Water on a cool surface and water in the bottle are different observations
At dawn, a panel is covered with droplets but its collection bottle is nearly empty. Another panel looks less wet yet has delivered more water. The difference need not mean that one observation is wrong. Formation, retention, drainage and later evaporation occupy different parts of the same water account.
Dew collection connects atmospheric vapour with a surface cool enough for condensation. A collector may also intercept fog or rain, and its material and exposure influence how much water reaches the outlet. This guide builds the explanation from the surface energy balance to a defensible yield statement, with original calculations and changed-context checks.
We will not infer drinking-water suitability, landscape dew deposition or dependable supply from a wet panel. Use supplied observations for the exercises. The aim is to understand what a collector has actually demonstrated before multiplying a small amount into a much larger claim.
Hydrology subject guide · Learning that transfers into life · How X Works hub
Choose your reading route
Route 1: Understand surface cooling · Chapters 1–3
Route 2: Account for collected water · Chapters 4–5
Route 3: Check sources and timing · Chapters 6–8
Route 4: Compare devices honestly · Chapters 9–10
Route 5: Transfer and report · Chapters 11–12
Full chapter index · Independent worked case · Questions and practical checks
Chapter index
Understand surface cooling · Chapters 1–3
Distinguish dew from other routes to a wet surface
Use surface temperature rather than air temperature alone
Follow the energy that must leave during condensation
Account for collected water · Chapters 4–5
Separate formation, retained storage and drainage
Define the area before comparing yields
Check sources and timing · Chapters 6–8
Test source attribution with independent observations
Keep the reading time and missing water visible
Use controls and detection limits without inventing certainty
Compare devices honestly · Chapters 9–10
Compare formation and collection performance separately
Do not turn a collector yield into landscape deposition
Transfer and report · Chapters 11–12
Perform changed-weather and changed-purpose checks
Report a yield another reader can interpret
CHAPTER 1 OF 12
1. Distinguish dew from other routes to a wet surface
Dew forms when water vapour condenses on a sufficiently cool surface. Fog interception collects droplets that were already suspended in air. Rain brings falling drops. These routes can all leave water on a panel, but they do not require the same conditions or represent the same flux.
A familiar cold drink helps establish the distinction. Water appearing outside a cold container can come from ambient vapour, not leakage through its wall. The analogy explains condensation; it does not show how an unpowered outdoor collector becomes cold, or whether all water collected outdoors has the same origin.
A panel exposed during fog and a light shower has several possible inputs. Calling the final bottle volume dew without other evidence assigns a source that the bottle itself does not identify. The observation remains useful as collected water, even when a more specific source partition is unresolved.
Begin a report with the route being tested and the observations available to separate it from alternatives. Surface temperature, humidity, rain detection and fog observations answer different parts of that question. A well-labelled mixed collection can be more informative than a confidently named dew yield based only on an outlet reading.
CHAPTER 2 OF 12
2. Use surface temperature rather than air temperature alone
Condensation depends on conditions at the receiving surface. Air temperature from a nearby sensor is not automatically panel temperature. A surface can exchange radiation with its surroundings and sky while also gaining or losing heat through contact and moving air.
Suppose a fictional air mass has temperature 18°C and a supplied dew-point temperature of 14°C. A panel at 16°C is below air temperature but still above that dew point. Cooling relative to the air alone has not established condensation. A panel at 12°C would meet the temperature part of the liquid-condensation condition under these simplified circumstances.
This example deliberately supplies the dew point rather than asking the reader to derive it from an unverified formula. In a real study, humidity and temperature measurements, their locations and uncertainties matter. A coarse station value can miss the conditions close to the surface.
When temperature differences are small, avoid a binary conclusion unsupported by sensor precision. A nominal panel temperature of 13.9°C and dew point of 14.0°C do not settle the question if their uncertainties are much larger than 0.1°C. The next useful step is better local evidence, not additional decimal places in the subtraction.
CHAPTER 3 OF 12
3. Follow the energy that must leave during condensation
Condensation releases latent heat. Continued dew formation therefore requires the surface system to dispose of that heat while remaining cool enough. Radiative exchange can provide cooling, but heat gained from air, supports or sunlight can oppose it. A passive collector is not free from an energy balance simply because it has no powered refrigerator.
