How far into the future can we anticipate winter weather?

As an example of predictability limits in the EPS, we show the graphic below for EPS forecast ensemble mean and normalized spread of 500-hPa heights from a cold-season case valid 312 hours after the ensemble initial time (the time of interest for this case). For this forecast cycle (and a number of consequent ones) there was little indication that the large-scale flow would be conducive to a winter weather event in the southern and eastern U.S. There is some suggestion of split polar and subtropical jet streams, especially in the separate centers of high spread in the 500-hPa heights (maxima south of Hudson Bay and over Texas, northeastern Mexico and adjacent Gulf of Mexico waters). This split flow was the only hint that winter weather might be possible, with any potential for winter weather in the southern and eastern U.S. dependent on the timing and phasing of northern and southern stream systems. Even at the short range, such processes can be difficult for NWP models to predict accurately!

Typically at this time range (13 days), the most a forecaster can expect from the ensembles is a general idea of flow regime, and thereby a general idea of possible weather types at that time range. Oftentimes even the flow regime will be unclear, and nothing can then be inferred from the ensemble at this time range. However, occasionally there are times when the predictability at this time range is as good as or better than at 3 to 5 days! An example is shown on the NCEP ensemble page here. This MREF forecast example was developed by Dr. Zoltan Toth at the Environmental Modeling Center (EMC).