Sounds intriguing, actually. Swell data is basically time series—periodic, noisy, with a few strong predictors like wind speed and fetch. I’d start with a spectral analysis to see the dominant periods, then fit a regression or a small neural net that takes recent wind, pressure, and tide phase as inputs. Cross‑validate with a rolling window to keep the model fresh. The key is to keep the pipeline tidy: ingest, clean, transform, train, evaluate. Once you have a solid error metric, you can start feeding the predictions into an interactive chart for the surfers. Ready to dig in?

Sounds good. Let’s first gather the wind and tide logs, then run a quick FFT to pick out the dominant periods. After that, we can fit a regression model that predicts swell height from recent wind, pressure, and tidal phase. I’ll set up the pipeline and keep the data clean—no surprises. Ready when you are.