Evidence checked October 31, 2026. This is explanatory analysis, not a property assessment, engineering design or insurance recommendation. All examples involving the invented town of Estuary Junction, its buildings, storms and numerical results are hypothetical.
1. The date beside the blue line
NOAA's explanation of high-tide flooding, updated September 23, 2026, describes a problem that does not require a spectacular storm: rising relative sea level can combine with local conditions to push water onto streets. That current explanation offers a useful starting point for reading flood maps. The landscape may look familiar while the water level against which it is assessed changes. A map without an explicit time horizon can conceal that distinction. NOAA high-tide flooding explanation
Imagine a town considering a new library expected to serve residents for several decades. On the desk are three maps. One is an effective regulatory map. Another displays a model of today's rainfall flooding. A third shows a coastal scenario for a future year. The library parcel is outside a shaded area on the first map, inside a shallow inundation area on the second, and beside a deeper area on the third. The disagreement is not automatically evidence that somebody made a mistake. The maps may answer different questions.
The town must identify those questions before choosing which picture to trust. Is the displayed boundary tied to river flow, rainfall accumulation, coastal water level or a combination? Does it represent current conditions, historical observations, or a future scenario? Does it show the ground getting wet, a building's floor being reached, or a road becoming unusable? These differences matter more than which map has the most convincing colors.
The central argument of this article is that flood information should travel with its time horizon and physical definition. A useful map is not a promise that everything outside its line is safe. It is a modeled or observed statement about a particular mechanism, threshold and set of conditions. Reading it that way makes the evidence more useful, not less.
2. An annual probability is not an appointment
A US Geological Survey explanation dated June 20, 2019, describes the familiar hundred-year flood as an event with a one-percent chance in a given year under the stated frequency model. It also explains why such floods can occur in successive years. The terminology describes probability, not a schedule that gives a community a century to recover before the next event. USGS recurrence-interval explanation
A useful mental experiment is an annual lottery with a losing ticket that is returned to the container after every draw. Drawing that ticket this year does not remove it next year. This analogy assumes the same probabilities each time and independence between annual outcomes. Real hydrological processes can be more complicated, so the analogy should explain the statistical convention rather than claim that the weather literally behaves like shuffled tickets.
For Estuary Junction, a map labelled one-percent annual chance should therefore not be translated into “unlikely during the library's life” without another calculation. Even a small annual probability accumulates over many years. Under a deliberately simplified constant, independent annual probability of 0.01, the chance of at least one exceedance in thirty years is one minus 0.99 raised to the thirtieth power: about 26 percent. That is an illustration, not an estimate for the proposed library.
It is equally important not to turn the calculation into certainty about damage. Exceeding a modeled water level is not identical to flooding the occupied floor. Flooding the floor is not identical to a particular financial loss. The statistical statement concerns a defined threshold. Each step from that threshold to a human consequence requires further information about terrain, building design, access and vulnerability.
3. Rainfall rarity and flood rarity are different objects
NOAA's precipitation-frequency FAQ emphasizes that a rainfall frequency estimate needs a duration, location and area definition. It also cautions against treating point estimates as directly applicable to every watershed-scale use. These qualifications are essential when a headline calls a storm a thousand-year event: the description may apply to rainfall measured over a particular interval, not to the resulting river flow everywhere downstream. NOAA precipitation-frequency definitions
Consider two invented storms delivering the same total rain to the same drainage basin. One concentrates its rainfall into a short burst over already wet ground. The other spreads it across a longer period when the soil and drainage system have more capacity to receive water. Their totals match, but their runoff timing need not. A flood model must represent how rainfall becomes water moving across land and through channels, rather than simply transferring a rarity label from the rain gauge to the river.
Location matters too. A narrow storm can strike one tributary while largely missing another. At their confluence, the timing of the two flows influences the downstream peak. A map derived from a single uniform design storm may be useful for a specified task, but it should not be presented as a photograph of every storm capable of producing the selected flood probability.
For the library project, the practical question is what input the model actually used. Was it an observed storm, a synthetic rainfall pattern, a statistical flow estimate or a combination of models? Which durations were examined? How were spatial differences represented? The answers reveal whether a disagreement between maps concerns climate, hydrology, hydraulic routing or simply the definition of the event being compared.
