A pressure rise is not always a water-level rise
A submerged pressure logger measures the weight of the water above it per unit area, plus atmospheric pressure if it is an absolute sensor. A higher reading can mean more water above the logger, denser water at the same depth, higher air pressure, or a logger that has moved lower. To report water-surface elevation, first separate those possibilities.
This guide follows the conversion from raw pressure to a water-surface elevation tied to a datum. Start with the freshwater calculation, then work through an estuary where the same pressure has two plausible explanations. The examples show why a salinity profile, a surveyed sensor elevation and a separate surface observation answer different questions.
Hydrology subject guide · Learning that transfers into life · How X Works hub
Choose your reading route
Route 1 · Chapters 1–3: Build the pressure-to-level calculation
Start with the quantity you need
Start with the sensor, subtract air pressure where appropriate, and carry depth to a datum.
Route 2 · Chapters 4–6: Read an estuary with changing density
Work out how denser water can imitate a rise
Calculate the false rise and decide what a local density observation cannot tell you.
Route 3 · Chapters 7–9: Find the failure with independent evidence
Check a fixed surface with a changing column
Compare the pressure record with a surface record, installation log and timing checks.
Route 4 · Chapters 10–12: Transfer and report the result
Carry uncertainty through the conversion
Use the method under changed conditions and preserve uncertainty.
Full chapter index · Independent worked case · Questions and practical checks
Chapter index
Build the pressure-to-level calculation
Start with the quantity you need
Subtract air pressure in the correct place
Carry the depth from the sensor to the datum
Read an estuary with changing density
Work out how denser water can imitate a rise
Build pressure from layers, not just the bottom reading
Decide what a density profile can resolve
Find the failure with independent evidence
Check a fixed surface with a changing column
Diagnose changes in installation without guessing
Keep timing and dynamic pressure in view
CHAPTER 1 OF 12 · Build the pressure-to-level calculation
1. Start with the quantity you need
A river gauge can report stage, the elevation of the water surface relative to an agreed reference. That is different from depth above the channel bed and different again from depth above a submerged instrument. If a sensor sits a little above the bed, the last two quantities differ even when the bed does not move. Write the quantity on the top of the worksheet before converting a number.
Imagine a sensor mounted on a pier. Its pressure port faces water at a particular elevation. The port does not know where the pier top, the channel bed or the map datum sits. It responds to the pressure around it. A measurement of pressure therefore needs both a physical model and a surveyed connection if the product is to be a surface elevation.
In quiet water of uniform density, the water pressure above atmospheric pressure is density times gravitational acceleration times the vertical depth above the port. In symbols, p = ρgh. The result follows because a column with area one square metre and height h has mass ρh and weighs ρgh newtons. This is a useful first-principles model, not a guarantee that every instantaneous pressure fluctuation is hydrostatic.
Here the target is surface elevation at a fixed station, not water velocity or salt transport. A pressure-derived water level may later feed a discharge calculation, but the gauge does not measure discharge. Errors at this first conversion can travel downstream into calculations that look unrelated unless you keep the observation chain visible.
CHAPTER 2 OF 12 · Build the pressure-to-level calculation
2. Subtract air pressure in the correct place
An absolute logger in open water can measure atmospheric pressure plus the water-column contribution. Suppose it reports 123.10 kPa while a suitable barometer reports 101.52 kPa at the corresponding time. Their difference is 21.58 kPa, or 21,580 Pa. The subtraction is made in pressure units before converting to a depth. The absolute reading of 123.10 kPa is not the water load.
A properly functioning vented instrument references atmospheric pressure through its vent system and normally reports gauge pressure directly. In that case do not subtract atmospheric pressure again. Check the instrument configuration and the saved data description, because an export can already contain a compensation that was not applied in the raw channel.
For a scale check, take the freshwater approximation ρ = 1,000 kg/m³ and g = 9.81 m/s². A 1 kPa mismatch corresponds to about 0.102 m of freshwater head. A barometer that is a little early, a little far away, or at an elevation that requires accounting can therefore matter. Do not assume two pressure records describe the same air pressure merely because they have the same date.
