Choosing climate and weather data is often harder than analyzing it. This practical checklist helps you match a local research question with the right type of record, compare coverage and limitations, document your decisions, and avoid conclusions that the data cannot support.
Overview
A strong local climate data project begins with a clear match between the question and the measurement. A school project about heat in a neighborhood may need daily air temperature from nearby stations. A study of flood risk may require precipitation totals, river observations, elevation, land cover, and records of past extreme events. A question about coastal change may depend on tide-gauge records, satellite observations, and a consistent reference level.
There is no single “best climate dataset.” The most useful source is the one whose variable, location, time span, resolution, uncertainty, and access terms fit the project. Before downloading anything, write down five decisions:
- Place: What area are you studying, and how close must the measurements be?
- Variable: Do you need air temperature, precipitation, sea level, particulate pollution, soil moisture, or event records?
- Time period: Are you examining a single storm, a season, a recent change, or a long historical baseline?
- Time resolution: Will hourly, daily, monthly, or annual observations answer the question?
- Comparison: What baseline, neighboring location, or earlier period will provide context?
For introductory work, a well-documented station record or gridded product is usually easier to explain than a complex model output. For advanced research, combining sources can be valuable, but each source should remain identifiable throughout the analysis. If you use satellite imagery, review the guidance in How to Use Satellite Imagery to Track Changes on Earth before treating an image as a direct measurement.
Checklist by scenario
The table below is a starting point rather than a ranking. Product names, download procedures, and access requirements can change, so confirm the current documentation before beginning.
| Research need | Useful source type | Coverage and resolution | Historical depth and uncertainty | Licensing and access questions |
|---|---|---|---|---|
| Temperature | Local weather-station observations; quality-controlled station archives; gridded analyses for area-wide context | Stations provide point measurements, often at subdaily or daily intervals. Gridded products estimate conditions across cells and may smooth local effects. | Station length varies. Check missing dates, station moves, sensor changes, exposure, and whether values are observed or estimated. | Check download format, attribution requirements, commercial-use terms, and whether an account or API key is required. |
| Precipitation | Rain gauges; radar-derived precipitation; satellite or blended precipitation products | Gauges measure a specific site. Radar and satellite products provide broader coverage but may use estimates, corrections, or modeled values. | Short intense storms can be difficult to represent. Check accumulation period, units, gauge density, terrain effects, and validation notes. | Confirm whether the record can be redistributed and whether the source requires a citation or metadata file. |
| Sea level | Tide-gauge observations; satellite altimetry; coastal elevation and datum data | Tide gauges are local and continuous where available. Satellite measurements cover broad ocean areas but are not a substitute for every coastal site. | Interpretation depends on the vertical reference, instrument history, land movement, tides, and gaps. A local record may not represent an entire shoreline. | Record the vertical datum, processing version, and any restrictions on derived products. |
| Air quality | Regulatory or research monitoring stations; calibrated low-cost sensors; modeled or satellite-based concentration estimates | Monitors provide point observations, while models and satellites offer wider spatial coverage. Time intervals may range from minutes to daily averages. | Check pollutant definition, instrument method, calibration, detection limits, missing values, and whether a reading is preliminary or validated. | Look for rules on republishing measurements, displaying maps, and identifying sensor providers. |
| Soil moisture | In situ probes; satellite soil-moisture products; land-surface or hydrological models | Probes describe a small area and depth. Satellite products cover larger cells and usually represent near-surface conditions rather than the full root zone. | Vegetation, soil type, snow, frozen ground, and surface roughness affect results. Check depth, units, retrieval quality flags, and spatial scale. | Read the product guide before combining measurements from different depths or processing versions. |
| Extreme events | Storm, flood, wildfire, heat, drought, or disaster event databases combined with observations | Coverage may be based on reports, thresholds, satellite detection, damage records, or station observations. Definitions can vary by region. | Recent events may be incomplete or provisional. Check the event definition, threshold, start and end dates, and whether one event can appear as multiple records. | Review attribution rules and whether maps, photographs, or derived event boundaries have separate terms. |
Match the source to the question
For a question such as “Was this summer unusually warm at my school?” begin with the nearest suitable station, then compare the same calendar period across a clearly defined historical baseline. For “How did rainfall vary across the county?” a gridded or multi-station product may be more appropriate than one gauge. For “Did air quality change near a road?” monitor placement, wind direction, averaging period, and pollutant selection matter as much as the numerical values.