A careful report can still use plain language. It might say that a study tests drainage under a specified rainfall scenario rather than claiming that every property experiences an identical hundred-year flood. This is not pedantry. It prevents a label attached to one measurement from silently becoming a claim about an entire landscape.
4. A river model and a street model do not see the same pathways
Water can reach a building from several directions. A river can leave its channel. Rain can accumulate faster than a local drainage system removes it. Coastal water can raise the downstream boundary of that drainage system. An obstruction can redirect flow along a street. Mapping one pathway carefully does not establish that the others have been assessed.
GAO's October 25, 2021 review found limitations in the flood products it examined, including treatment of heavy-rainfall hazards and future climate information. That is a historical finding about the reviewed program, not a claim that every map available in 2026 has the same omissions. The report's live recommendation record also documents subsequent planning action. Its enduring relevance is the need to inspect a product's actual scope. GAO flood-mapping review
In Estuary Junction, suppose a river study includes the main channel and its adjacent floodplain but does not explicitly model every street drain. A separate surface-water study routes intense rainfall across the town. A building outside the first study's inundation area may still be exposed in the second. Neither result needs to be dismissed; they can be complementary pieces of a broader assessment.
The reviewer should ask whether each pathway was included, excluded or approximated. “Not mapped” and “modeled as dry” are not synonyms. Neither are “outside the study area” and “outside flood risk.” A public viewer should make these distinctions visible through its legend, coverage layer or explanatory notes.
The most useful question is therefore not whether the address is in a flood zone in the abstract. It is which flood mechanisms have been examined for that address and which have not. A physical pathway does not cease to exist because a particular regulatory or engineering product was designed for another purpose.
5. Elevation is a measurement, not a backdrop
An inundation map often appears to place water onto a perfectly known ground surface. In reality, the ground representation comes from a measurement and processing chain. The USGS 3D Elevation Program provides elevation models at different resolutions, source data and metadata, along with standards and documented quality information. Those distinctions should remain visible when elevation data become the foundation of a local flood model. USGS elevation standards and specifications
Suppose the library's entrance sits near a shallow modeled flood boundary. A small difference between surveyed threshold elevation and the model's ground elevation could change whether the doorway is shown as reached. That does not mean every map needs a building survey before it can provide useful screening. It means the precision of the decision should match the precision of the underlying measurements.
Horizontal resolution is not the same as vertical accuracy. Dividing a surface into smaller squares does not automatically make every elevation correct. A beautifully detailed raster can preserve a systematic offset or an incorrect representation of a roadside embankment. Conversely, a coarser regional product may be adequate for identifying where more detailed investigation is warranted.
The analyst also needs to know what the surface represents. Tree canopy, building roofs, ground, bridges and water surfaces can require different treatment. An apparent ridge in a dataset might be a real barrier to flow, an elevated structure under which water passes, or an artifact of processing. The hydraulic model's interpretation matters as much as the image on the screen.
For a consequential site decision, the project should identify the elevation source, collection date, reference system and relevant quality checks. Where a specific threshold drives the decision, an appropriately qualified survey may be needed. The purpose is not to demand impossible certainty; it is to avoid treating a convenient background layer as unquestionable ground truth.
6. The vertical reference can move every result
A statement that water reaches an elevation of three metres is incomplete unless the reference is known. Elevation values are measured relative to a datum or another defined reference, not relative to an abstract universal zero. Ground survey, tide information and model outputs must be compatible before their numbers can be subtracted meaningfully.
In a hypothetical error, one contractor supplies an entrance elevation in one reference system while another provides water levels in another. Their spreadsheet calculates a comfortable margin. The arithmetic is flawless, but the subtraction compares unlike quantities. A later conversion could reduce, eliminate or reverse the apparent margin. The failure is not in the addition and subtraction; it is in the meanings of the inputs.
Units create a simpler version of the same problem. Feet and metres should never meet in a calculation without an explicit conversion. Less obvious distinctions include whether a value represents ground elevation, a floor threshold, a water surface or a depth above local ground. A column headed “height” is not enough for reliable technical exchange.
Estuary Junction's proposed data dictionary should therefore give every elevation field a unit, reference system and physical description. The entrance threshold might be recorded separately from the surrounding pavement. A flood depth should identify which ground surface was used. A coastal water-level scenario should state its baseline and conversion before being compared with inland survey data.