For a deliberately simplified steady interval, let available cooling after all non-latent heat terms be Qavailable. A condensation mass M over area A and duration t requires approximately M × Lv of latent heat removal, where Lv is latent heat per unit mass. Thus M/A cannot exceed Qavailable × t/Lv under that specified balance.
Take a supplied net cooling capacity of 20 W/m² available for condensation for six hours and a rounded Lv of 2.45 MJ/kg. Available energy is 20 × 21,600 = 432,000 J/m². Dividing by 2,450,000 J/kg gives about 0.176 kg/m², equivalent to 0.176 mm liquid water. This is a conditional upper bound, not a yield prediction.
Actual formation can be smaller because vapour delivery, changing weather, start-up cooling or other processes limit it. The calculation also assumes the supplied 20 W/m² already accounts for non-latent gains and losses. Subtracting those terms a second time would double count them; ignoring them when the supplied cooling has not included them would overstate the bound.
CHAPTER 4 OF 12
4. Separate formation, retained storage and drainage
Choose a boundary around the panel and its drainage path. Let C be condensation, F intercepted fog, P caught rain, E re-evaporation, D water drained into the bottle and ΔS the increase in water retained on the panel and gutter. With no other routes, ΔS = C + F + P − E − D.
Rearranging gives C = D + ΔS + E − F − P. A bottle measures D, not every term. If the panel begins dry and ends with droplets, some formed water remains outside the bottle. If the panel loses retained water after sunrise, a later reading may miss part of the night’s formation.
In a fictional rain-free, fog-free interval, drainage is 0.10 mm, retained water increases by 0.06 mm and 0.02 mm re-evaporates. Condensation is 0.18 mm under the stated account. Calling the bottle’s 0.10 mm the formation amount would understate it, while calling 0.18 mm delivered yield would overstate delivery.
Those two labels are both worth keeping. A surface-science question may need formation; a collection-system question may need drained and recoverable water. Neither observation is automatically superior. The right quantity depends on the claim, and the balance shows why the quantities differ.
CHAPTER 5 OF 12
5. Define the area before comparing yields
Reported water per square metre requires an area definition. A tilted panel has an actual surface area and a smaller horizontal footprint. A device may also have a separate radiating area, condensing area and installation footprint. Similar unit labels can hide different denominators.
For a flat fictional panel with surface area 2 m² inclined 30° from horizontal, the horizontal projection is 2 × cos 30°, about 1.732 m². A collected 0.30 litre is 0.15 litre per square metre of panel, or about 0.173 litre per square metre of projected area. Neither number is wrong when its denominator is explicit.
If another device reports per square metre of total installation footprint, its value cannot be ranked fairly against a panel-area yield without reconciling the definitions. Supports, spacing and access may occupy area that does not condense water. A larger numerical yield can partly reflect a smaller denominator.
Do not treat this projection calculation as a correction for fog capture. Wind-driven droplets relate to the area normal to their motion, while radiative exchange depends on view and geometry. Different processes require different spatial descriptions. One convenient projected area does not describe every term in the water and energy balances.
CHAPTER 6 OF 12
6. Test source attribution with independent observations
A useful attribution plan asks which competing input could explain the water. Rain detection addresses falling precipitation. Fog observations address suspended droplets. Surface temperature relative to dew point supports or challenges condensation. None alone measures the complete partition.
Suppose a fictional bottle contains 0.20 litre after a night with no detected rain. That observation removes one possible explanation only within the rain instrument’s detection and placement limits. It does not exclude fog or local splashing. No-rain is narrower than all-water-was-dew.
A paired sheltered or warmed surface can sometimes help investigate condensation, but its exposure is altered too. Covering a panel may change radiative view, airflow and droplet interception together. The difference between the pair should not be assigned uniquely to one mechanism unless the control’s consequences have been examined.
For a supplied classroom experiment, explicitly state that fog, rain and splash are absent if the exercise aims to isolate condensation. For a field result, replace that assumption with evidence and limits. Scientific clarity comes from identifying which alternatives were actually constrained, not from adding a long instrument list after the conclusion.
