A warm lake surface does not tell you how much heat the whole lake stored
A lake can store energy while the weather is warm and release it later. If an evaporation calculation assigns all available energy immediately to vaporising water, it can confuse changing lake heat storage with evaporation. The missing term is not an optional correction: it is part of the energy account being used.
This guide follows temperature observations through water volume, heat-content change, time averaging and an explicitly bounded evaporation estimate. You will see why equal surface temperatures can conceal different heat stores, why mixing can move heat without adding it, and why a small temperature uncertainty can matter greatly over a short interval.
The numerical lakes are fictional. Their constants and exclusions are supplied teaching assumptions, not a complete field method or design recommendation. Read the connected water-budget and atmospheric-flux guides for the different quantities they measure. Here the central question is what happened to energy inside the lake before a residual is called evaporation.
Hydrology subject guide · Learning that transfers into life · How X Works hub
Choose your reading route
Route 1: Define the energy account · Chapters 1–3
Route 2: Calculate heat storage · Chapters 4–5
Route 3: Estimate evaporation carefully · Chapters 6–8
Route 4: Check timing and representation · Chapters 9–10
Route 5: Transfer and report · Chapters 11–12
Full chapter index · Independent worked case · Questions and practical checks
Chapter index
Define the energy account · Chapters 1–3
Choose the lake boundary and interval
Keep temperature, heat content and heat flux distinct
Do not substitute the surface for the whole water column
Calculate heat storage · Chapters 4–5
Weight temperature change by represented volume
Use depth and basin geometry together
Estimate evaporation carefully · Chapters 6–8
Convert stored energy into a comparable flux
Subtract warming storage and add released storage with the right sign
Convert latent heat to evaporated water only after checking the budget
Check timing and representation · Chapters 9–10
Recognise why short intervals amplify inventory error
Treat mixing, inflow and changing volume as different mechanisms
Transfer and report · Chapters 11–12
Use independent evidence to test the residual
Explain what the corrected estimate does and does not show
CHAPTER 1 OF 12
1. Choose the lake boundary and interval
An energy budget applies conservation to a defined water body during a defined period. Energy can enter or leave through radiation, turbulent exchange, flowing water and other relevant boundaries. Some can remain stored in the lake. State whether the account includes bottom sediments, ice, inflows and changing water volume before writing the equation.
For the simplest open-water teaching model, let net incoming energy over an interval equal latent heat used for evaporation, other specified outward energy transfers, and increased lake heat storage. A positive storage change means the water body retained energy. It cannot simultaneously be assigned to an unobserved outward flux without double counting.
Use one interval for all terms. A temperature difference between Monday morning and Thursday afternoon cannot directly correct a radiation total from Monday midnight to Thursday midnight unless their timing is reconciled. Conservation is a strong principle, but incompatible observation windows can make a correct equation into an incorrect calculation.
CHAPTER 2 OF 12
2. Keep temperature, heat content and heat flux distinct
Temperature describes a thermal state. Heat content in a simple water-volume model also depends on the amount of water and a stated heat-capacity approximation. Heat flux describes energy transfer per area per time. Degrees Celsius, joules and watts per square metre are therefore not interchangeable units.
Using a stipulated volumetric heat capacity of 4.18 MJ/(m³·°C), warming 1 m³ of water by 1°C increases its sensible heat content by 4.18 MJ. Warming 100 m³ by the same temperature difference requires a hundred times that energy under the same approximation. Equal temperatures do not imply equal stored energy.
Heat content is usually expressed relative to a reference temperature. For constant water volume, a consistent reference cancels when calculating the change. If volume changes, the reference and water-flow energy terms need compatible handling. Avoid interpreting a changing reference-dependent inventory as newly stored environmental heat without accounting for water entering and leaving.
CHAPTER 3 OF 12
3. Do not substitute the surface for the whole water column
A surface observation can be valuable for surface exchange, but it does not uniquely establish temperatures at depth. A lake with a warm shallow layer over cold water can share its surface temperature with a much more uniformly warm lake. Their volume-weighted heat contents can differ considerably.
Suppose two equal-volume layers are 20°C and 10°C. Their volume-weighted mean is 15°C. A surface reading of 20°C is not the mean for this two-layer model. Applying 20°C to both layers would overstate heat content relative to a common reference.