This is a modest administrative practice with a substantial physical consequence. A model can be sophisticated, a survey can be accurate, and a final result can still be wrong if their reference frames are mismatched. The public need not master geodesy to ask whether the people producing the map have reconciled those frames and documented the reconciliation.
7. Hydraulic detail changes where the water goes
Once water enters a channel or street network, its movement depends on geometry, storage and resistance. The relevant model is not simply a bathtub filled to a common height. A constriction can produce different upstream and downstream conditions; a storage area can delay water; a raised route can separate areas until overtopping occurs.
The US Army Corps of Engineers' versioned HEC-RAS 6.1 guidance warns against forcing a good calibration with unrealistic parameter values. It also identifies terrain quality, off-channel storage and unusual flood events as matters requiring attention. This article uses those methodological warnings, not the old manual version as a claim about the latest software release. HEC-RAS calibration guidance
For the invented town, a road embankment divides a low area into two compartments connected by a culvert. If a model erases the embankment, it may let water spread too easily. If it keeps the embankment but omits the culvert, it may trap too much water. Increasing the map's visual resolution does not, by itself, resolve either conceptual error.
A useful review starts with the pathways most likely to control the result. Where does water enter? Where can it leave? Which structures constrain it? Which storage spaces are represented? What happens when a threshold is crossed? These questions are more informative than asking whether the model is simply one-dimensional or two-dimensional, as though the number of dimensions guaranteed correctness.
The method should fit the physical problem and available evidence. A complex model with poorly known inputs can offer an elaborate but uncertain answer. A simpler model can be useful within a narrower scope. Neither deserves automatic trust based on its label alone.
8. Calibration is a test of an explanation
A flood model should be compared with evidence where appropriate observations exist. But fitting one event is not the same as demonstrating that the model explains the system under all relevant conditions. Several combinations of assumptions can sometimes produce similar results at a limited set of observation points.
Imagine that Estuary Junction has three high-water marks from one historical storm. A modeller adjusts the representation until the simulated levels closely match those marks. That is useful. Yet a second event may reveal that the model moved water too quickly, or stored it in the wrong place, while coincidentally matching the first event's peak at those three locations.
The review should therefore distinguish the evidence used to tune the model from evidence used to challenge it afterward. Where independent events are available, they can test whether a successful fit extends beyond the conditions that shaped the parameters. Where observations are sparse, the limitation should be explicit rather than disguised by a precise-looking map.
Observation quality also matters. A water mark can be mislocated, a timestamp uncertain or a measurement affected by local conditions not represented in the model. Treating every observed value as exact can force the model toward an implausible explanation. The appropriate comparison considers uncertainty in observations as well as uncertainty in model inputs.
For the library study, a useful verification note would identify the events compared, the locations assessed, the kinds of disagreement observed and the conditions not tested. It would not merely say “validated” beside a software logo. The question is what aspect of performance was established: peak level, timing, extent, flow or some combination. Those are related but different tests, and a map intended for one decision may need stronger evidence on one dimension than another.
9. The future is a family of conditions
The 2022 US interagency sea-level technical report includes regional and local scenarios, near-term observation-based trajectories and information about changing extreme-water-level probabilities. Its public explanation describes site-specific land motion and other local factors. The lesson for a flood-map reader is that a future year alone does not define the water level: the scenario and local reference matter too. US sea-level report overview
A map labelled 2050 might represent one emissions assumption, one sea-level trajectory, one land-development pattern and one drainage configuration. Another 2050 map might use different assumptions. Their disagreement should be investigated as a difference between conditional projections, not necessarily a contest between one true forecast and one false forecast.
In the library example, officials could examine several plausible combinations rather than choose the most reassuring picture. One scenario might preserve current drainage capacity. Another could test reduced performance during a power interruption. A third might examine a proposed upstream development pattern. These are editorially proposed planning exercises, not claims that every combination is equally likely or that they replace professional design standards.
The decision question becomes: which choices perform acceptably across the conditions the community considers relevant? Raising vulnerable equipment may help under several scenarios. Reserving space for a future adaptation may have value even when its exact timing remains uncertain. By contrast, a design that succeeds only under one optimistic assumption is more sensitive to error in that assumption.
Scenario analysis is most useful when the assumptions remain attached to the output. A future flood line detached from its scenario can acquire unwarranted authority. A carefully labelled family of results can support a more honest decision without pretending that uncertainty has disappeared.
