When Good Forecasting Computer Models Go Bad

When Good Forecasting Computer Models Go Bad
SEATTLE - The basis of forecasting models are created from taking a current picture of the atmosphere. But we can only take weather observations at certain locations around the globe, so we have to average in the areas not observed.

That, and even the most complex mathematical modeling equations we've developed can't account for all of the subtleties of Mother Nature. And once a model misses one thing, that error will become magnified as the forecast extrapolates out further in time. Add the fact that the oceans are nearly devoid of observation points, and you make it easy for mistakes to occur from averaging over large distances.