USGS streamgage revision ledger › proposed observation record

30-Day Streamgage Publication-Difference Observation Record

A bounded, source-bound record of what differed between dated captures of USGS daily-mean discharge for the streamgages you name — delivered as a PDF methods record and a row-level CSV, priced at $249 once after a qualified scope. Built for a hydrology data-QA lead at a water utility. This page is an offer test. Nothing on it takes payment, and no paid scope is open today.

A missing or mixed-custody seal stays NOT COMPARABLE; it never becomes evidence that a streamgage changed.

Request the observation-record scope →

What this is, stated plainly

Public streamgage values can move after publication. A value that was provisional in one week can be revised, requalified, or superseded in the next. USGS documents this openly, and the daily-values service will hand you the current version of a record. What the service does not hand you is a preserved version history showing exactly what it returned to us before — paired with what it returned later, canonicalized the same way on both sides, and written up so a colleague can follow the comparison without trusting anyone's memory.

That preserved pairing is the proposed product. We hold dated captures of daily-mean discharge rows for public streamgage site numbers. For one retrospective 30-day window, we compare eligible rows across trustworthy seals, classify each outcome using a fixed precedence, and write out every field a reviewer would need to check our work: both seal dates, both source URLs, both retrieval times, the USGS qualifier strings on each side, the values on each side, the assigned state, the changed axes, and the canonical row hashes we computed. You are not buying the public data. You are buying a bounded comparison packet and a QA handoff artifact that can accompany a ticket, review memo, or model-input change log.

Current archive gate

Paid scopes are closed. They stay closed until the archive holds at least 30 trustworthy daily seals under the definition in the Method section below. The counts here are read from the archive at page render and are not hand-maintained.

Measure	Current value	What it means
Latest trustworthy seal	2026-08-09	The newest custody-clean date in the archive. Nothing after it can be compared.
Rows in trustworthy seals	58,963	Daily-mean observation rows on dates that passed the custody test.
Qualification threshold	30 seals	No paid scope opens before this archive-wide depth exists.
Window coverage floor	25 of 30 dates	A requested retrospective window must retain at least 25 trustworthy daily seals.

If the trustworthy count is below 30, no scope can be qualified, whatever the calendar count says. A high calendar count paired with a low trustworthy count is exactly the situation this gate exists to catch. A legacy date that cannot bind its rows to one receipt set is visible in the excluded count and cannot contribute to a deliverable.

Request the observation-record scope → The form records coarse interest only. It takes no payment and no site numbers.

Why pay when USGS data is free?

Every source row underlying this record is public, and USGS-authored information is generally in the public domain. You can fetch the current version yourself for nothing. The honest reason a buyer might pay is not access. It is the preserved earlier side, repeatable comparison method, and a bounded handoff packet. If those do not matter to your workflow, do not buy this.

1. The earlier captured version is not refetchable from the current response. The daily-values service can return a historical date as USGS serves it now. It does not recreate the response our system received on an earlier collection date. A publication-difference record needs both captured versions.
2. Canonicalization is identical on both sides. A difference is only meaningful if both sides were normalized the same way. The record states the canonical fields and carries a row hash on each side.
3. The artifact is a handoff, not a query result. The PDF states what was compared, which dates were withheld, and why. The CSV carries row-level provenance and uncertainty instead of forcing the next reviewer to reconstruct the method.
4. The scope closes. One 30-day window, one list of 1 to 10 preflighted site numbers, one delivery. It does not become a subscription or indefinite monitoring obligation.

If your team already captures dated versions, binds each capture to a source receipt, excludes incomplete local days, and preserves a QA-ready diff, you already have the asset and should not buy ours.

Who fits

• A hydrology data-QA lead at a water utility who needs to explain why a daily-mean-discharge input differed between two review cycles.
• A team preparing a model-input change log where the reviewer will ask which rows differed, which qualifier changed, and which dates could not be compared.
• A group handing a USGS-derived dataset to another internal group and wanting the handoff to carry a source-bound statement of what was observed.
• A reviewer who needs the absence of an observed difference recorded explicitly rather than inferred from silence.

Who does not fit

• Anyone who needs an official or certified USGS record. This is neither.
• Anyone making a flood, safety, release, emergency, or real-time operations decision. Use the authoritative current source and qualified staff.
• Anyone seeking engineering advice, a compliance determination, a water-rights position, or support for a dispute or settlement.
• Anyone needing forward-looking monitoring or alerts. This proposed product uses one retrospective window after the archive-depth gate passes.
• Anyone needing a version from before our archive began. We cannot produce a captured version we do not hold.
• Anyone whose requested site is authored by a non-USGS agency. Preflight declines it.