10. Nonstationarity changes the probability question
The familiar annual-probability calculation assumes a stable probability when the same number is used for every year. A changing climate, evolving catchment or altered drainage network can challenge that assumption. The statistical distribution relevant to a long-lived asset may not remain fixed throughout its service life.
A hypothetical calculation illustrates the difference without estimating any real town's risk. Suppose an independent annual exceedance probability is one percent for the first fifteen years and two percent for the next fifteen. The probability of no exceedance throughout the thirty years is 0.99 to the fifteenth power multiplied by 0.98 to the fifteenth power. Subtracting that product from one gives about 36.5 percent. The result exceeds the roughly 26 percent obtained with one percent throughout.
This arithmetic does not establish that probability doubles anywhere, and independence is still an explicit simplifying assumption. Its purpose is to show why the period and assumptions must be stated. An analyst should not present a single thirty-year probability as though it were independent of the model used to describe those thirty years.
For public discussion, a useful report could separate the present-day probability estimate from the method used to examine change. It could show how the conclusion responds to plausible alternatives and identify which decisions are robust across them. It should also avoid implying that a longer record automatically solves a changing-system problem. Historical observations remain valuable, but the question is how they inform the conditions under consideration.
The flood map's time horizon is therefore not an ornamental label. It is part of the mathematical definition of the claim. A planning process that ignores it can be precise about the past while remaining poorly matched to the asset it is about to build.
11. Compound flooding requires a joint question
At an estuary, the sea and the river do not take turns merely because separate reports describe them separately. High coastal water can affect how readily inland water drains. Heavy rainfall can coincide with elevated downstream conditions. The relevant assessment may need to consider their joint behavior rather than selecting a single worst value from each independent study.
The tempting shortcut is to combine the one-percent river event with the one-percent coastal event and call the result a one-percent combined event. That label is not justified without examining dependence and the way the system responds. The joint probability is a separate question. Even under independence, multiplying probabilities describes simultaneous occurrence of the specified events, not necessarily the probability of a particular combined flood level.
Estuary Junction's hypothetical library lies inland from a tidal outlet. A coastal barrier proposal reduces one route of water entry but changes the drainage conditions during a rainstorm. The town should ask whether the analysis includes the conditions under which the outlet is restricted and how stored inland water is removed. The fact that a barrier addresses one hazard does not prove that every interacting pathway improves.
Timing can be as important as peak magnitude. The river's largest flow may arrive before, during or after the highest coastal level. A model that pairs only simultaneous peaks may be conservative for one question but unrealistic for another. A model that assumes no overlap may miss an important stress condition.
The public-facing explanation should therefore state which combinations were examined and why. It need not publish every technical simulation to make clear that the analysis considered interacting pathways. The objective is to understand the system's response, not to win a contest over which individual hazard has the most dramatic label.
12. A defence changes risk without erasing it
A flood defence can substantially alter the pathways reaching a community. It does not make every possible water level, equipment failure or operational condition irrelevant. A map that assumes a defence performs as intended answers a different question from a map examining overtopping, breach or other degraded performance.
For the invented town, suppose a raised embankment protects a low district from a specified river condition. Behind it, the library's entrance appears dry. The planning team still needs to understand rainfall drainage within the protected area, access routes, the consequences of higher water and the assumptions about maintenance. These questions do not negate the defence's benefit. They identify the conditions under which that benefit exists.
Residual risk should not be communicated as a vague reminder that anything is possible. It should be tied to explicit failure modes or exceedance conditions relevant to the design. If a pump is assumed to operate, say so. If the analysis excludes a breach scenario, say so. If emergency access depends on one low crossing, show that dependency separately from the building's floor elevation.
This is especially important when a defence encourages confidence and additional development. The physical protection and the number of people relying on it can change together. An assessment of consequences should examine the population and services behind the defence rather than treating the original construction date as the end of the story.
A responsible public description can acknowledge both achievement and limitation: the project reduces exposure under defined conditions, while planning addresses conditions beyond those assumptions. That statement is more useful than calling the area either protected or unprotected without explaining what either word means in practice.
13. The building is not the whole service
A library can remain physically dry while becoming unavailable. Its access road may flood, power equipment elsewhere may fail, or staff and users may be unable to reach it. A parcel-level inundation result therefore cannot, on its own, establish continuity of service.
