Daily-mean flood screening: Seine at Paris-Austerlitz, France¶
Find the observed record, inspect its usable period, and take a reproducible flood-screening example into your own work. This is a maintainer reproduction; independent hydrologist review is pending.
Open the gauge · Open the recorded analysis
What this snapshot contains¶
| Property | Retained evidence |
|---|---|
| Agency | Eaufrance / SCHAPI (Hub'Eau) |
| Station | hubeau_hydrometrie/F700000103 |
| Observations | 7,542 daily mean discharge values, m³/s |
| Actual period | 2006-01-28 to 2026-09-21 |
| Agency licence | etalab-2.0 |
| Input CSV SHA-256 | 06735536efee67994b0c8f1941aad8c5cf2b721bc4a3cef0788e2df175be9f54 |
The recorded example compares GEV L-moments, LP3, and GEV MLE on complete-year annual maxima. Each uncertainty interval belongs to its named estimator. Record length, extrapolation, fit spread and maxima trend are checked separately. The result's grade applies to this screening question and its tested checks.
Recorded conclusion: Screen the 100-year daily-mean flood at Seine at Paris-Austerlitz, France; identify estimator uncertainty and limitations: 100-year return level, GEV (L-moments) 1951 m3/s (not established).
Daily means can understate instantaneous flood peaks. Stationarity and independence remain assumptions. No channel topology, regulation assessment, or model/gauge catchment comparability was established. This is not a certified design flood or a complete Bulletin 17C analysis.
Keep and reproduce the evidence¶
Download the observations CSV, self-contained report, complete study, and provenance. Use Tools → Open saved study for the complete file; it restores the recorded results without rerunning the analysis. These URLs become available when this adoption build is deployed. For immutable citation, use a named release or the source commit recorded in the provenance, not a mutable deployment URL.
To regenerate all three examples, save the three retained CSVs under their original
names (fish-river-us-cold.csv, kingston-uk-cold.csv, seine-fr-cold.csv) and run
python -m examples.adoption_reference --input-dir retained-csvs --out regenerated.
Set AQUASCOPE_REVISION to the checked-out commit. The script makes no agency or
language-model calls; it regenerates statistics, checks, figures and reports.
See validation scope, saving and sharing, and Python, R and QGIS handoffs.