griteater
Member
In addition to what Fro and CR stated, I like to view the Control as an extension of the Euro Op run, to just see what it shows past day 10. As with any Op or ensemble member, you don't take it to the bank, you just look at the overall features. I like to look at what it is showing with the Pac Jet, etc., as it has more definition / detail than you are going to see on an Ensemble Mean out at range.
But here's something interesting too...there probably have been more recent papers written on ensembles (I'm sure), but there is a paper from 2005 that showed how the Op run is more accurate in days 0-5, but the Control run is more accurate from days 5-10 (again, I don't know if that has changed, but that's what is stated in the paper).
Source: http://www.iapjournals.ac.cn/fileDQKXJZ/journal/article/dqkxjz/2005/6/PDF/226zyj.pdf
Here is a chart and select text from that paper:
..."For the short lead-time (0–96 hours), the high resolution GFS is the best, and the individual ensemble perturbation forecasts are far behind either the GFS (due to the resolution and initial error) or CTL (due to the initial error). After a short lead-time (120 hours), the model resolution is not as important as the first 96 hours to improve the model forecast skills; as unexpected, CTL is slightly better than GFS from 144 hours to 264 hours lead-time in this experiment period."
..."Interestingly, both GFS and CTL are still better than any of the individual ensemble members."
..."The resolution plays a key role in the success of the short-range forecasts while the influence of the resolution is much smaller than that of the initial conditions for medium-range forecasts."