Schmid and Read’s heat-budget overview connects surface heat exchanges with lake thermal structure. The practical lesson here is to match a measurement to its role. A surface temperature used in an exchange formula and a set of depth-representative temperatures used for heat storage answer different parts of the budget.
CHAPTER 4 OF 12
4. Weight temperature change by represented volume
For a fixed-volume layered model, heat-storage change is ΔU = ρc Σ(VᵢΔTᵢ), using a compatible volumetric heat-capacity approximation ρc. Vᵢ is the water volume represented by layer i. The calculation weights temperature changes by water volume, not merely by the number of sensors.
Let the upper layer contain 10,000 m³ and warm by 2°C, while the lower contains 20,000 m³ and cools by 0.5°C. The weighted volume–temperature change is 20,000 − 10,000 = 10,000 m³·°C. Multiplying by 4.18 MJ/(m³·°C) gives +41,800 MJ, or +41.8 GJ.
An unweighted average of the two temperature changes is +0.75°C. Multiplying that by all 30,000 m³ would give 94.05 GJ, more than twice the correct layered result. The error is not in the thermodynamics; it comes from assigning equal weight to unequal water volumes.
CHAPTER 5 OF 12
5. Use depth and basin geometry together
A temperature profile is a set of observations through depth. To turn it into a volume-weighted inventory, each observation or interpolated layer needs a represented water volume. In a basin that narrows with depth, equal depth intervals need not contain equal water volumes.
A rectangular teaching tank makes depth weighting simple because cross-sectional area is constant. A real basin can have shelves, separate lobes and deep holes. The geometry belongs to the heat-content calculation just as much as the temperature measurements do. More depth sensors cannot repair an incorrect volume model.
Keep the spatial basis of profiles visible. A profile over the deepest point may not automatically represent shallow arms or sheltered bays. If a basin-wide estimate uses several regions, combine their temperatures and volumes consistently. Report uncertainty from spatial representation separately from sensor uncertainty.
CHAPTER 6 OF 12
6. Convert stored energy into a comparable flux
To compare interval heat-storage change with a mean surface-energy flux, divide by the defined surface area and interval duration. A heat-content change is an energy amount; dividing by time gives power, and dividing by area gives power per square metre.
For +41.8 GJ over a 10,000 m² lake during one day, storage change per area is 4.18 MJ/m². Dividing 4.18 million joules by 86,400 seconds gives about +48.4 W/m². This is the interval-mean storage rate under the stated fixed-area approximation, not an instantaneous flux at every time or location.
If the same inventory change occurs over ten days, the mean storage rate is about +4.84 W/m². A storage difference cannot be quoted as a flux without its time window. When surface area changes appreciably, define the area convention and use a budget formulation compatible with the changing geometry.
CHAPTER 7 OF 12
7. Subtract warming storage and add released storage with the right sign
Define A as the interval-mean energy available for latent exchange after all other explicitly represented terms. Then LE = A − S, where S is the positive-into-storage rate. This simplified equation is only as complete as the terms included in A. It does not prove that an unexplained residual is entirely evaporation.
If A = 100 W/m² and S = +30 W/m², latent heat flux is 70 W/m². Ignoring storage would assign 100 W/m² to evaporation and overestimate the latent term. If S = −30 W/m², the lake releases stored energy and latent flux becomes 130 W/m² in this stipulated account.
The sign follows conservation, not the calendar. Warming and cooling periods can differ between depths and locations. Determine the measured inventory change rather than automatically assigning a summer-positive or autumn-negative correction. A convenient seasonal label cannot substitute for compatible observations.
CHAPTER 8 OF 12
8. Convert latent heat to evaporated water only after checking the budget
Using a supplied latent heat of vaporisation of 2.45 MJ/kg for an illustrative calculation, evaporating 1 mm of water from 1 m² requires about 2.45 MJ. One millimetre corresponds to approximately 1 kg/m². Actual applications should use an appropriate thermodynamic treatment rather than treating this rounded teaching constant as exact under all conditions.
A latent flux of 70 W/m² over one day supplies 6.048 MJ/m². Dividing by 2.45 MJ/kg gives about 2.47 kg/m², or 2.47 mm of evaporation. Using 100 W/m² instead would give about 3.53 mm. The omitted +30 W/m² storage term changes the answer by about 1.06 mm for that day.
