PhysWall inverts a closed, non-linear physical law at a single measurement point — and refuses when the inverse is not unique.
A stream gauge read two ways. Move the stage and watch what the two rating curves say — the one fitted before Hurricane Harvey, and the one USGS published after it.
This is a rating-curve comparison, not an inversion of a physical law, and it says so rather than implying otherwise.
⚠ worked out in your browser
Four curves on one gauge is not a defect in the record. It is shifting control — the USGS term for a stage-discharge relationship that has moved, "usually due to erosion or deposition of sediment at the stage measurement site."
And after a flood large enough, "the USGS may resurvey the channel cross section and produce a new rating curve." That is what happened here: the river went through Harvey, the channel moved, and the curve moved with it.
⚠ So the 3.4 ft spread at the flood mark is the finding, not the failure. A tool that hid it behind one curve would be hiding the thing worth knowing. Choosing between them is a question about the river rather than about the arithmetic. And the exponent amplifies it: a 5% stage reading is an 8.1% discharge, which is what an uncertainty budget is for.
Move a control and the answer below updates. There is nothing to press.
The verdict above uses the post-Harvey curve. The pre-Harvey curve is shown only as a percentage difference, and that hides the larger term.
at 10,000 cfs
route A, pre-Harvey crosses at 24.663 ft
route B, post-Harvey crosses at 22.132 ft
window 2.531 ft
inside that window the two routes disagree
on which side of the mark you are:
stage route A route B A B
22.132 8,416 10,000 below below
23.000 8,948 10,645 below ABOVE
24.663 10,000 11,924 below ABOVE
And the uncertainty on this page does not cover it. ±5% on stage becomes ±8.1% on flow through the exponent, and at every stage the gap between the curves is roughly twice that band:
stage route gap ±5% band covered 13.0 605 342 no 20.0 1,320 689 no 23.0 1,697 865 no 41.9 4,958 2,292 no
⚠ The choice of route is the largest term in the answer, and it was the one term not displayed. A 2.5 ft window on a flood mark, behind a single number, is exactly the failure this page was built to expose — and it was in the page. Found by an outside reviewer who ran the arithmetic rather than reading the verdict.
Neither curve is authoritative. The post-Harvey one is fitted to the peaks that follow the scour and is the right one for a reading taken today; the pre-Harvey one says what the same stage used to carry. Reporting one without the other is the error.
Grey dots are annual peaks, 1929–2025; amber are the nine since Harvey. Small green dots are USGS field measurements — a current meter in the water, 95 of them since 1980 — and the dotted green line is fitted to the post-Harvey ones. Three records, two instruments, and the amber and green both sit above the grey curve.
validated: something outside was in a position to move this, and it did. Three outside records touched it and one of them broke a whole version.
The claim is that the channel scoured — that the same stage now carries more water. Both curves are fitted to USGS annual peak discharge, which USGS measured and published: 8 peaks from November 2008 to Harvey, 9 from Harvey on. If the post-Harvey peaks had sat on the pre-Harvey curve, the claim would be wrong.
They do not. The median is ×1.26 more flow at the same stage, and the field-measurement route gives ×1.42 independently. And USGS say the same thing in their own rating remark, which we did not write either.
⚠ The first version of this check failed, and the failure was ours. The pre-Harvey curve was fitted to daily mean discharge against daily mean stage, then compared against instantaneous annual peaks. Those are different quantities — the daily mean stage is lower than the instantaneous peak, so the same coefficient is not the same coefficient. Post-Harvey peaks appeared to fit the old curve better, which is the opposite of scour and would have been a real finding had it been true. Both curves are now fitted to peaks.
And a second, independent route says the same thing. USGS field measurements are a current meter in the water — velocity times cross-section, not derived from any rating. 38 measurements before Harvey, 57 after:
route instrument more flow
at same stage
annual peaks stage + rating ×1.26
field measurements current meter, direct ×1.42
window: Nov 2008 to Harvey, 8 peaks and 11 field
measurements. Deliberately short -- see below.
