Menu image Banner image Banner image

Frequently Asked Questions

What are computer models?  A mathematical lattice of data is built with several dozen layers vertically and a grid horizontally whose spacing is stated as the resolution, like 1.3 km. Under it is a model of the terrain like heights and water versus land. The model consists of formulas which take the current temperature, humidity, motion, etc at each lattice point and compute new values for a few seconds later for each point as affected by the lattice points around it. Updating all the lattice points is one computational iteration. Each forecast run starts with the weather as it is reported and iterates the model over and over to estimate the future, looking ahead hours or days.

How accurate are the forecasts?  They are quite good but being forecasts they are not perfect. It is even hard to measure how good because of the Salish terrain complexity and the natural variability of winds. Comparisons at buoy 46088 in the middle of the Strait of Juan de Fuca are fairly stable, but there are few such samples out on the water. Most stations are on land where winds can be locally sheltered or accentuated.

With this App you can judge for yourself: the Lines and History screens have comparisons to measured winds and the Grids show detail within each microclimate.

Do some stations consistently differ?  As you use the information you will notice that the available stations behind Whidbey Island are sheltered and often report lighter winds than the computer models. On the other hand stations at "corners" like West Point, Alki Point or Race Rocks will sometimes accelerate close in to bump up over forecasts.

Why the UW models?  The computer forecasts by the University of Washington at 1.3km steps are among the highest resolution in the country, with terrain models to match. This puts a number of computed grid points inside our straits and sounds where 12km or larger grids cannot. The UW models undergo constant improvement by a dedicated research team driven by meeting the needs of real-world users, for state-of-the-art reliability.

Why stop at Nanaimo BC?  The high-resolution wind forecasts we use from the UW only go as far North as that. A future investigation is to see if we can utilize comparable models and a forecast discussion from Canadian sources.

How come computer models are uncertain?  The short answer is because they are of necessity approximations, constrained by the amount of computing horsepower available. The terrain models are not always as high a resolution as the weather grid, a critical issue along the Pacific coast.

Sets of models are run (called ensembles), each making different tradeoffs in resolution, math complexity, heuristics and input data handling. Forecasters compare these different models to see how closely they agree, which provides one measure of uncertainty – observations that appear in this App's Forecast Discussion report.