A negative latent residual deserves investigation rather than automatic clipping. It might be compatible with net condensation under the chosen convention, or it might reveal missing energy terms or inconsistent measurements. State which interpretation is supported. Setting every inconvenient residual to zero can hide a systematic budget problem.
CHAPTER 9 OF 12
9. Recognise why short intervals amplify inventory error
A storage rate is calculated from a difference between two heat inventories divided by elapsed time. If the inventory difference contains uncertainty, shortening the interval makes the same energy error larger when expressed as W/m². High time resolution does not automatically supply high accuracy in a differenced quantity.
Andreasen and colleagues examined daily lake evaporation inferred from longer energy-budget observations and discussed problems associated with short-window heat-storage estimates. Their research motivates a time-scale check, not a universal minimum interval for every lake or instrument arrangement.
Distinguish independent random errors from common or drifting bias. A constant additive temperature bias can cancel from a fixed-volume change if it affects both inventories identically. A bias that changes between surveys does not cancel. A stable-looking temperature record can still yield a noisy or biased heat-storage difference.
CHAPTER 10 OF 12
10. Treat mixing, inflow and changing volume as different mechanisms
Mixing can redistribute heat internally. In an idealised insulated system, blending warm and cold layers changes their temperatures without changing total heat content. A warmer bottom observation during mixing is therefore not proof that heat entered through the lake bed.
Flowing water can carry energy across the lake boundary. If inflow and outflow differ in temperature, their net energy contribution can matter even when water volume remains constant. The heat-budget convention must retain these terms rather than assigning their effect to evaporation or storage error.
Changing volume adds another distinction. Water-level change can alter the amount of water represented by a profile and the surface area used for flux normalisation. Keep the water budget beside the energy budget. A simpler fixed-volume calculation can be a useful teaching fence, but it should not quietly become a complete account of a strongly fluctuating reservoir.
CHAPTER 11 OF 12
11. Use independent evidence to test the residual
Compare an energy-budget evaporation estimate with an appropriately supported independent method or water-budget constraint. Each method has its own uncertainty and spatial support. Agreement is more informative when the methods do not inherit the same dominant assumptions or forcing errors.
A good comparison identifies what would discriminate a storage problem from an evaporation problem. Additional depth observations can test the inventory; a flow-temperature record can test an omitted advective term; an atmospheric flux observation can constrain latent exchange. Collecting more surface temperatures alone may not distinguish these alternatives.
When methods disagree, retain the disagreement and diagnose the chain. Do not adjust heat-storage estimates merely until evaporation matches a preferred reference. That turns a comparison into a fitting exercise and removes its independence. Report the tested interval and conditions rather than claiming an entire season was validated by one convenient day.
CHAPTER 12 OF 12
12. Explain what the corrected estimate does and does not show
A clear result identifies the lake boundary, represented water volumes, temperature-change evidence, area and interval. It states the sign of storage and the remaining budget terms before reporting evaporation. Distinguish direct measurements, modelled terms and residual calculations.
Do not infer a climate trend from one warm surface day or one corrected evaporation estimate. Long-term comparison needs compatible methods, representative records and attention to changing lake geometry or observation practice. A lake can reveal important change without every isolated anomaly being a trend measurement.
For retrieval practice, explain how two equal surface temperatures can accompany different heat contents, how mixing can warm the bottom without adding energy, and why a survey-temperature error can create a larger apparent flux over one day than over ten. Then recalculate after changing depth weights or storage sign. Those are the connections that make the budget useful.
Independent worked case and evidence checks
Work with a fictional fixed-area lake of 10,000 m² containing 30,000 m³ water. Represent it by an upper volume of 10,000 m³ and a lower volume of 20,000 m³. Use 4.18 MJ/(m³·°C) for volumetric heat capacity, constant water properties and no ice. These are supplied simplifications, not a survey design.
At the first observation, upper and lower temperatures are 20°C and 10°C. At the second, one day later, they are 22°C and 9.5°C. The upper layer gains 83.6 GJ and the lower loses 41.8 GJ. Net lake storage increases by 41.8 GJ.
The volume-weighted mean temperature changes from (10,000 × 20 + 20,000 × 10)/30,000 = 13.333°C to 13.667°C. The increase is one-third of a degree. Multiplying 30,000 m³ by one-third of a degree and the stated heat capacity reproduces the 41.8 GJ inventory change.