CHAPTER 7 OF 12
7. Keep the reading time and missing water visible
The time of collection influences what remains observable. Water can form before dawn, drain gradually and re-evaporate as conditions change. A bottle protected from loss and a panel still exposed to the air may preserve different portions of the night’s water.
Imagine a fictional panel forms 0.16 mm, drains 0.09 mm by dawn and retains 0.07 mm. By mid-morning, another 0.03 mm has drained and 0.04 mm has evaporated. A final bottle reading gives 0.12 mm delivered. It does not show the maximum retained store or the original 0.16 mm formation.
A time series of panel mass, drainage and temperature could constrain that sequence more strongly than one morning bottle reading. But adding instruments does not guarantee closure if their clocks, areas or boundaries differ. Align the intervals before subtracting the terms.
For a multi-night study, state whether the panel and bottle are reset or allowed to carry water between intervals. Water draining after midnight may have formed before it. Calendar labels are useful bookkeeping conventions, not physical barriers that force every droplet into one night’s formation account.
CHAPTER 8 OF 12
8. Use controls and detection limits without inventing certainty
Small water amounts are sensitive to measurement handling. A collection vessel can gain moisture from its own exposure or lose it before weighing. A blank or control may help quantify a background contribution, but only when its exposure represents the relevant background process.
In a fictional weighing exercise, a sample cloth gains 0.040 kg while an appropriate control gains 0.006 kg from the specified background. The adjusted water amount is 0.034 kg under that control model. If the control was stored differently or used a different material, subtracting its gain would need further justification.
Now suppose the combined weighing uncertainty is 0.010 kg. The small adjusted signal should not be presented with milligram-level confidence because the balance display has many digits. Instrument resolution, repeatability and the complete measurement procedure can be different limits.
A below-detection result is not proof of no condensation anywhere. It states that the chosen method did not distinguish the amount within its detection rule over its support. Preserve that label when summarising nights. Replacing every non-detection with zero can change averages and apparent collection frequency.
CHAPTER 9 OF 12
9. Compare formation and collection performance separately
Surface properties can influence cooling, droplet retention and removal in different ways. A design that forms more condensate may not deliver the most bottle water. An outlet comparison alone cannot identify which stage produced an improvement.
Use two fictional equal-area devices. Panel A forms 0.24 mm and delivers half, yielding 0.12 mm. Panel B forms 0.18 mm and delivers 80%, yielding 0.144 mm. B has lower formation but higher delivered collection. Calling it a better condenser based solely on the bottle would merge formation and removal performance.
The ETH Zurich research programme explicitly connects radiative cooling with condensate removal. Its particular device and experimental conditions should not become a guarantee for a different foil, climate or scale. The transferable point is the need to examine both cooling and the path by which formed water becomes recoverable.
For an honest comparison, retain matched area definitions, weather intervals and starting stores. Repeat across relevant conditions rather than selecting one favourable night. If devices stand in different exposures, a performance difference may combine design and site effects. Pairing and replication help address that question but do not erase it by name alone.
CHAPTER 10 OF 12
10. Do not turn a collector yield into landscape deposition
A deliberately designed panel is not a representative sample of every surface nearby. Its radiative exposure, material, thermal connection and inclination can differ from soil, leaves or roofs. Collector yield therefore cannot simply be multiplied by catchment area and called measured landscape dew.
Suppose a fictional panel delivers 0.10 mm and a catchment covers 1 km². Multiplying gives 100 m³, but that arithmetic only answers a hypothetical uniform-depth calculation. It does not establish that the landscape condensed or retained that amount. The missing step is representativeness, not conversion from millimetres to metres.
A catchment question also asks where the vapour came from and what happens to deposited water. Vapour condensed within a chosen boundary can have arrived from outside or been recycled from nearby evaporation. Local surface wetting and net external water input need not be identical.
Keep the boundary visible. A panel can demonstrate a useful collection mechanism without establishing groundwater recharge, ecosystem water uptake or new basin supply. Each broader claim requires observations suited to its own pathways and time scale. Modest, correctly bounded findings are more useful than impressive volume expansions unsupported by the sampling design.