If air pressure actually changes the water system, do not confuse that physical response with the atmospheric term added to an absolute logger reading. Removing air pressure from the instrument signal is not an instruction to remove a real atmospheric influence on the water surface. The two operations answer different questions. Keep the raw records so you can revisit which correction was made.
CHAPTER 3 OF 12 · Build the pressure-to-level calculation
3. Carry the depth from the sensor to the datum
Using the compensated pressure of 21,580 Pa, divide by 1,000 × 9.81. The depth above the pressure port is about 2.200 m. If the port elevation is 0.650 m above the station datum, add it: the surface elevation is about 2.850 m above that datum. This result assumes the water density, local pressure model and port elevation are suitable for the intended precision.
Now use a datum under which the port elevation is −0.350 m. The same depth gives a surface elevation of about +1.850 m. Nothing physical moved. The coordinate origin changed. An elevation difference of exactly one metre can be a reference change, not a water event. This is why a chart labelled simply “level, m” is less useful than a chart naming its datum.
Suppose instead that the datum stays fixed but the sensor is reinstalled 0.060 m lower. At the same surface and density, its depth increases by 0.060 m. If the reduction still uses the old port elevation, the published surface appears to rise by 0.060 m. A pressure and an installation log answer different parts of the problem; neither replaces the other.
Check what point was actually surveyed. A casing cap, a hanging attachment and a pressure port are distinct physical points. If the instrument hangs from a marked cable, record the vertical relationship and whether it has been rechecked. The durable evidence is not just a number for installation depth. It is a path from a named reference to the active pressure port.
CHAPTER 4 OF 12 · Read an estuary with changing density
4. Work out how denser water can imitate a rise
Take a fixed port with exactly 4.000 m of water above it. Under the teaching assumptions ρ = 1,000 kg/m³ and g = 9.81 m/s², the pressure due to the water is 39,240 Pa. Replace the column with denser water at 1,025 kg/m³, leaving the surface and sensor fixed. The water pressure becomes 40,221 Pa. The logger correctly sees a rise of 981 Pa.
If the conversion still assumes freshwater, the second reading gives 40,221 ÷ 9,810 = 4.100 m. The reported surface rises 0.100 m even though the actual surface has not moved. This is not an instrument drift problem: the sensor can be functioning perfectly while the model that converts its reading is wrong for the changing water.
Now reverse the question. Suppose the pressure rises by 981 Pa but you have no density or independent water-level record. A rise of 0.100 m in freshwater would produce the same difference under this model. A range of smaller or larger real changes combined with density change could also fit. The pressure record by itself does not select a unique explanation.
If you use an approximate density, carry it as an explicit assumption and examine the plausible error. For a uniform column the sensitivity is proportional to its depth: the same fractional density mismatch translates into a larger absolute depth error at a deeper port. That means a correction that was negligible in one deployment may matter in another, even for the same station and the same type of sensor.
CHAPTER 5 OF 12 · Read an estuary with changing density
5. Build pressure from layers, not just the bottom reading
An estuary can have fresher water near the surface and denser water below. In hydrostatic conditions the pressure at the port adds the contributions of all layers above it. For a simple two-layer model with thicknesses h₁ and h₂, use p = g(ρ₁h₁ + ρ₂h₂). For a smoothly varying column the equivalent idea is to integrate density with depth.
Take a fictional 4 m column with 3 m at 1,000 kg/m³ and 1 m at 1,025 kg/m³. The mass per unit area is 4,025 kg/m², giving 39,485.25 Pa of pressure under the teaching value of gravity. A freshwater conversion returns 4.025 m. A conversion using only bottom density returns about 3.927 m. In this example it is worse to use the unrepresentative bottom reading than to use the freshwater assumption.
Move the boundary so 2 m of the column is in each layer, keeping both layer densities and the total depth fixed. The mass per area rises to 4,050 kg/m². A freshwater conversion now returns 4.050 m, an apparent increase of 25 mm from the previous freshwater estimate. A bottom density sensor would see the same local value at both times and miss the change in layer thickness.