A surface-only interpretation would apply the upper layer’s 2°C warming to the whole lake, giving 250.8 GJ. That is six times the actual net change in this two-layer example. The surface observation can be correct while its whole-lake expansion is wrong.
Divide the correct +41.8 GJ by 10,000 m² and 86,400 seconds. The mean storage rate is about +48.38 W/m². If independently stipulated available energy after other terms is 120 W/m², the latent residual is 71.62 W/m².
Over a day, that latent term corresponds to about 6.188 MJ/m². Dividing by 2.45 MJ/kg gives approximately 2.53 mm of evaporation. Rounding differences are expected because intermediate rates were rounded. Carry adequate precision internally and report only justified precision in the final educational result.
Ignoring storage would give 120 × 86,400/2.45 million, about 4.23 mm. The excess is roughly 1.71 mm. It is not an unexplained water-loss measurement; it is energy that the supplied temperature inventory shows was retained in the lake.
Reverse the temperature change in a separate day so the lake releases 41.8 GJ while the same 120 W/m² available term is stipulated. Storage rate is −48.38 W/m² and latent residual becomes 168.38 W/m². The latent term can exceed the current available surface-energy term because stored energy is being released.
Now test mixing without external energy change. Start with equal volumes at 20°C and 10°C in an insulated two-layer tank. Complete mixing gives 15°C throughout under constant properties. The top cools and the bottom warms, while total heat content remains unchanged. Measuring only the bottom would misrepresent an internal redistribution as net gain.
Change to the lake’s unequal volumes. Mixing 10,000 m³ at 20°C with 20,000 m³ at 10°C gives approximately 13.333°C throughout. An arithmetic mean of 15°C would not conserve energy because the cold layer contains twice as much water.
Inspect a spatial-representation error. Suppose the deep profile actually represents only the central 60% of the lake volume, while a warm shallow region represents 40%. If their mean temperature changes are 0.2°C and 1.0°C, the whole-volume mean change is 0.52°C. Using the central value everywhere would give 0.2°C and understate the stipulated warming.
For a separate short-window uncertainty exercise, take mean depth as 5 m and suppose the estimated volume-mean temperature difference has an error of +0.1°C. The resulting inventory-change error per surface area is 4.18 × 5 × 0.1 = 2.09 MJ/m². Mean depth here is the volume-to-area ratio, not a claim of uniform basin depth.
Over one day, 2.09 MJ/m² corresponds to about 24.19 W/m². Over ten days, the same difference error corresponds to 2.419 W/m². The longer interval reduces the flux expression of this fixed endpoint error, but it also supplies a longer-period average rather than resolving daily behaviour.
Using the teaching latent heat, 2.09 MJ/m² is about 0.853 mm of evaporation-equivalent energy. That is a total-interval equivalent. It becomes about 0.853 mm/day over one day or 0.0853 mm/day over ten days. Keep amount and rate separate when describing the effect.
Now consider a common +0.2°C bias in both inventories at fixed volume. It cancels from each layer’s temperature change. If the bias changes from +0.2°C initially to +0.3°C finally, an artificial +0.1°C difference remains. Shared calibration and changing calibration are not the same uncertainty model.
Introduce a simplified advective term with equal inflow and outflow volumes of 1,000 m³ during a day, incoming water at 18°C and outgoing water at 16°C. The net sensible-energy input is 4.18 × 1,000 × 2 = 8,360 MJ, or 8.36 GJ. The arbitrary temperature reference cancels because the compared volumes are equal.
Across the 10,000 m² lake, that advective contribution averages about 9.68 W/m² for the day. Omitting it changes the residual available for latent exchange or storage. Its sign should follow the documented boundary convention, not an assumption that every inflow necessarily cools the lake.
Unequal flows need more care. If lake volume changes, an inventory relative to 0°C and one relative to 10°C differ by a term associated with that changing water mass. A complete compatible energy and water account preserves the same physical outcome. Do not compare reference-dependent partial inventories while leaving the corresponding flow terms out.
Test profile timing. Suppose the first survey samples early morning and the second late afternoon. A shallow surface-warming signal can differ from a same-time-of-day comparison. If the required quantity is a daily inventory change, document whether the surveys represent compatible endpoints rather than assigning the whole diel difference to the intended daily budget.
