Graph says it all.
(Hat tip Green, Inc. blog)
Wind potential varies over the year by a factor of two. There's absolutely no reasonable way of storing half a year of electricity on a grid scale. So these huge seasonal variations are terminally bad - either you build enough wind capacity to function through the August nadir, and throw away the excess in other months - at huge loss - or you run natural gas turbines for half the year - with huge CO2 emissions. Or both.
The authors don't actually suggest any solution, besides the brief remark that "energy-rich chemical species such as H2 could provide a means for longer-term storage." Ehh... no.
The paper has some other interesting stuff - a physical model to extrapolate 100m-level winds from ground meterological measurements; and some correleation coefficients between different locations.
One thing that is certainly missing is shorter-term wind fluctuations - timescales not of months, but days. How much variation is left when you sum together a very large region? I think that could be worse than the seasonal mode. I've thought up a model for using METAR data to answer this question, but I'm lazy and haven't done anything in weeks. Also, major puzzle pieces are missing, such as a 100m-level extrapolation method (maybe just linear?), a way to distribute farms (to maximize power, minimize variation) and something approaching a coherent methodology. :(
A couple snapshots to show where I am (data from NOAA; shows cube of wind speed on July 10, 12-1 AM):