CHAPTER 11 OF 12
11. Perform changed-weather and changed-purpose checks
First change a clear night to a cloudy one while keeping the panel identical. The radiative environment changes, so an observed clear-night yield should not be carried forward unchanged. Humidity may also change. A higher humidity does not automatically outweigh every change in heat exchange.
Next increase wind. Moving air can supply vapour but can also exchange heat with a cooled surface; fog interception can change at the same time. The net response depends on conditions and design. A one-direction rule such as more wind always means more dew would conceal those competing effects.
Then change the question from nightly surface condensation to a dependable daily delivery. You now need the distribution of collection across many relevant days, carryover storage and a stated demand. One successful demonstration is not a reliability study. Missing and zero-yield periods belong in the denominator.
Finally change the purpose from volume to water quality. The collection amount provides no assessment of substances deposited from the air or contacted on the device. This guide has not established suitability for drinking. A quality claim would need its own representative sampling and competent evaluation rather than an assumption that condensation guarantees purity.
CHAPTER 12 OF 12
12. Report a yield another reader can interpret
A useful result names the device, condensing area, installation area if relevant, observation window, drainage amount, retained-water treatment and evidence for source classification. Include weather conditions and the uncertainty that most affects the claim. State whether the result describes formation, delivered collection or a mixture of atmospheric inputs.
For example: 0.12 litre per square metre of panel delivered over the stated night, with surface retention measured separately, no rain detected by the described method and fog occurrence unresolved. That statement is less sweeping than pure dew production, but another reader can use it without guessing what was ruled out.
Keep performance and service claims apart. An original calculation can show how much panel area a hypothetical demand would require under a supplied constant yield. It cannot establish that every panel receives identical exposure or that the yield persists through the year.
For retrieval practice, explain why droplets can be abundant while the bottle is nearly empty, why a panel colder than air may still be too warm for dew, and why a high per-area yield can depend on its denominator. Those answers demonstrate a connected model of heat, water and measurement rather than a memorised promise about collecting water from air.
Independent worked case and evidence checks
Independent case: Aisha compares two fictional 1.5 m² panels over the same rain-free, fog-free eight-hour interval. Panel A drains 0.18 litre, retains an additional 0.09 litre and loses a supplied 0.03 litre to re-evaporation. Panel B drains 0.21 litre, retains an additional 0.03 litre and loses a supplied 0.015 litre. Both start with the stated baseline stores and have no other transfers.
For A, condensation is 0.18 + 0.09 + 0.03 = 0.30 litre, or 0.20 mm over panel area. Delivered depth is 0.18/1.5 = 0.12 mm. For B, condensation is 0.255 litre, or 0.17 mm; delivered depth is 0.14 mm. A forms more water, while B delivers more to the bottle.
The delivered fractions are 0.18/0.30 = 60% and 0.21/0.255, about 82%. These fractions describe the supplied interval and account. They are not permanent material constants. A different starting wetness or drainage interval could change them even without changing the panel coating.
Use a rounded latent heat of 2.45 MJ/kg. A’s 0.20 kg/m² condensation requires 0.49 MJ/m² latent heat removal; over eight hours that averages about 17 W/m². This does not mean radiation alone was measured at 17 W/m². It is the latent component implied by the supplied formation amount.
Now change the source evidence. A fog observation indicates that 0.045 litre of B’s total input was independently assigned to intercepted droplets. With the same remaining terms, B’s vapour-condensation amount is 0.255 − 0.045 = 0.210 litre, or 0.14 mm. Bottle collection remains 0.21 litre. Source classification changes without changing the outlet reading.
If the 0.045 litre is not independently known, do not subtract it simply because fog occurred. Fog presence establishes a plausible route, not its amount. Report mixed collection and the unresolved partition. This is an evidence check, not a requirement to force every case into an exact source balance.
Finally suppose the intended service requires 20 litres per day and a fictional planning assumption assigns a constant 0.14 litre/m² daily delivered yield. Division gives approximately 143 m² of panel. That is conditional sizing arithmetic. Seasonal variability, installation footprint, drainage reliability and quality remain outside the calculation. State those missing links before calling the number a feasible supply system.