Proposed scope after qualification

• Buyer: one hydrology data-QA lead at a water utility.
• Sites: 1 to 10 public, USGS-authored streamgage site numbers, supplied only after coarse-scope review.
• Window: one retrospective 30-day window entirely inside the trustworthy archive.
• Coverage floor: at least 25 trustworthy daily seals inside that window.
• Series: daily-mean discharge only.
• Deliverables: one PDF methods record and one row-level CSV.
• Timing: within three business days after qualified scope and payment.
• Proposed price: $249 once.

Method: what makes a date trustworthy

The custody test

A date is trustworthy only when it has exactly one collection run and the collected_at timestamp matches across its source receipts and parsed rows. If a date carries two runs, or the timestamp on a receipt disagrees with the timestamp on an entity row, the date is a legacy mixed-run date and is excluded. Excluded dates are counted and disclosed. They are not silently dropped and cannot contribute a change.

This matters because a first-write-wins store can retain one run's source receipt while a retry fills a row the first response did not contain. Rather than guess which response backs that later row, the read model refuses the entire date. The collector now refuses a second same-date run before any network fetch or blob write, so future dates cannot acquire that mixed custody shape.

The settling buffer

A daily mean is eligible for comparison only after a conservative two-date settling buffer. USGS daily values follow local standard time, so a date that appears complete in UTC may still represent a partial local day. The read model excludes the trailing two observation dates relative to each collection date. This nationwide rule is deliberately more conservative than guessing a local-day boundary from a run timestamp.

An observation that fails the buffer at either capture is NOT COMPARABLE with an incomplete-day reason. It can never become CHANGED. Losing recent rows is less costly than manufacturing a publication difference from an unfinished daily mean.

Primary states and precedence

Every row receives one primary state. The precedence is fixed and applied in this order, so an inconclusive row cannot be promoted into a stronger claim.

Order	State	Meaning
1	UNKNOWN	The source, receipt, or parsed evidence is unreadable or insufficient to classify.
2	NOT COMPARABLE	One or both sides fail the custody, source, depth, or settling rule. Never evidence of change.
3	ROW APPEARED	The eligible row is present on the later side and absent on the earlier side.
4	ROW WITHDRAWN	The eligible row is present on the earlier side and absent on the later side.
5	CHANGED	The row is eligible and present on both sides, with at least one tracked axis different.
6	NO DIFFERENCE OBSERVED	The row is eligible and present on both sides, with no tracked axis different.

The final phrase is precise: no difference observed, not no change occurred. The record speaks only to the copies our system captured on its stated dates.

CHANGED axes

A CHANGED row carries one or more axes. They are not forced into a mutually exclusive label:

• VALUE — the reported daily-mean-discharge value differs.
• QUALIFIER — the USGS qualifier string differs, including a provisional or approved marker.
• DATA STATE — one side carries a value and the other carries a no-data state.

A row can change on more than one axis at once. The CSV preserves each axis, so a reviewer can distinguish a numeric difference from a qualifier transition or a no-data transition.

What each CSV row carries

Field group	Contents
Identity	Public USGS site number, parameter code, statistic code, and observation date.
Seal dates	The earlier and later trustworthy seal dates used for comparison.
Source receipts	Source URLs and retrieval timestamps for both sides.
Qualifiers	The USGS qualifier strings captured on both sides, including provisional markers.
Values	The daily-mean-discharge value captured on both sides.
State	The single primary state from the precedence table.
Axes	For CHANGED rows, every axis that differs.
Hashes	The canonical row hash computed on each side.

What the hashes attest—and what they do not

The canonical row hashes attest to our captured and canonicalized rows: that the row content reported in the deliverable yields the stated hash under the documented canonicalization. That is the whole claim.

The hashes are not proof of what USGS transmitted, and they are not an official USGS signature. We did not sign on USGS's behalf and USGS did not sign our packet. A hash can faithfully identify our copy even when the capture itself was incomplete, which is why custody, source receipts, and UNKNOWN states remain necessary.

Preflight declines before payment

The page is not checkout. Even after the archive reaches its depth gate, each proposed scope is counted before any invoice exists. We decline when:

• A supplied site number carries a non-USGS agency code.
• The requested 30-day window contains fewer than 25 trustworthy seals.
• Daily-mean-discharge coverage is missing for any requested site. An ineligible site is not silently omitted while full price remains.
• The requested window begins before the trustworthy archive.
• The source receipts or row custody needed for the requested record are inconclusive.

A decline is not evidence about the streamgage. It says our archive cannot support the requested product under its own rules.

