Now that so many people have smartphones, the use of weather apps has increased dramatically. But you have to be careful using these apps, because sometimes the information they provide, which appears to be based on science, may be more “wish-casting” than forecasting.
Dan’s Wild Wild Science blog at the American Geophysical Union has a great posting on when the apps provide information that is too good to be true. If you use smartphones to look up hourly rain probabilities or look for detailed forecasts more than a week in advance, you need to be especially careful.
If you’re wondering which forecasting sources give the best results, you might be interested in this older article from OmniNerd.com, which compares the forecast errors for high and low temperatures for a number of different sources.
In general, forecasts are a lot better than they were even a decade ago due to increases in computing skill and better models. The National Weather Service seems to have some of the best forecasts for both high and low temperature but only go out to about six days (a low value on the graph below is a more accurate forecast). A few other companies go out farther in time (in one case, much farther) but those forecasts have basically no skill. In fact, one of the owners of a well-known commercial forecasting firm (I won’t mention which one) told me they do their 45-day forecasts “purely for entertainment value”. So don’t use those forecast numbers to make work plans for more than a few days out. I will say that climate forecasts for 2-3 weeks ahead do have some skill because they look at general trends, not give specific numbers.
