original GRACE mission and its follow-on are related but distinct measurement periods and processing histories. Stitching them into a single trend needs attention to calibration, gaps and the chosen release. These distinctions matter for comparisons across missions, regions and time. If a graph hides its mask, baseline or correction methods, it does not supply enough context for confident attribution or a precise long-term trend.
If a graph has an interval without mission observations between two segments, drawing an uninterrupted line does not fill that missing evidence. An apparent step could be geophysical or processing-related. In this worked example the figures are invented to demonstrate a calculation or a diagnostic distinction; they are not a reported GRACE measurement. Preserve units, signs and denominators so the reasoning can be reproduced without access to an illustration.
Report the time boundaries, gap and common processing conventions. Where appropriate include independent altimetry-Argo checks rather than assuming an unbroken instrumental record. Where the source does not disclose a needed parameter, mark the inference as unresolved rather than inventing a correction or claiming unwarranted precision. A stronger report can be written after the relevant source field is verified.
29. What the map cannot establish by itself
A regional ocean mass anomaly does not reveal which glacier, river, groundwater store, wind system or atmospheric condition caused it. Attribution needs additional evidence and a model of transfers. A useful translation carries the original technical boundary into ordinary language. Resist a visually appealing stronger claim if the input does not establish the physical process, sampling period or uncertainty involved.
A positive mass anomaly in a basin could combine changes in net water input, redistribution and atmospheric loading. Merely reading a positive pixel cannot distinguish these causes or quantify source contributions. In this worked example the figures are invented to demonstrate a calculation or a diagnostic distinction; they are not a reported GRACE measurement. Preserve units, signs and denominators so the reasoning can be reproduced without access to an illustration.
Use a causal chain that says what is observed, what is inferred, and what remains hypothetical. An accessible explanation should disclose this limit rather than convert a map into an unsupported causal narrative. Where the source does not disclose a needed parameter, mark the inference as unresolved rather than inventing a correction or claiming unwarranted precision. A stronger report can be written after the relevant source field is verified.
30. Preserving technical meaning in translation
The words anomaly, equivalent, residual, correction, reference period, mass and pressure perform precise jobs. A casual translation that replaces ‘equivalent thickness’ with ‘sea level’ or ‘uncertainty’ with ‘error’ can change the scientific claim. Before drawing a conclusion, ask what change to the selected baseline, coastal mask or processing release would alter the result. The answer shows whether the observation is robust and which assumptions remain visible.
The English phrase ‘monthly ocean bottom-pressure anomaly, in centimetres EWT relative to the chosen mean’ should retain all four qualifiers in another language. Dropping ‘monthly’ or ‘relative’ erases key conditions. In this worked example the figures are invented to demonstrate a calculation or a diagnostic distinction; they are not a reported GRACE measurement. Preserve units, signs and denominators so the reasoning can be reproduced without access to an illustration.
Ask a reviewer to reconstruct the observation’s unit, reference frame, algorithm and limitations from the translated passage. Preserve the capacity to verify the conclusion, not merely a smooth sentence. Where the source does not disclose a needed parameter, mark the inference as unresolved rather than inventing a correction or claiming unwarranted precision. A stronger report can be written after the relevant source field is verified.
Independent transfer questions and fully checked answers
Convert +2.4 cm EWT into mass per area at 1000 kg/m³.
2.4 cm equals 0.024 m. Multiply by 1000 kg/m³ to obtain 24 kg/m². It is an equivalent mass anomaly, not automatically 2.4 cm of observed surface height. For independent checking, write the variable you started with, the transformation applied, and the resulting unit. Decide whether the result is an observation, a product-derived estimate, or only a classroom illustration. State the additional evidence needed before attributing a regional pattern to a particular cause.
Convert a +5 kg/m² mass anomaly to approximate pressure.
Multiply by 9.81 m/s². The pressure equivalent is +49.05 Pa. At reference density 1000 kg/m³, the same mass is 0.5 cm EWT. For independent checking, write the variable you started with, the transformation applied, and the resulting unit. Decide whether the result is an observation, a product-derived estimate, or only a classroom illustration. State the additional evidence needed before attributing a regional pattern to a particular cause.
A map is −1.2 cm EWT. Does it indicate negative absolute ocean mass?
No. It indicates a mass-related signal below the product’s reference mean by 1.2 cm equivalent thickness. The ocean itself still has a large positive absolute mass. For independent checking, write the variable you started with, the transformation applied, and the resulting unit. Decide whether the result is an observation, a product-derived estimate, or only a classroom illustration. State the additional evidence needed before attributing a regional pattern to a particular cause.
Altimetry gives +4 mm/year and steric change +1 mm/year. What is the illustrative residual mass term?
The simple residual is +3 mm/year if periods, masks, conventions and errors are reconciled. A different GRACE result is a budget difference to investigate, not a value to force into equality. For independent checking, write the variable you started with, the transformation applied, and the resulting unit. Decide whether the result is an observation, a product-derived estimate, or only a classroom illustration. State the additional evidence needed before attributing a regional pattern to a particular cause.
Why should a coastal grid cell be treated cautiously?
Land ice or water-storage gravity signals can leak into ocean values under smoothing and finite resolution. Inspect masks, correction methods and nearby land trends before attributing a pixel entirely to ocean mass. For independent checking, write the variable you started with, the transformation applied, and the resulting unit. Decide whether the result is an observation, a product-derived estimate, or only a classroom illustration. State the additional evidence needed before attributing a regional pattern to a particular cause.
What does a missing monthly file mean?
