Running a surf forecasting website for almost a decade we’ve learnt a very great deal about accuracy and people’s perception of it. We’ve seen surfers arguing time and time again over which of two forecast websites is more accurate when we know they both show fundamentally the same model data. We’ve changed our logo or page designs only to be told that we’ve ruined the accuracy of the forecast! The key issue for most surfers is, as we’ve already touched on, the gap between what a model forecast is telling you and what you think it’s telling you. Or more simply the gap between the incoming swell and the waves on the beach.
Short term predictions of incoming swell and wind strength and direction. If you compare the forecast to an offshore wave buoy and look at the forecast for the next 24hrs you’ll find an extremely good correlation with actual wave heights most of the time. There are exceptions and you can check against a buoy at any time to see when these do occur. Going out in time the accuracy diminishes somewhat. Check out the graphs below, they compare the short range forecast (the blue line) with the actual measured height, period and wind strength (the grey line).
Note the two measurements of period use a slightly different system (and we’re missing some wind data) but none the less you can see for yourself just how accurate the forecast data itself is. This means that if you can learn how to read model data, how to spot what it’s telling us you can be confident in the predictions for surf you make from it.
The model swell forecast was never intended to tell you exactly what the surf will do as it enters the shallow water at your beach, it’ll appear very inaccurate if this is your expectation, a swell of 6ft doesn’t always make a 6ft wave on the beach – in fact it might more often than not NOT make a 6ft wave on the beach. Working out the likely local conditions from model data is what this website is designed to help you with.
Forecast accuracy will diminish as the forecast looks further forward. You’ll notice fairly regularly that storms that appear 5-10 days out on the forecast get bigger or smaller as they come closer in time, in fact you might sometimes see the model predict storms and swells that don’t happen at all. The longer range the forecast the more likely you are to see these issues.