Estuary Junction's hypothetical project team could map the facility as a small network. The occupied floor is one node. Electrical equipment, communications, sanitation, delivery access and pedestrian routes are others. The question is which connection fails first under each relevant condition and what practical alternative exists.
This network view changes some design priorities. Protecting a building's most expensive room may not preserve its usefulness if the only accessible entrance is cut off. A backup power source may have limited value if its fuel or maintenance route is unavailable. Conversely, a modest relocation of sensitive equipment could preserve recovery capacity even when temporary closure remains unavoidable.
The model also needs an appropriate temporal measure. Maximum depth says little about whether disruption lasts an hour or several days. Arrival time can matter for moving equipment or closing a facility safely, while duration can influence cleaning and restoration. These quantities should not be inferred from a static flood-extent image unless the underlying analysis supports them.
For public planning, the service question is concrete: what does the institution need to keep doing, for whom, and during which conditions? A flood map contributes physical information to that decision. It cannot define the service obligation by itself. Treating the building as part of a connected system prevents a narrow engineering success from being mistaken for uninterrupted public benefit.
14. Depth is not damage, and damage is not welfare
A modeled depth can be translated into a damage estimate only by adding assumptions about what is exposed and how it responds. The same depth outside two buildings can produce different losses because their thresholds, materials, contents and uses differ. A monetary estimate adds another layer of valuation.
Suppose the invented town compares two sites with similar modeled water depth. One houses easily moved furniture above the likely entry level. The other contains essential electrical equipment in a lower space. A generic building curve may not capture that difference adequately. The reviewer should ask whether the damage model represents the actual asset or merely an average class.
Expected annual loss can summarize a range of possible events, but it is not the loss expected every year. A hypothetical average of ten thousand units of currency could arise from frequent small disruptions, rare large losses or a mixture. Those patterns have different implications for cash reserves, emergency capacity and public tolerance even when the mean is identical.
Distribution also matters. A town-wide benefit can coexist with increased exposure for a smaller group. Avoided property damage does not automatically measure disruption to people with fewer alternatives, lost access to essential services or the difficulty of recovery. A planning report should make clear which consequences it values, which it reports separately and which it cannot quantify reliably.
This does not require abandoning monetary analysis. It requires interpreting the number as one modeled outcome rather than as a complete measure of social value. The flood map, exposure inventory and consequence model each contribute a different piece. Combining them should make those pieces more transparent, not hide their assumptions inside a single benefit figure.
15. A narrow boundary can conceal broad uncertainty
Maps need boundaries to remain readable. Yet a sharp line can imply a level of certainty that the underlying terrain, flow estimate and hydraulic representation do not support. A pixel immediately outside a displayed line is not necessarily meaningfully different from a pixel immediately inside it.
In Estuary Junction, imagine a sensitivity analysis in which modest, plausible changes to uncertain inputs move the modeled edge across the library entrance. The appropriate conclusion is not that the model is worthless. It is that the decision is sensitive near that location and needs further investigation or a design that does not rely on a narrow margin.
Different uncertainties should remain distinguishable. Measurement uncertainty concerns what was observed. Parameter uncertainty concerns values used to represent processes. Model-structure uncertainty concerns which processes and pathways are represented. Scenario uncertainty concerns future conditions. Combining everything into an unexplained error bar can make the output look scientific while leaving users unable to respond intelligently.
A useful public display could show a central estimate alongside a clearly described sensitivity envelope, or provide separate scenario layers rather than a single authoritative edge. Such a display must explain what the range means. It should not label a collection of plausible model runs as a confidence interval unless the statistical construction supports that interpretation.
The decision-maker needs to know which uncertainty can be reduced and which must be managed. A better survey may resolve a local elevation question. It cannot settle every future climate scenario. A refined drainage model may improve understanding of an outlet. It cannot guarantee that maintenance assumptions will hold forever. Distinguishing those cases helps the town spend its next research dollar where evidence can actually change the decision.
16. Updating the map can reveal more than changing water
When a revised map places more land inside a hazard area, several explanations are possible. The estimated hazard may have changed. The terrain data may have improved. A previously omitted pathway may have been added. A new modelling method may produce a different representation. The boundary alone cannot identify the cause.