It means that period lacks a usable value in the chosen record. It is not zero mass anomaly. Mark the gap or document an interpolation method if scientifically appropriate. For independent checking, write the variable you started with, the transformation applied, and the resulting unit. Decide whether the result is an observation, a product-derived estimate, or only a classroom illustration. State the additional evidence needed before attributing a regional pattern to a particular cause.
What happens if a source says its Level-3 product has GAD added back?
The background atmosphere-ocean model contribution is handled as a defined processing operation. Preserve that detail and do not subtract or restore GAD again without checking the data product convention. For independent checking, write the variable you started with, the transformation applied, and the resulting unit. Decide whether the result is an observation, a product-derived estimate, or only a classroom illustration. State the additional evidence needed before attributing a regional pattern to a particular cause.
Could two GRACE processing centres yield different trends?
Yes. Inversion, correction, smoothing, resolution and coastal treatments can differ. Compare release metadata and uncertainty before deciding whether the physical estimates truly disagree. For independent checking, write the variable you started with, the transformation applied, and the resulting unit. Decide whether the result is an observation, a product-derived estimate, or only a classroom illustration. State the additional evidence needed before attributing a regional pattern to a particular cause.
What is the appropriate unit for an EWT trend?
A mass-equivalent trend may be stated as mm EWT/year or cm EWT/year, with reference conventions. A pressure trend uses Pa/year and a surface-mass-density trend kg/m²/year. Never drop the per-year denominator. For independent checking, write the variable you started with, the transformation applied, and the resulting unit. Decide whether the result is an observation, a product-derived estimate, or only a classroom illustration. State the additional evidence needed before attributing a regional pattern to a particular cause.
Why does a one-month peak not prove sea-level acceleration?
An acceleration is change in a trend over time, not a single anomalous month. Seasonal cycles and errors can produce peaks. Estimate a supported long-period trend and its time dependence with appropriate error treatment. For independent checking, write the variable you started with, the transformation applied, and the resulting unit. Decide whether the result is an observation, a product-derived estimate, or only a classroom illustration. State the additional evidence needed before attributing a regional pattern to a particular cause.
Frequently asked questions
Is GRACE measuring pressure directly?
No. It measures gravity-related information from orbit, and subsequent inversion and processing can express mass changes as an ocean bottom-pressure anomaly. The geophysical signal and product definition must be read together. For a scientific report, always retain the processing version, named monthly reference and unit, and link the reader to the primary dataset rather than an uncited screenshot.
Why does the map say centimetres when the topic is mass?
Equivalent water thickness converts mass per area into the thickness of a hypothetical standard-density water layer. This convenient convention makes the map readable, but it should not be confused with actual local water-surface height. For a scientific report, always retain the processing version, named monthly reference and unit, and link the reader to the primary dataset rather than an uncited screenshot.
Can GRACE detect tides and local port flooding?
Monthly gravity products are not suitable substitutes for tide gauges or high-frequency coastal pressure observations. Different observing systems solve different temporal and spatial problems. For a scientific report, always retain the processing version, named monthly reference and unit, and link the reader to the primary dataset rather than an uncited screenshot.
What explains disagreement with satellite altimetry?
Altimetry senses total sea-surface height, including density-driven steric effects. Their budgets also differ in masks, time periods, processing and error structures. Compare mass plus steric contributions with altimetry on matched terms. For a scientific report, always retain the processing version, named monthly reference and unit, and link the reader to the primary dataset rather than an uncited screenshot.
Why do published historical numbers get revised?
Releases may use updated low-degree coefficients, GIA assumptions, gravity inversions, corrections and filtering. Data reprocessing can change an old estimate without changing when the underlying measurement happened. For a scientific report, always retain the processing version, named monthly reference and unit, and link the reader to the primary dataset rather than an uncited screenshot.
What does global mass gain prove about a particular coast?
Very little by itself about exact local flooding. Relative sea level at a coast also depends on ocean dynamics, vertical land motion, tides, storm processes and bathymetry. A global mass term is one component of a larger risk assessment. For a scientific report, always retain the processing version, named monthly reference and unit, and link the reader to the primary dataset rather than an uncited screenshot.
What is a safe translation of a positive monthly anomaly?
Say the inferred ocean bottom-pressure field is above its declared reference mean by a stated water-equivalent amount during the named month, and identify the dataset release. Explain uncertainty and avoid calling it a direct observation of coastline sea-level rise. For a scientific report, always retain the processing version, named monthly reference and unit, and link the reader to the primary dataset rather than an uncited screenshot.
Related verified eduKate learning
For a different ocean height measurement, read the independent eduKate guide on satellite altimetry and geostrophic currents at https://edukatesg.com/2026/10/28/how-to-translate-easily-satellite-altimetry-sea-level-anomaly-dynamic-topography-geostrophic-currents/ . A separate explanation of stratification and density is at https://edukatesg.com/2026/10/30/how-to-translate-easily-ocean-mixed-layer-depth-density-threshold-barrier-layer-stratification/ . For the language of scientific explanation, see https://edukatesg.com/project-type/how-english-works/ and for evidence-based science see https://edukatesg.com/how-science-works/ .
Primary data products and methodological sources
NASA CSR GRACE RL06v04 monthly ocean bottom-pressure anomaly description.
NASA JPL GRACE-FO RL6.3 ocean bottom-pressure anomaly dataset.
NASA JPL PO.DAAC JPL RL06v04 ocean-pressure anomaly product.
NASA GRACE GAD Level-2 background ocean coefficient product.
Leuliette and Miller (2009), closing sea-level rise budget using altimetry, Argo and GRACE.