I mentioned in my last blog post that sometimes it can be hard to trust a long-range forecast. The forecast is always subject to change, but anything more than three days out is subject to what could be a great deal of change. Looking ahead seven to ten days is about as dependable as income from playing a scratch-off lottery ticket – I wouldn’t do it. You might be wondering why.
There are several factors that help to make a forecast accurate. When dealing with computer models, the better the initial data input is, the better the output should be. Therein lies the problem. Despite having more available information than ever before to give the models about the current situation, we still have large holes in the puzzle that is the world’s weather.
So, where do we get the initial conditions that go into the computer models? The hourly readings that are taken and reported at weather stations across the globe are included in the model input. Most of those stations are on land. Some are on buoys along coast lines. Instruments on airplanes send back information about what’s going on in the atmosphere as the planes travel through the many levels above us, and, even higher above the surface, satellites send us information about clouds, land, and ocean temperatures to help fill in the gaps.
Yet, there are still gaps. Our largest question mark is what is happening over vast expanses of the ocean. The satellites and passing ships can tell us something, but not as much as we get from our land-based weather stations. There are other areas on land with sparse populations and where reporting stations are few and far between that are also problematic. As long as there are areas without reporting, there will be missing information and weather model outputs will be affected.
So what do the models use to look farther down the line and predict next week’s weather? Large scale features, meaning weather patterns on a global level, are a big factor. Unfortunately, very small things that can affect those patterns are not included well in the computer models. There are also tiny errors that the models make, which in the short term forecasts are negligible, but become larger as they go farther out over time.
The computer models’ accuracy has improved in recent decades, but we still have a long way to go in long-range forecasting. I’m not saying don’t pay any attention to extended forecasts ever, but I am saying that if you’re planning an outdoor event, don’t cancel it based on the 7 day forecast because it is subject to change.