The depth-average density in the first example is 1,006.25 kg/m³ and in the second it is 1,012.50 kg/m³. Those averages weight density by thickness above the port. They are not simple averages of the two reported values unless the layers have equal thickness. Do not use a surface average, bottom measurement or two-point arithmetic mean as a substitute without checking what column it represents.
CHAPTER 6 OF 12 · Read an estuary with changing density
6. Decide what a density profile can resolve
A profile can help estimate the water load, but it must represent the right place, interval and height range. A profile taken an hour earlier can miss a moving interface. A cast made across the channel may sample a different freshwater layer. A profile ending above the pressure port leaves part of the integration unobserved. The sampling problem is about spatial and temporal support as well as instrument precision.
For the simple teaching model, suppose the total depth is already known from a separate instrument. You can calculate the hydrostatic pressure predicted by the sampled layers and compare it with the compensated logger pressure. The comparison is stronger than choosing a density specifically to force the logger to agree with the surface measurement. It provides an independent expectation from observations with distinct failure modes.
When depth is not independently known, a well-characterised density profile can support a conversion, but keep its vertical reference clear. If the profile is expressed as depth below a moving surface, you may need to solve for the unknown water depth consistently rather than selecting a fixed grid that already assumes it. If the profile and the port are tied to the same vertical datum, the relation can be posed in terms of the unknown surface elevation.
Salinity is not density. Temperature and pressure also matter, and the variables expected by an equation of state must be used consistently. For seawater work, the TEOS-10 references explain the relevant thermodynamic variables and conventions. The rounded densities here illustrate arithmetic. They do not provide a salinity-to-density lookup table for field use.
CHAPTER 7 OF 12 · Find the failure with independent evidence
7. Check a fixed surface with a changing column
Suppose a separate surface gauge says the water-surface elevation is steady over an hour. The pressure-derived freshwater stage rises by 25 mm. A contemporaneous profile suggests the denser lower layer grew from 1 m to 2 m in a 4 m column. As the previous calculation shows, that layer change predicts 25 mm of apparent freshwater rise. The observations support a density explanation, provided the records are aligned and the instrument elevations are stable.
Now imagine the separate surface gauge also rises by 25 mm. It is tempting to conclude the two devices agree, so no density effect exists. But their observation support may differ, or several effects may offset each other. If the profile still predicts a density contribution, compare predicted pressure with measured pressure under the actual surface observation. Investigate the residual instead of treating agreement between final plots as proof.
The independent surface observation might be a correctly referenced staff reading in safe conditions or a suitable non-contact gauge. Neither is automatically exact. Waves, a sloping surface, mismatched timestamps and sensor placement can affect comparisons. State how the two stations relate, what each averages, and whether the observed difference exceeds their joint resolution for the question.
One paired measurement establishes agreement only at that instant. To distinguish a fixed offset from an error that grows as the estuary changes, repeat across different water levels and density structures. A flat difference suggests a different next check from a residual that changes with bottom salinity, interface depth or barometric pressure. Those patterns help choose tests; they do not replace them.
CHAPTER 8 OF 12 · Find the failure with independent evidence
8. Diagnose changes in installation without guessing
A sudden jump just after servicing is a useful clue. Compare the old and new elevations, the orientation of the pressure port, the mounting arrangement and the logger configuration. A step aligned with a service date can be physical or instrumental; a date match does not establish the amount or sign of any correction.
Suppose the independent surface record is unchanged, the water density appears steady and a re-survey finds the port 0.035 m below its previous elevation. At about 1,000 kg/m³, that would increase water pressure by roughly 343 Pa. If the pressure jump is about that size, the evidence supports a port elevation correction. Use the surveyed change and a documented effective time, not the value needed to flatten the final chart.
A gradual discrepancy can be harder. A settling support, fouling or a blocked vent, barometer drift, changes in water density and clock drift can each leave a different signature, but those signatures can overlap. Check calibration, reference elevations and raw channels before fitting a correction. A polynomial trend cannot tell you which of those causes it is approximating.
Keep a simple event ledger. For each event record what changed, when the change became effective, the old and new settings or elevations, and the evidence. If the interval is uncertain, mark it as such. Treat a section between the last verified old setup and the first verified new setup as a bounded uncertainty, not a point where a guessed step must be inserted.
