The short answer
Recent USGS water data may be provisional and can change after quality review. Record the station, parameter, unit, observation time, approval status and retrieval date; a missing reading is not zero.
The reading is a record, not a verdict
You check a river gauge on Monday, save a value, and return later to find the historical series has changed. That does not automatically mean that either visit was useless. Recent USGS real-time data are often provisional: the agency makes them available before the full review process is complete, and they can be revised.
Start by recording the station name and identifier, the measured parameter, its unit, the observation time and time zone, the value, and its approval status. Also record when you retrieved it. Those details let someone else find the same series and understand whether a later value is a revised version of the same observation or a new observation entirely.
What provisional and approved mean
Provisional means the data are subject to review and revision. USGS says that instrument malfunctions and physical changes at the measurement site can affect values. Its review may use field observations and other quality-control information that were not available when a real-time value was first published.
Approved data have passed the agency’s applicable review, but approval is not a guarantee of perfect accuracy or a promise that every use is appropriate. The USGS also has procedures for revising approved records. If a decision depends on an exact historical figure, use the currently published record, keep its status and access date, and ask the source about its limitations.
Identify the quantity before comparing numbers
Gauge height, or stage, describes water height relative to that station’s reference datum. Discharge describes the volume of water passing a point per unit of time. They answer different questions and do not have interchangeable units. A stage at one station cannot simply be compared with the same number at another station using a different datum.
Use the original station page to check the parameter, unit, datum, timestamp and any qualification. An instantaneous reading is also different from a daily mean. If you compare dates, match the same parameter and time basis; if those differ, explain the difference instead of treating the values as a single trend.
A missing point is not a zero
A chart gap can arise for several reasons, including telemetry interruptions, maintenance or a value withheld while it is assessed. USGS uses flags to communicate some periods when data are unavailable or masked. The gap alone does not tell you which cause applies, or that the river stopped flowing.
Do not fill a missing interval with zero. Check the agency’s station notes and flags, use any stated coverage limitations, and show the gap in your record. If you need a comparison across a missing interval, explain whether a different authoritative source or a documented estimation method can answer the question.
Use a reproducible observation note
A useful note might read: “Station ID: [identifier]; parameter: [stage or discharge]; value and unit: [value]; observed: [date, time and time zone]; status: provisional; retrieved: [date and time]; source URL: [link].” These placeholders are a template, not a real observation.
When returning to the record, compare the same timestamp and parameter. Note the old and new values and their statuses rather than silently replacing one. For a chart, identify the observation window, missing periods and any subsequent revision; a recently refreshed web page does not make an old observation current.
Know when a gauge is not enough
A gauge is a measurement at a particular station. Its reading alone does not describe conditions at every downstream location, forecast how the river will behave, or establish that a road is safe. For decisions involving flooding, consult current National Weather Service forecasts and warnings, local emergency officials and road authorities.
Public Atlas provides a route back to agency data, not an emergency warning service. If a source is stale, unavailable or flagged, check the issuing agency directly and avoid turning the last known value into a claim about conditions now.
A handoff note for a colleague
Give the next reader enough detail to reproduce your lookup: the station ID and link, parameter and unit, datum if the parameter is stage, observation time and zone, value, status, and date of retrieval. Add the series interval and missing periods for a chart. Where a value was revised, preserve both versions and state which you used.
For a decision involving a particular river reach, include the physical location and the reason you selected the gauge. A gauge downstream, upstream or on a tributary may answer a different question. This record is evidence for further review; it does not certify a location as flood-safe.
Before citing a number publicly
Check the underlying station page again immediately before publication. Ensure the timestamp you quote is the observation time, not the time your browser loaded the page. Identify the status, units, datum where relevant, and whether the number is instantaneous or a daily summary. An unlabeled value can mislead even when its arithmetic is correct.
If the figure is from a provisional series, say so in the same passage as the number. Link directly to the agency and avoid describing a later approval or correction as an uninterrupted change in water level. If a source does not provide enough information to support the comparison, leave the comparison out and explain the limitation.
Your next steps
- Record the station, parameter, unit, observation time and status.
- Keep the retrieval date and source link.
- Leave missing values as missing and note any flags.
- For flood decisions, consult official forecasts and local authorities.
Common questions
does provisional mean wrong?
No. Provisional is a status describing the stage of review, not a finding that a value is wrong. It also is not a precision score. An unreviewed number may be useful for timely awareness, while an approved number is usually more suitable for a historical comparison that calls for a reviewed record. The important question is whether the data are suitable for the intended use. A quick personal note, a news report, an engineering calculation and an emergency decision have different needs. For a consequential use, consult the USGS station information and a qualified professional rather than treating a site label as a universal quality guarantee.
why did a chart change but the timestamp did not?
A value can be revised for an existing observation time after additional review. Conversely, a page may have acquired a new observation with a different timestamp. Compare both the observation timestamp and the retrieval date before deciding what changed. If the earlier reading was in a downloaded file, retain that copy with its source and status. Check the same station and parameter in the current record and document the difference. If the source gives a revision note or data flag, record it; do not invent a cause when no explanation is available.
Sources and editorial record
- USGS: Why streamflow data may be revised
- USGS: Provisional data statement
- USGS: How web flags explain data gaps
Prepared with AI assistance from the linked sources. Independent human review pending. “Public Atlas Editorial” is a publication label. First edition: 2026-10-28. Scheduled source review: 2027-01-28. Data examples update with their source snapshots; the guide text is versioned separately.
Changes to this guide
- 2026-10-28 — First AI-assisted edition published after source checks, with human review pending.