Exact workflow

1. Interest. Submit work email, company, approximate gage count, intended QA use, preferred retrospective window, and deadline. Do not send site numbers on the public form.
2. Depth check. Until the archive holds 30 trustworthy seals, the request stays an interest signal and no paid scope opens.
3. Coarse scope. After the depth gate, a human establishes the intended window and confirms the product is a QA observation record, not an official or professional determination.
4. Site preflight. Public site numbers are checked for USGS authorship, daily-mean-discharge coverage, trustworthy seal depth, and window bounds.
5. Decision. An unsuitable scope is declined with its counted reason. Nothing is charged.
6. Qualified order. Only a scope passing every source gate can proceed to payment.
7. Delivery. The PDF methods record and row-level CSV are prepared within three business days after qualified scope and payment.
8. Close. Delivery ends the engagement. No subscription or indefinite watch begins.

Data minimization

The public request form collects work contact details and coarse scope. It accepts no streamgage site numbers, source rows, file upload, credentials, private URL, personal dataset, or system access. Site numbers are public identifiers and are exchanged only after the archive and coarse-scope gates pass.

The proposed analysis needs no customer operational readings and no customer data. The deliverable compares public, USGS-authored material captured by our own collector. It contains the public site identifiers, public observation fields, and our source/custody metadata.

What this record does not do

• It does not promise that any difference will be found. An all-NO DIFFERENCE OBSERVED result is valid.
• It is not an official USGS record and is not a certified record.
• It provides no flood, safety, emergency, or real-time operational guidance.
• It provides no engineering advice or professional judgment about a stream.
• It makes no compliance determination and reaches no water-rights or settlement conclusion.
• It is not indefinite monitoring, alerting, or a subscription.
• It cannot create history from before the trustworthy archive.

Official source and use boundary

The method begins with the source's own caveats:

• Provisional values are subject to revision. See the USGS Provisional Data Statement.
• Approved values may still be revised. The USGS Data Disclaimer states that further analysis and review can produce later revisions.
• The daily-values service supports current and historical dates. The official Daily Values Service details explain the API, output, qualifiers, and efficient query boundary.
• USGS-authored information is generally public domain. The USGS copyright FAQ requests source acknowledgment and warns that separately marked third-party material can differ. This product preflight is limited to USGS-authored site rows.

These links support the source boundary; they do not endorse this product. The source agency remains authoritative for its data.

Evidence we actually have

This section keeps a demand test from impersonating demand. Every count below was read from an owned source on 2026-08-09.

• Revenue ledger: 36,429 total events and 7 payments totaling $1,094. All 7 payments are attributed to blog partnerships. Exact searches for streamgage, stream gage, USGS, discharge revision, and river flow returned 0. Nobody has paid us for this product.
• Search Console: the latest owned snapshot covers 2026-07-16 through 2026-07-29 and contains 110 unique sampled queries. It has 0 matching queries, 0 matching impressions, and 0 matching clicks for streamgage, stream gage, USGS, river flow, or daily mean discharge. That snapshot predates this page, so it is not a verdict on the offer.
• Lead store: 1 generic close-file request exists. Its authenticity and whether it matches this product are both UNKNOWN. It is not counted as streamgage interest.

There are no testimonials because there are no customers for this product. There is no market-size claim and no guarantee. The purpose of the page and form is to make the offer specific enough that a real buyer can accept, reject, or challenge it while the data gate remains visible.

Frequently asked questions

Can I buy this today?

No. Paid scopes stay closed until the archive holds 30 trustworthy seals. The current gate above shows the counted depth.

What if my preferred window has 30 calendar dates but only 24 trustworthy seals?

It is declined before payment. The product needs at least 25 trustworthy seals in that exact window. A calendar date that fails custody contributes no evidence.

What if no values differ?

The qualifying rows receive NO DIFFERENCE OBSERVED. That is a statement about our captures, not proof that no change occurred anywhere or at any time.

Can you compare stage, instantaneous values, water quality, or another statistic?

Not in this scope. It covers daily-mean discharge only. Missing coverage is a decline, not a custom substitution.

Can you compare a date before your archive began?

No. We can only compare trustworthy copies we captured. We do not estimate a missing earlier version.

Is the hash proof that USGS published a value?

No. It identifies our canonical row. The source receipt and retrieval metadata describe our observation, but neither is an official USGS signature.

Does a ROW WITHDRAWN result mean USGS deleted or repudiated a value?

No. It means an eligible identity appeared in our earlier captured response and not in our later captured response. The record does not assign a cause.

Does submitting the form start collection for me?

No. The existing collector continues its ordinary public-source schedule. The form does not schedule work, create a private watch, or open a paid order.

Does buying this start monitoring?

No. If the offer later qualifies, it covers one already-observed retrospective window and one delivery. Nothing renews.

Register interest

If the proposed workflow, price, and strict gates describe something you would use, send the coarse scope. The form records interest only—no payment, site numbers, files, or obligation.

Request the observation-record scope →