Two instruments, two records, the same direction and the same order of magnitude. That is what scour looks like.
⚠ An outside reviewer found the second error, and it was the strongest criticism available: “you mixed two rulers.” The window used to start in 1990 and so ran across 27 October 2008, the day the daily record’s baseline dropped 2.50 ft overnight. Both independent records step across that date by the same amount — field measurements +30%, annual peaks +27% — and whether that is a datum re-reference or a real channel change, a curve fitted across it is fitted to two regimes.
The window now starts after the step. It costs points: 8 peaks instead of 27, 11 field measurements instead of 38. The figures fell from ×1.47 and ×1.46 to ×1.26 and ×1.42. The claim survives at a smaller number, which is the only honest place for it to land.
⚠ And there is a fifth kind of gap here, which the four axes do not name. The four are: the machine (ε), the wrong law, the wrong object, and which definition. This is none of them — it is the gap between the instrument and the record kept about it. A datum was re-referenced, or a channel scoured, and whichever it was, the documentation that would settle it does not exist. The same shape turned up at a second station: buoy 46042 went silent on 21 August and does not appear in the maintenance schedule, because the schedule was last updated on 5 August. Nothing is wrong with the instrument record and nothing is wrong with the paperwork; they simply do not meet.
⚠ The 2008 step is undetermined: both records are consistent with a datum re-reference and with a real channel change, and nothing in the data separates them. The window steps around the question rather than answering it, which is the correct move when the evidence does not reach.
What moved it, in order:
1 91 annual peaks broke the first version outright 2 USGS rating 16.0 gave the mechanism: scour 3 95 field measures confirmed it on a second instrument
The first is the one that earns the label. Our opening claim — that post-Harvey peaks would sit on the new curve — came out backwards against USGS’s own peak record, and the reason was a quantity mismatch of ours. Outside data was in a position to take something away, and it took the whole first answer.
⚠ Not alpha: nobody outside this project has run a breaking round against the corrected version, and the current meter is USGS’s — we have held nothing in this channel.
A stage reading alone cannot tell you whether the gauge moved or the creek did. On 27 October 2008 the baseline at this station dropped 2.50 ft in a day — from 3.96 to 1.46 — which looks exactly like a sensor falling over.
It was not. The rating curve fitted before that date still predicts the flow afterwards to within 3%, so the datum was re-referenced deliberately and the discharge record is continuous across the break. One route could not have told you that.
⚠ Harvey is the opposite case. USGS rating 16.0, effective 1 November 2018, carries their own remark: “Extended and changed rating based on scour at low end from Hurricane Harvey”. The exponent barely moved — 1.549 to 1.551, which is 0.1% over eleven years — and the coefficient rose 14%. The channel shape is the same and it now carries more water at the same stage.
⚠ This tool says whether a reading is consistent with the channel it was measured in. It does not forecast anything. For flood warnings, the National Weather Service.
Zero measurements of our own. Every figure on this page was measured and published by USGS or NOAA, and we have held nothing in this channel.
annual peaks USGS 08074000, peak-flow record
91 water years, 1929-2025
rating 16.0 USGS stage-discharge table, effective
1 Nov 2018, with their own remark
field measures USGS field-measurement record
95 current-meter visits since 1980
daily record USGS daily values, 2000-2009
rainfall NOAA GHCN-Daily, station USW00012960
The two curves are ours — log-log fits to their measurements. The measurements are theirs. That distinction is the difference between citing and claiming, and it is why this page carries their rating remark verbatim rather than paraphrasing it.
⚠ This tool is free. Nothing on this page was measured by us. It reads USGS records — public data, measured and published by a public agency — and charging for a reading of a public record would be charging for arithmetic that is not ours.
⚠ The same engine runs all nine tools here. It asks how much a measurement allows you to conclude, and refuses the same way in every field — two roots and it will not choose, one source entered twice, the law does not hold here. How to check that without trusting us →
A checking tool, not professional advice. It tells you what a measurement does and does not support; what to do about that is your decision.