CHAPTER 9 OF 12 · Find the failure with independent evidence
9. Keep timing and dynamic pressure in view
The hydrostatic conversion describes pressure from the weight of water overhead. Flow around an exposed port, waves, rapid accelerations and the instrument’s sampling arrangement can add complications. For wave work, a submerged pressure signal need not translate point-for-point into a surface wave record using the static formula. Use a method suited to the target time scale and installation.
Even when the intended product is a slowly changing stage, sampling matters. An instantaneous logger value and a ten-minute average from a separate gauge need not match if there are waves or fast fluctuations. Decide what quantity the comparison uses: an instant, a centred average, or a value supported over a specified interval. Record enough metadata to reconstruct that choice.
A clock mismatch can create an apparent residual that changes sign. If the water surface is rising, an early timestamp could make one record look low when matched with a later reading. During a fall, the same mismatch can make it look high. Test timing against a reliable reference and compare series on common windows before concluding that the residual is a density signature.
A smoothing filter can reduce visible noise but cannot distinguish sensor motion from water movement or density changes. It can also remove or shift a short feature you care about. Keep an untouched series, state what filtering was used, and re-check the interpretation at the scale where the decision will actually be made.
CHAPTER 10 OF 12 · Transfer and report the result
10. Carry uncertainty through the conversion
Consider a nominal depth of 4 m and an uncertain representative density around 1,010 kg/m³. A small density uncertainty of 3 kg/m³ corresponds to about 4 × 3 ÷ 1,010, or 12 mm of depth uncertainty in a uniform model. A pressure uncertainty of 80 Pa corresponds to roughly 8 mm of depth. An uncertain port elevation of 10 mm adds an uncertainty directly to the surface elevation.
If those three terms were independent standard uncertainties in a suitable local linear model, their root-sum-square would be about 17 mm. That arithmetic is not appropriate merely because three bounds appear in a spreadsheet. Bounds, systematic offsets and standard uncertainties are different. If density error persists through the day, averaging many readings will not shrink it like independent sensor noise.
State separate uncertainties when the model is incomplete. If the layer boundary is not observed, you can report a range for plausible column densities rather than a precise standard uncertainty. If the port elevation might have changed between visits, describe which dates are affected. The aim is to tell a reader what part of the result is a measurement, a conversion, and an assumption.
A practical report can show an uncorrected freshwater-head series next to the converted stage and explain when they diverge. That is not an invitation to pick whichever curve looks best. The difference reveals the contribution of the chosen model and helps a second reader identify whether a published threshold crossing depends on an uncertain correction.
CHAPTER 11 OF 12 · Transfer and report the result
11. Transfer the reasoning to another setting
A shallow freshwater pond has little salinity variation, so you may start with a nearly constant density for a modest precision target. That does not eliminate the need for barometric treatment, a verified sensor elevation and a realistic sensor range. If the pond dries and the port is exposed, the open-water column formula does not give a meaningful negative water depth.
In a well, a pressure logger can be used to estimate a water level in the casing under suitable conditions. A changing pressure due to real aquifer response is not the same as an atmospheric measurement offset in an absolute logger. Interpretation depends on the installation, well hydraulics, and what the water level is intended to represent. Do not transfer an estuary density correction as if it were a groundwater response model.
For a classroom test, imagine moving the same instrument from a 4 m estuary to a 0.5 m tank. A 2.5 percent density difference would create about 12.5 mm of false depth in the tank versus 100 mm at 4 m. Now move the sensor down 30 mm in each setting. That change contributes 30 mm of apparent rise in both settings if the installation record is not updated. The two effects scale differently.
Finally imagine the tank is pressurised above its water surface. Ambient room air pressure may no longer be the relevant pressure above the column. You need the pressure at the actual boundary. The transferable rule is to identify what loads the port, which reference it measures against, and which part of the system a new observation would constrain.