A second comparison changes the reporting denominator across time rather than area. A fictional ten-night record has delivered depths of 0.20 mm on two nights and no detected collection on eight nights. The mean over successful nights is 0.20 mm; assigning zero to the eight non-detections for an explicitly simplified planning exercise gives an all-night mean of 0.04 mm. Both summaries can be calculated, but they answer different questions. A report about dependable delivery must not present the successful-night average as though it described every night.
If the eight values were missing because nobody measured them, the 0.04 mm average would not be justified either. Missing observations and observed non-detections are different. A study can report the two measured-night results and the incomplete coverage, or use a separately supported model for the missing intervals. Counting unobserved nights as zero may understate yield; omitting all difficult nights may overstate it. Preserve the observation status before choosing the summary statistic.
For a matched-device experiment, alternate positions or use replicated placements where the study design permits. In a fictional two-position example, A collects 0.12 mm at position west while B collects 0.18 mm at east; after a position swap under another matched interval, A collects 0.19 mm at east while B collects 0.13 mm at west. The small illustration suggests a position effect worth testing. It does not establish an exact material advantage because weather and other conditions can still differ between intervals.
An informative learning report would therefore present device, position, weather and interval together, then explain which comparisons are matched. More nights increase evidence only when their conditions and missingness are understood. Ten panels exposed during the same brief fog event are not automatically ten independent samples of an annual climate. The question determines the relevant replication: surface comparison, source attribution, seasonal frequency or dependable delivery. Count evidence at the scale of that question rather than using the largest available number.
Check the final wording once more. A finding that a collector delivered water under the observed conditions is already meaningful. It need not be promoted into a claim that the whole catchment received the same dew depth, that the device will produce the annual average every day, or that the bottle is ready for a particular use. Keeping these boundaries explicit preserves the result for later comparison and makes the next experiment easier to design.
Worked comparison
| Fictional equal-area comparison | Panel A | Panel B |
|---|---|---|
| Condensation without fog input | 0.30 litre | 0.255 litre |
| Delivered water | 0.18 litre | 0.21 litre |
| Formation depth over 1.5 m² | 0.20 mm | 0.17 mm |
| Delivered fraction | 60% | About 82% |
Questions and practical checks
Is all water collected at night dew?
No. Fog, rain, splash or carried-over storage may contribute. The outlet does not identify origin. Use source-specific observations or report mixed collected water with the unresolved alternatives.
Why can a wet panel have a nearly empty bottle?
Water may remain attached to the panel or in the gutter, or may evaporate before reaching the outlet. A bottle reading measures drainage. Formation requires a fuller storage and loss account.
Does colder than air mean cold enough for condensation?
Not necessarily. Compare panel temperature with local dew-point conditions, not just air temperature. Measurement uncertainty also matters when the difference is small.
Can the energy bound predict actual yield?
It limits yield only under its stated heat-balance assumptions. Vapour supply, changing conditions, start-up cooling and retention can reduce delivered water. A bound is not a forecast or an independently measured flux.
Does collected water demonstrate drinking-water quality?
No. This guide evaluates quantities and evidence chains, not suitability for drinking. A separate representative quality assessment is needed for that claim.
Continue with the connected subjects
Distinguish sources and surface conditions
Fog collectors: droplets, wind-normal flux and retained water
Dew point: compare surface conditions with saturation temperature
The cold-can example: condensation outside a container
Follow energy and water accounts
Lake heat storage and compatible energy windows
Budget closure and the limits of source identification
Compare the observation with another measurement route
Choose the quantity first: these companion studies distinguish water lost from a maintained surface, precipitation accumulated in a vessel, and water condensed or intercepted on a collector. Their readings are connected by water and energy accounts, not interchangeable labels.
Evaporation pans: exposure, fetch and coefficient transfer
Precipitation storage gauges: accumulated catch and maintenance
Atmometers: artificial surfaces and reference-crop estimates
Sources and scope
ETH Zurich: the research device, radiative shielding and condensate removal
FAO: water-depth units and the latent-heat conversion used in the teaching calculations
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.