How to Build a Small Climate Data Exhibit Around One Local Question

TLDR

A trustworthy climate data exhibit project starts with one answerable question, not a collection of alarming facts. Choose one main indicator, name the dataset, explain what the chart measures, and disclose its geography, period, units, baseline, method, access date, and main limitation. Clearly separate observations from reconstructions and projections. Before opening the exhibit, ask someone unfamiliar with it to explain the conclusion and caveat without help from a presenter.

The strongest small exhibits make their evidence chain visible: question, measurement, dataset, graphic, conclusion and limit. A visitor should be able to understand not only what the display says, but also how to inspect the original evidence. That makes a modest annotated chart more useful than a wall crowded with disconnected statistics.

Begin with a question the data can answer

“How is our climate changing?” is too broad for one display. A better question defines the variable, place and period. Examples include: “How has the number of very hot days changed at the nearest long-running weather station?” or “How do projected summer temperatures in this county differ under two emissions scenarios?”

The second question is valid, but it requires more explanation because it concerns modeled futures rather than recorded conditions. Regional projections may also involve climate downscaling methods that translate broad model output into finer geographic information. For a first exhibit, a well-documented observational record is usually easier to explain.

Possible question	Evidence needed	Main caution
How has annual or seasonal temperature changed at a local station?	A quality-controlled station series with a stable variable definition	Station moves, instrument changes, missing values and local land-use changes may need treatment
How has global surface temperature changed?	A global temperature analysis such as NASA GISTEMP	Explain anomalies and avoid presenting a global average as a local thermometer reading
How has Arctic sea ice changed during the satellite era?	A documented satellite-derived extent or area record	State the season, metric and period; extent and area are not interchangeable
What could local summers be like later this century?	A regional projection with named scenario, model method and reference period	A projection is conditional, not an observation or a day-specific weather forecast
Did climate change cause one local flood, fire or heatwave?	Event-specific attribution research plus relevant observations	A photograph or single event record cannot establish attribution by itself

If the question is about a flood, wildfire or heatwave, resist turning the event into standalone proof of climate change. An exhibit can document the event and show relevant long-term trends, but causal attribution requires evidence suited to the event, location and mechanism. State exactly what the supplied analysis establishes rather than moving from “consistent with” to “caused by.”

Choose one evidence pathway

Most small exhibits need one primary indicator. Add a second only when it helps explain the mechanism or tests the same conclusion from a genuinely different angle. Three temperature charts with slightly different styling do not necessarily provide three independent lines of evidence.

The evidence type determines what can be claimed. Instrumental observations come from devices such as weather stations, ocean instruments and satellites. These records differ in length, geographic coverage and processing. For an exhibit about ocean observations, an explanation of how buoys, drifters and floats measure the ocean can help visitors understand that “ocean data” is not one uniform measurement.

Reconstructions estimate conditions before widespread instrumental coverage using climate-sensitive records such as tree rings, corals or ice cores. These are not direct thermometer readings. Each proxy responds to particular environmental factors and must be calibrated and tested. A display using them should explain what tree-ring climate proxies measure or the corresponding method for the proxy selected. Climate records therefore have different time spans and interpretation limits.

Projections are model-based, conditional statements about possible futures. Their results depend on assumptions that can include future emissions, scenario design, model structure and downscaling. NOAA’s Climate Explorer provides access to climate data and scenarios, but projected values still need to be labeled as projections rather than observations or weather forecasts.

Find data by variable, place, period and observing system

Before downloading anything, write a one-sentence data specification: “I need monthly mean air temperature for this station from year A through year B,” or “I need modeled annual days above this threshold for this county under these scenarios.” NOAA’s guide to finding climate data recommends defining practical dimensions such as geography, time period, variable, and instrument or observing system.

For a global temperature example, NASA’s GISTEMP v4 combines NOAA GHCN v4 land-station data with ERSST v5 ocean data. NASA publishes the analysis, data products and methodology. Its temperature index is expressed as an anomaly—a departure from a reference-period average—rather than an absolute global temperature. NASA’s GISTEMP documentation lets visitors inspect that evidence trail.

Do not silently substitute a convenient visualization for the dataset named in the label. If you download values from one agency but reproduce a chart design from another, document which source supplied the numbers, which transformations you performed and whether the definitions match.

Explain anomalies and baselines without jargon

A temperature anomaly is the difference between a measured or estimated temperature and the average for a defined reference period. If a location’s reference-period average for a month were 10°C and the observed value were 11°C, its anomaly would be +1°C. This simple example illustrates the calculation; it is not a reported climate value.

The baseline establishes the chart’s zero line. Changing the baseline generally shifts anomaly values up or down, but it does not manufacture or erase the underlying change through time. Two valid charts can therefore display different anomaly values for the same year if they use different reference periods, spatial coverage or analytical methods.