When the project concerns a landscape rather than a single point, consider a blended approach: station observations for ground truth, satellite imagery for spatial patterns, and a gridded product for consistent comparison. Explain what each source contributes instead of merging them into one unexplained series.
What to double-check
- Variable definition: “Temperature” might mean an instantaneous reading, daily mean, daily maximum, or a value adjusted by a processing method. “Rainfall” may refer to an accumulation over a particular interval.
- Units and time zones: Confirm degrees, millimeters or inches, local time or coordinated time, and whether timestamps mark the beginning or end of an interval. Keep conversions in a separate, documented step.
- Station metadata: Note latitude, longitude, elevation, surrounding land use, operating dates, sensor changes, and periods of relocation. Distance alone does not prove that two sites are comparable.
- Spatial scale: A grid cell may cover a large area. Avoid presenting a cell average as the exact condition at a particular home, field, or building.
- Completeness and quality flags: Count missing observations and inspect flagged values. Do not silently replace gaps with zeros, averages, or data from another source.
- Baseline choice: State the reference period and why it fits the question. A short comparison can show recent variability, while a longer record may better describe the range of conditions, provided the record is consistent enough.
- Uncertainty: Preserve error ranges, quality flags, confidence information, or product-level uncertainty where available. A precise-looking chart is not automatically a precise result.
- Version and access date: Save the product name, version, URL, download date, search filters, and documentation used. If possible, retain the original file and a cleaned copy.
For calculations, keep enough precision during intermediate steps and round only at the end. The Scientific Notation and Significant Figures guide can help when values span very different scales.
Common mistakes
- Choosing the easiest download instead of the best measurement: A convenient spreadsheet may not match the location, variable, or time scale required.
- Mixing incompatible records: Combining local observations, satellite estimates, and model outputs without explaining their differences can create misleading trends.
- Using a distant station without checking conditions: Elevation, coastlines, urban surfaces, vegetation, and topography can produce different local climates.
- Calling weather a climate trend: A single hot day or storm is an event. To discuss climate, define a longer comparison and describe variability as well as averages.
- Removing inconvenient values: Outliers may be errors, but they may also represent the event your project is studying. Investigate before deleting.
- Ignoring revisions: Preliminary observations and processed products can be corrected. Record whether your analysis uses a provisional or finalized version.
- Failing to preserve the method: A chart without a source, filter, unit, baseline, and processing note cannot be reliably checked or repeated.
When to revisit
Revisit your dataset choice before each new project, especially when a seasonal planning cycle begins, a major weather event prompts comparison, or you expand from one location to a wider region. Also check again when a provider changes its file structure, product version, variable definition, access method, or documentation.
Use this short pre-analysis checklist:
- Write the research question in one sentence.
- Identify the required variable, location, time span, and resolution.
- Compare at least two plausible source types before downloading.
- Read the metadata and record units, definitions, quality flags, and uncertainty.
- Choose and justify a baseline or comparison location.
- Download the original files and save the access date and product version.
- Inspect missing values, outliers, timestamps, and spatial coverage.
- Label every chart with the source, period, units, and processing notes.
A dataset is not trustworthy merely because it comes from a familiar website. It becomes useful when its measurement, scale, limitations, and documentation fit the question. Keep this checklist with your project files, review it when tools change, and update your source notes before starting the next local climate or weather investigation.
