Does Buoy Type Affect Wave Model Skill? Discus vs. Waverider
Published 21-Oct-2026
A natural question when judging wave model skill across a mixed buoy roster: do the models do better or worse depending on what kind of buoy is doing the measuring? This roster has two dominant hull types -- the large 3-meter Discus buoy and the smaller Waverider buoy -- so it's a fair comparison to make.
The numbers
Using wave-height skill (percentage of forecasts within 1 ft of the observed value, across all 4 models, over the trailing 14-day lookback window) averaged per station, then compared across the 20 active buoys with enough recent data to score (13 Discus, 7 Waverider; 4 currently-offline buoys and the roster's one Ionomer Foam buoy excluded):
| Buoy type | Stations | Mean % within 1 ft | Std dev |
|---|---|---|---|
| 3-meter Discus | 13 | 56.1% | 16.6% |
| Waverider Buoy | 7 | 45.3% | 8.1% |
Welch's t-test on the per-station means: t = 1.96, p = 0.066 -- suggestive but not conclusive at this sample size (13 vs. 7 stations).
The gap holds at every lead-time bin, and is largest at short lead times, where models generally have the most skill to lose:
| Lead time | Discus | Waverider |
|---|---|---|
| 0-3 days | 73.4% | 55.6% |
| 3-5 days | 57.3% | 44.5% |
| 5-7 days | 48.8% | 40.9% |
| 7-10 days | 41.4% | 38.5% |
The likely real driver: where these buoys sit, not what kind they are
Looking at where each buoy type actually sits in this roster:
- Almost every Discus buoy is well offshore in open water -- e.g. 46059 sits 357 nm west of San Francisco, 51000 sits 245 nm NE of Honolulu, 41040 sits 470 nm east of Martinique.
- Almost every Waverider buoy sits in a nearshore channel/pass entrance (Southwest Pass, Aransas Pass Channel) or at a small Pacific/Caribbean island (Saipan, Palau, Pohnpei, American Samoa) -- places with complex local bathymetry and sheltering that a 0.25° global wave model structurally can't resolve, regardless of what's measuring the waves there.
In other words, "Waverider vs. Discus" in this particular roster is almost a proxy for "nearshore/complex-coast vs. open-ocean." It looks like the smaller Waverider hull was chosen specifically for those nearshore/channel deployments -- not that the sensor itself is less accurate. The skill gap above is best read as a model-resolution limitation at complex coastal/island sites, not a buoy-instrument limitation.