AccuWeather's 25 day forecast and we thought it would be worth tracking how their accuracy fares as the forecast changes and as we go out in time. I am grading their "snapshot" forecast from last Wednesday as a starting point, but will also compare their forecast for April 23rd (the rain and 58 degree high) as another frame of reference to show how the forecast changes over time.
First, the first six days of the forecast fared well, which is not surprising. Accu's forecast for April 4th-9th had an average error of 1.3 degrees per day, with their highest error on Sunday (forecasting 66 instead of 71) and achieving three perfect forecasts on Wednesday, Thursday, and Friday. The standard of what is usually considered an accurate forecast for three days is out is an error of three degrees or less...so that forecast was achieved.
high of 58 degrees and nearly three-quarters of an inch of rainfall. If you look on the left hand side, you'll see the forecast (as of this morning) changed just a bit. The prediction for two weeks from yesterday is now for temperatures near 80 with plenty of sunshine.
Lovers of Spring will be thrilled with the news...loathers of heat will shudder at the thought. Those who are skeptical of the use of long range forecasts in a quantifiable sense (i.e. giving specific temperature, rainfall, wind forecasts) will point to this as further evidence that long range forecasting has a long, long ways to go before we can "go there" with these types of forecasts...which I would agree with.
AccuWeather's rationale for using 25 day forecasts was to give out "critical information" to help people make decisions based on how the weather is "likely" to impact your plans. Given there's a 21 degree swing so far and the outright removal of a chance of rain, I'm not sure we can take much confidence in specific temperature forecasts with much more value than a grain of salt.