CHAPTER 12 OF 12 · Transfer and report the result
12. Make the next reader able to reproduce the result
Begin the data note with a sentence that names the output: water-surface elevation at the station in a specified datum, based on a named pressure sensor. Include the reference survey, effective port elevation, raw units, timing basis and sampling window. Say whether the sensor is absolute or vented and whether the pressure record supplied to the conversion had already been compensated.
Describe how density was chosen or estimated. If you used a uniform approximation, state it and show the sensitivity relevant to your case. If you integrated a profile, give its sampling position and time, reference, coverage and interpolation choices. If a part of the column was unobserved, say how you handled it and what independent check was available.
As a quick review, calculate one row by hand from raw pressure to the reported elevation. Test what happens if the water gets denser but does not get deeper. Then test what happens if the port moves down but the true surface does not move. Both can make the pressure rise, yet they call for different corrections and different supporting evidence.
The goal is not to make every stage trace look smooth. It is to preserve the difference between an observed pressure, an estimated water depth and a datum-referenced surface elevation. If the evidence does not separate density, height or installation changes, leave the conclusion conditional. That is a useful finding: it tells you what to measure next.
Independent worked case and evidence checks
Try a changed-context calculation. An absolute sensor reports 131.210 kPa while a matched barometer reports 101.190 kPa. The compensated pressure is 30,020 Pa. At 1,020 kg/m³, the inferred depth is approximately 3.000 m. Using freshwater density instead gives about 3.060 m.
If the surveyed port elevation is −0.420 m, the density-aware elevation is about 2.580 m in the station datum. The freshwater-assumption elevation is about 2.640 m. If the port later moves 0.050 m down, the correct reference is −0.470 m after the change; otherwise the reported surface carries a further 50 mm installation bias.
What would you observe independently? A staff reading or appropriate surface gauge tests the surface, a vertical density profile tests the water column, and a port re-survey tests the datum tie. If only raw pressure and time survive, the three causes cannot be separated from the pressure series alone.
Worked comparison
| Fictional column above fixed port | Pressure from water | Depth with freshwater conversion |
|---|---|---|
| 4 m at 1,000 kg/m³ | 39,240 Pa | 4.000 m |
| 3 m at 1,000 plus 1 m at 1,025 kg/m³ | 39,485.25 Pa | 4.025 m |
| 2 m at 1,000 plus 2 m at 1,025 kg/m³ | 39,730.50 Pa | 4.050 m |
| 4 m at 1,025 kg/m³ | 40,221 Pa | 4.100 m |
Questions and practical checks
Does a submerged logger measure the surface directly?
No. It measures pressure at its port. A model estimates depth above the port, and a surveyed port elevation connects that depth to a datum.
Can I use the bottom salinity to correct the whole column?
Only when evidence supports that it represents the column for the required precision. A stratified estuary often requires depth-resolved information.
Should I always subtract a barometer?
Subtract a compatible atmospheric pressure from an uncompensated absolute reading. Do not repeat that subtraction on a functioning vented gauge-pressure channel or a value already compensated.
Why compare to a surface gauge?
It provides an independent observation that can help separate true surface movement from a change in water density or sensor installation, provided timing and references match.
Can a pressure logger determine salinity and water depth at once?
Not from one hydrostatic pressure reading alone. Different depth and density combinations can give the same pressure.
Is the teaching conversion valid for waves or a port exposed to fast flow?
Not automatically. Dynamic effects and the instrument response may matter. Use a method appropriate to the frequency range and installation.
Continue with the connected subjects
Read the water column
Estuarine stratification: check the layers and the depth of each observation
CTD profiles: quality-check temperature, salinity and density
Check related instruments and derived quantities
Pressure records, seiches and water-level sampling
See where stage enters an index-velocity discharge calculation
Tide-gauge readings, vertical datums and storm-surge residuals
Sources and scope
US Geological Survey: Use of Submersible Pressure Transducers in Water-Resources Investigations
TEOS-10: official seawater thermodynamic framework
Gibbs SeaWater toolbox: density functions and input conventions
The numerical cases in this guide are original fictional teaching examples, not measurements from a real site. This is an educational explanation, not a field safety instruction, engineering approval or an official certification. For a real project, use qualified practitioners, current product documentation and site-specific evidence. Sources were checked on 5 November 2026.