The invented town could publish a change ledger alongside a revised map. It would distinguish updated observations, altered assumptions, corrected errors and new scenarios. Where practical, the modeller could rerun old and new configurations to isolate important contributions. Not every difference can be decomposed neatly, but a documented explanation is better than presenting the revision as self-explanatory.
This matters for both reassurance and alarm. A smaller shaded area does not automatically mean the physical hazard declined. A larger one does not automatically mean a sudden environmental deterioration occurred. Better knowledge can change a map without changing the landscape at the moment of publication.
Historical versions remain useful if labelled correctly. A newsroom investigating what officials knew during an earlier decision needs the version then available, not only today's replacement. A current planning application needs current applicable information and a clear account of limitations. Those are different uses of the archive.
For Estuary Junction, the library decision record should preserve the map version, study date and scenario actually used. If later information changes the assessment, the town can then examine whether adaptation is needed without rewriting history. Version control supports accountability by making it possible to distinguish an unreasonable original decision from a reasonable decision made under evidence that subsequently changed.
17. A sensitivity exercise can change the investigation
A useful sensitivity analysis changes a defined assumption and observes what happens to the result. It is not a collection of arbitrary extreme cases assembled to make a project look either frightening or safe. The changes should be physically interpretable, documented and relevant to the decision being considered.
Consider an entirely invented screening exercise for the library entrance. The initial model produces a peak water surface of 4.10 metres in a specified reference system, while a preliminary entrance survey gives 4.35 metres in that same system. The apparent difference is 0.25 metres. These numbers are fictional and are not a recommended design margin. They merely make the calculation inspectable.
The team then runs three separate sensitivity tests. A revised inlet representation raises the modeled water surface by 0.08 metres. A different plausible downstream condition raises it by 0.18 metres. A revised surface pathway lowers it by 0.05 metres. Those changes should not automatically be added together: each came from a separate run, and the system may respond nonlinearly when conditions are combined. A combined scenario requires its own simulation or another justified analytical treatment.
The exercise reveals something more useful than a single uncertainty label. The entrance result is especially sensitive to the downstream condition. That finding could justify checking the outlet representation and the evidence supporting its boundary condition before commissioning a broad new survey of unrelated terrain. Sensitivity identifies where an assumption matters; it does not by itself establish that the alternative value is correct or probable.
Suppose the team also examines a second candidate site whose entrance remains outside the modeled water under all three tests. That does not prove the second site is universally safe. Other pathways or scenarios may matter there. It does suggest that the specific decision is less sensitive to the uncertainties explored, a legitimate comparison if the two sites were assessed consistently.
The final report should preserve both the numerical results and the rationale for selecting the tests. Otherwise a reader cannot distinguish a meaningful investigation from a conveniently chosen set of cases. It should also record when the analysis changes the next action: obtain better observations, improve a representation, modify the design, or accept a clearly stated residual uncertainty. A sensitivity study earns its place in a planning process when it informs that next action, rather than merely adding another appendix to an already crowded folder.
18. The next useful question is local and specific
A resident does not need to become a flood modeller to ask useful questions. The most productive questions connect the map to an actual pathway or decision: does the study include rainfall flooding on this road, what elevation defines the entrance, which future period was examined, and what happens when the downstream outlet cannot discharge freely?
For the hypothetical library hearing, officials could organize discussion around four drawings rather than one polished map. The first shows water pathways. The second shows the building and its essential connections. The third compares relevant scenarios and horizons. The fourth identifies unresolved observations or design dependencies. Together they explain why a decision might require more than checking whether a parcel lies inside a colored zone.
The answers should identify where professional judgement enters. A model does not decide the acceptable interruption of a public service. A public meeting does not replace engineering assessment. A useful process connects those roles: specialists explain what the physical evidence supports, while accountable decision-makers explain the service standard and trade-offs they choose.
The ultimate test is whether the map helps the community make a better decision. That might mean choosing another site, changing the design, preserving adaptation options or commissioning a targeted survey. It might also mean proceeding because the assessed risks are manageable under the stated conditions. The evidence should support the choice actually made, not merely decorate a decision already settled.
A flood map is most valuable when its limits can be read alongside its results. Its line has a mechanism, its probability has assumptions, its elevation has a reference, and its projection has a time horizon. Keeping those facts attached to the picture turns a reassuring or frightening image into a usable account of how water could reach the places people depend on.