NOAA tide flooding
Predict whether a tide gauge's daily maximum observed water level will exceed its published minor flood threshold tomorrow - from observations alone.
About this release
122 NOS stations, 2006-2025, 2,440 station-years of verified water levels: will a station's maximum observed water level exceed its own published minor flood threshold tomorrow? NOAA publishes the observations, the verified maxima and the annual counts, so access is not the contribution - the frozen panel, the observations-only feature contract and the leakage gate are.
The release carries two levels, each a complete task instance. The temporal level is the ordinary panel split. The station-disjoint level holds 41 stations out of training entirely, and it is the point of the dataset: it asks whether a model learned something transferable rather than one station's local habits. It scores the same on stations it has never seen.
The numbers are published with their floor. The baseline reaches 0.8633 (temporal) and 0.8690 (station-disjoint); a single unlearned feature - yesterday's maximum against the station's own threshold - already reaches about 0.83 and 0.85. The headroom is narrow, and a reviewer should see that here rather than discover it. Only about 2% of station-days are positive, so a constant answer agrees with the label 96% of the time: score AUC, never accuracy.
Splits
| Split | Rows | Positive labels |
|---|---|---|
| station_disjoint-eval | 29,930 | — |
| station_disjoint-holdout | 29,930 | — |
| station_disjoint-train | 461,580 | — |
| temporal-eval | 89,038 | — |
| temporal-holdout | 88,841 | — |
| temporal-train | 697,373 | — |
Qualification and baseline
PASSED - 32 checks per level, 0 failed, on both full artifacts, on the worker Gate 1.3.0.
one baseline only. No coding agent and no harness comparison was run, and none is implied.
Use this version
from datasets import load_dataset
ds = load_dataset('earino/noaa-tide-flooding', 'station_disjoint', revision='v2026.09')
ds = load_dataset('earino/noaa-tide-flooding', 'temporal', revision='v2026.09')The version links and checksums identify this artifact. Exclude the labelled holdout from the agent's workspace when running an evaluation.
Files and checksums
| Download | Bytes | SHA-256 |
|---|---|---|
| station_disjoint-meta.json | 867 | 752bf619ffdced446e95250d9787150b93224397b4f09d20db60f41e051fc86a |
| station_disjoint-private-holdout.csv | 4,228,360 | 37d772620231806f3aaf7025961d136c91febda3b82b7fe3442180874602c827 |
| station_disjoint-public-eval.csv | 4,229,477 | be50109320c4af5a925d9f34fedeb6905aed675986ddc9f33c59bd9ac0ba1705 |
| station_disjoint-public-train.csv | 64,636,168 | d7d2bff29c71680962f4421d239121e63beac242b600dd0c23c9b77ac046ce3c |
| station_disjoint-quality.json | 3,793 | 6cc116dd9d17576d02b1a5138e863f0f0d28df9a31f4d71fdf2593fe22f067ee |
| temporal-meta.json | 750 | 2f4e3052728c69dd07836d34e3c3cfab80811905d4f974ec0ae018e4a08dacc4 |
| temporal-private-holdout.csv | 12,475,947 | 89db2e79b60a7185db3e5b937e3b2d29fcbceab016e85c789473c8ac2e104d51 |
| temporal-public-eval.csv | 12,507,815 | a2d733a8297294a14f9c64b5f9182a274a954b41dcefd411ca6ef00dc74191d3 |
| temporal-public-train.csv | 97,931,048 | 60854702933b12cc0909f846bf062016830be189297012021340543c1560e339 |
| temporal-quality.json | 3,795 | 2448f84c683ff46168d8151d1d074433ae962f6ac8d78c14a813b0cca317405f |
Artifact fb49f932c7b4ea5ab532c20f7dce59a7d733de925d1e172054ff1d38abe1eeed
Limitations
- A derived research dataset from NOAA CO-OPS public-domain data, labelled as derived - not an official NOAA, NOS or NWS product, and not a flood warning or forecast service.
- Headroom over a single unlearned feature is narrow (+0.0368 and +0.0236). Recorded in measurements.json rather than left to be discovered.
- About 2% of station-days are positive. Accuracy is misleading on this task by a wide margin.
- The holdout is labelled and downloadable. Keep its labels outside the evaluated agent's workspace during an evaluation.
Source and terms
NOAA CO-OPS verified water levels and flood thresholds
NOAA National Ocean Service, Center for Operational Oceanographic Products and Services (CO-OPS), verified water level observations and flood thresholds, https://tidesandcurrents.noaa.gov/, accessed 2026-09-18
Source: LicenseRef-Public-Domain. Compilation: CC0-1.0. Code and documentation: MIT.
Our contribution: the frozen station list, the row selection, the derived binary label, the temporal and station-disjoint partitions, the feature construction and the packaging. Does not cover or relicense NOAA's data.
Suggested citation
NOAA CO-OPS. Verified daily maximum water level observations (product daily_max_min) and minor flood thresholds (station metadata API) for 122 NOS stations, 2006-2025. Retrieved 2026-09-18 from https://api.tidesandcurrents.noaa.gov/. Public Domain. Derived dataset: NOAA tide flooding prediction dataset, version 2026.09.