About
Who is behind this
CurveNumber is made by two people. One is a licensed professional engineer who has spent a career doing this work by hand and defending the result in front of reviewers. The other works on the method itself, in the standards community, where the argument about what a curve number can and cannot support is actually had.
Deliberately unfinished
Their names, credentials and biographies are not on this page yet, and that is not an oversight.
- [NAME], [PE LICENCE: STATE AND NUMBER], [ONE PARAGRAPH, WRITTEN AND APPROVED BY HIM]
- [NAME], [ROLE AND COMMITTEE, EXACTLY AS SHE OR HE WISHES IT DESCRIBED], [ONE PARAGRAPH, WRITTEN AND APPROVED BY THEM]
A professional licence and a standards role are somebody's own credentials. Putting either on a marketing page implies that the holder endorses what the page says, and neither of these has approved a word of it yet. A product whose argument is that a claim without an attributable source is not a claim would be a poor advertisement for itself if it started by writing someone else's credentials into its own copy. So the blanks stay until the people they describe fill them in.
Why it exists
A stormwater submission for a small site is rarely lost on the arithmetic. It is lost in the gap between a number and the account of where the number came from, which the engineer then reconstructs from memory, a spreadsheet with no history, and a PDF of a table whose edition nobody wrote down.
Everything in this product follows from taking that seriously. Provenance is not a feature bolted to the side of the calculation; a quantity cannot exist in this engine without it. An assumption is not allowed into the registry without a statement, in computed figures, of what it is worth in the answer and what evidence would displace it. And where nothing honest can be derived, the software says so and stops, because a plausible number in a slot that should be empty is the failure mode that costs an engineer their afternoon and their credibility with a reviewer.
What we will not do
- Claim a citation we do not have. Fifty-four values in the registry say that no external source exists, in those words, because the alternative is implying a reference that is not there.
- Record our own check as an independent one. The transcribed tables in this engine have been checked by nobody, the code says so, a test asserts it, and every report that uses them prints it. That is the one failure a provenance product cannot survive, so it is enforced rather than intended.
- Put a number where a question belongs. An unrated soil map unit is not group B. A pond is not CN 98.
- Let a language model produce a figure. The optional assistant, which ships switched off, may read every value in a record, explain it and argue with it. It may never become one. Every numeral in an answer is checked against the values the record holds before the answer is shown, and an answer that fails is refused rather than trimmed. A product that let a model fill a gap would be the same failure as filling it with a default, with better prose on it.
- Pretend to be further along than we are. This is version 0.8.0, it has never had a paying customer, peak discharge is not implemented and payment is not wired. All of that is on the overview rather than discovered later.
The registry is meant to outlive the product
The defaults registry is published free and is meant to be citable on its own. If this business fails, the useful thing it will have produced is a public list of the assumptions that sit silently inside small site runoff work, each with what it is worth in curve number units and what evidence would displace it. That list does not currently exist anywhere, and it should.
Contact
Write to support@curvenumber.com. It reaches one of the two people above, and there is nobody in between.
If you have the published NEH Part 630 Chapter 10 document to hand and want to settle the antecedent runoff condition row we have flagged as doubtful, that would be a genuinely useful thing to receive, and we would credit you for it.