Test a missing deep observation. If the deep layer’s change is unobserved, setting it to zero is an assumption, not a measurement. In the original case, that assumption would miss the −41.8 GJ lower-layer change and double the inferred net gain from 41.8 to 83.6 GJ. An uncertainty statement should identify that unobserved contribution explicitly.
For an independent check, suppose a separate fictional method estimates evaporation at 2.6 mm for the original day. It is close to the energy-budget estimate of about 2.53 mm, but the comparison is informative only after checking uncertainty and support. Two similar point estimates do not prove exact agreement or identify which individual budget terms are correct.
Finally, translate to water volume. An evaporation depth of 2.53 mm over 10,000 m² is approximately 25.3 m³. This connects the energy estimate to a water budget, but still inherits its assumptions. It does not become an independently measured loss merely because the units changed from energy to water.
Add a water-budget cross-check using a separate fictional day. A lake receives 100 m³ inflow, releases 80 m³ through the outlet and loses an estimated 25 m³ by evaporation, with no other boundary pathways. Storage decreases by 5 m³. If a water-level observation suggests storage instead increased by 10 m³, the accounts disagree by 15 m³. The disagreement cannot uniquely be assigned to evaporation: inflow, outlet flow, storage conversion and omitted pathways also need review.
Now retain the modelled 2.53 mm evaporation depth but change the reporting area from 10,000 to 12,000 m² in a separate fixed-area scenario. Evaporation volume becomes approximately 30.36 m³. Do not retain the old 25.3 m³ volume while changing the area label. Depth, area and volume must remain mutually compatible, and a changing real lake requires a time-consistent area treatment.
For independent retrieval, ask which observation would repair each problem. A new thermometer may constrain heat storage but not an unmeasured water diversion. A better area curve can improve water-volume conversion without independently determining atmospheric latent flux. The correct next step follows the particular broken connection.
The final explanation should therefore retain the entire chain: representative temperatures and volumes produced a heat-inventory change; area and time converted it to a rate; a declared energy balance produced a latent residual; an appropriate latent heat converted that residual to a modelled evaporation amount. Each conversion is useful only while its conditions stay attached.
Worked comparison
| Changed condition | Effect in the teaching model | Inference to avoid |
|---|---|---|
| Upper layer warms while larger lower layer cools | Net heat change can be modest | Surface warming equals whole-lake warming |
| Storage rate changes from +30 to −30 W/m² | At A = 100, latent residual changes from 70 to 130 | Storage sign is irrelevant |
| Equal endpoint energy error over 1 or 10 days | Different apparent mean-flux errors | Shorter intervals guarantee better accuracy |
| Warm and cold layers mix in an insulated tank | Temperatures change, total heat does not | Bottom warming proves an external heat source |
| Water inflow and outflow have different temperatures | Advective energy can cross the boundary | Every residual belongs to evaporation |
Questions and practical checks
Can surface temperature determine lake heat storage?
Not uniquely. Heat storage also depends on temperatures throughout representative water volumes and their changes over the relevant interval.
Why subtract a positive storage term?
Energy retained in the lake is not simultaneously available for an outward latent flux in the same budget. Released storage has the opposite sign.
Does lake mixing add heat?
Internal mixing can redistribute heat without adding it. A full budget must distinguish redistribution from exchange across the lake boundary.
Is the latent residual a direct evaporation measurement?
No. It is an estimate conditional on represented terms, observations and assumptions. Independent checks help test its interpretation.
Why can daily estimates be difficult?
Differencing uncertain inventories over a short interval can produce a large apparent storage-rate error. Better time resolution and better accuracy are different goals.
Continue with the connected subjects
Check budgets and atmospheric fluxes
Water-budget closure: distinguish residuals from identified fluxes
Evaporation depth and changing area: keep energy-derived depth compatible with water volume
Bathymetric survey comparisons: the geometry behind volume estimates
Eddy covariance: footprint, energy closure and independent comparison
Follow temperature and lake motion
Stream heat budgets: compare storage with advection and exchange
Lake seiches: motion, observation timing and aliasing
Apply the connected model
Lake-ice breakup: connect seasonal thermal change with a defined observation date
Sources and scope
Schmid and Read, 2022: heat budget of lakes
Andreasen, Rosenberry and Stannard: daily evaporation and longer energy-budget observations
USGS: Lake Mead and Lake Mohave evaporation, including heat-storage effects
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 30 October 2026.