Put the baseline in the chart subtitle or caption, not only in small-print methods text. Do not copy a number between products until you have checked the baseline, geographic domain, variable definition and version. If comparison is necessary, recalculate both series to a common baseline when the source data and method permit it, and disclose that transformation.

Build a label that can stand on its own

Every main chart or map should answer the following questions close to the graphic:

What variable is shown, and in what units?
What place or geographic domain does it represent?
What dates are covered?
Is it an observation, reconstruction or projection?
What reference period or baseline is used?
Which dataset and version supplied the values?
What processing or model method matters to interpretation?
When were continuously updated data accessed?
What is the most important limitation?
Where can a visitor inspect the original data and method?

An effective conclusion is narrow enough to survive contact with the methods. For example: “This record shows an increase in the selected heat metric over the displayed period” is stronger than “This proves every recent hot day was caused by climate change.” The IPCC assesses multiple evidence streams and concludes that human influence has warmed the atmosphere, ocean and land, but that global conclusion does not remove the need for event- and place-appropriate evidence.

Record an access date for a living dataset. NASA updates GISTEMP over time, so the final plotted value should not be treated as permanently fixed. A useful source line might identify the dataset, version, table or file, original URL, access date and any calculations performed by the exhibit team.

Make uncertainty specific and visible

“There is uncertainty” tells visitors very little. Name its source. Early observations may have sparse geographic coverage. A satellite record may cover only recent decades. A reconstruction may have a range that widens where proxy evidence is limited. A local projection may vary across models and emissions scenarios.

Use a source-provided uncertainty band or range where available, preserving its definition. In the label, explain what the band represents rather than calling it a generic margin of error. Do not invent error bars from visual judgment. If the source does not provide a quantitative range suitable for the chart, describe the relevant limitation in words.

The IPCC separates observed changes, paleoclimate evidence and model-based projections, and uses calibrated confidence language to express assessments of evidence and agreement. A small exhibit does not need to reproduce that entire framework, but it should preserve the basic distinction between what was recorded, what was reconstructed and what is conditional on a modeled future.

Separate observations from projections

Never let a solid line move invisibly from historical data into a modeled future. Use separate panels, a vertical divider or visibly different line styles. Label the transition year and explain the scenario in plain language. A projection describes what a model produces under stated conditions; it is not a prediction of the weather on a particular future date.

NASA’s Climate Time Machine demonstrates that different indicators can have different periods of coverage, including global temperature, sea ice, carbon dioxide, ice sheets, sea level and ocean warming. If you combine indicators, do not imply that all begin on the same date or come from the same observing system.

Design interaction around an evidence question

Interaction should help visitors inspect the claim, not merely press a button. Ask them to locate the baseline, identify where observations end, compare seasons, or find the geographic boundary. NOAA’s Science On a Sphere Explorer includes tools for probing, measuring and plotting environmental data in flat-screen and mobile formats, illustrating how interaction can be tied to evidence rather than decoration.

A low-tech version can work just as well: one printed chart, an annotated methods panel and a QR code leading to the original dataset or methodology page. If the display will be handled frequently or placed in a semi-exposed setting, durable labels produced through a commercial printing service may be practical. Keep the printed object secondary to the source trail, and check each agency asset’s specific credit and reuse conditions before reproducing it.

A practical build sequence

Write one question that names a variable, geography and period.
Choose one primary dataset and read its documentation before designing the chart.
Record the dataset title, version, file or table, URL and access date.
Check units, missing values, baseline, spatial coverage and any quality-control notes.
Create the simplest chart capable of answering the question.
Mark observations, reconstructions and projections distinctly.
Add a short conclusion and one specific limitation.
Create a source trail using a readable URL, QR code or both.
Have an unfamiliar reader interpret the chart without coaching.
Revise any label that the test reader misunderstood.

During the visitor test, ask three questions: “What does this chart measure?”, “What conclusion does it support?” and “What can it not establish?” If the reader cannot answer all three, the problem is usually not a lack of facts. It is an unclear title, hidden definition, missing baseline or overbroad conclusion.

Pre-display checklist

The exhibit asks one answerable climate question.
The primary dataset is named, versioned where possible and linked.
Variable, units, geography and period are visible.
The baseline or reference period is stated.
Observation, reconstruction and projection are correctly labeled.
Any projection names its scenario or conditions.
Uncertainty or limitations are described specifically.
The chart does not use one event as automatic proof of a broad causal claim.
The source access date and exhibit calculations are recorded.
Credits and reuse terms have been checked for every borrowed asset.
A test visitor can understand the claim and caveat without a presenter.

The next step

Start your climate data exhibit project with a blank index card, not design software. Write the question on the front. On the back, list the variable, place, period, dataset and main limitation. If those details fit together coherently, build one annotated chart and test it with a reader. An exhibit earns trust when its evidence can be followed—not when it contains the greatest number of climate facts.