Sources
Where metocean data comes from
Four families of sources coexist, with opposite strengths. None of them is sufficient on its own.
In-situ measurements
Wave buoys, ADCP current profilers, tide gauges, measurement masts. This is the reference, the only direct observation of the site. It is also the scarcest: a campaign rarely exceeds one or two years and almost always contains gaps.
Satellite observations
Altimetry and scatterometry give global coverage of wind and wave height. Temporal resolution stays coarse and the data degrades as you approach the coast.
Numerical models
Wave and ocean circulation models (WAVEWATCH III, HYCOM) and atmospheric models (WRF) produce fields that are continuous in space and time. They carry biases that must be quantified and then corrected against the available measurements.
Reanalyses
A reanalysis such as ERA5 replays several decades with a constant model and assimilation method. It is the only way to obtain a long, homogeneous series, which is a precondition for estimating any rare event.
An example of reconstructed data: wind from the ERA5 reanalysis and waves from a wave model, combined at a single point in the Iroise Sea during storm Ciarán (November 2023). Wind peaks above 33 m/s on 2 November; waves reach their maximum a few hours later and take four days to subside, long after the wind has dropped.
This is exactly where the work happens: cross-checking these sources, quantifying the bias of each, correcting models against available measurements, filling gaps without manufacturing information. Usable metocean data is always reconstructed data, and that reconstruction has to be documented to be defensible.