Paul Midford writes:

"Regarding 'numbers' on renewable energy, I am also not quite sure what he means." (2012/05/01)

A quick search failed to turn up the wind power numbers from METI to which Paul Midford refers, but there are even larger wind power capacity estimates from the less business-friendly Ministry of Environment in this March
2011 report
(http://www.env.go.jp/earth/report/h23-03/chpt4.pdf).
Specifically, the report gives capacity estimates of
1.3 TW for land and 1.6 TW for sea for a total of 2.9 TW, which is almost twice as high as the low end of METI's 1.5-1.9 TW range. However, for land, the MOE report whittles down potential capacity to 0.28 TW due to a number of legal and practical obstacles. Most of this capacity-0.27 TW-becomes economically feasible at a 20-year, 20 yen/kWh feed-in tariff (FIT) arrangement if manufacturing and installation costs can be reduced from current levels by 50% and 20% respectively. No legal or social constraints are considered for sea, but only 0.14 TW in capacity potential remains available under the same cost and FIT assumptions. Still, this gives us 0.41 TW, vastly more than the 0.05 TW in nuclear power capacity that was available before 3.11.
The gap is significantly overstated, though, since capacity potential is calculated on the basis of rated output, and the average wind speed in most the potential areas is well below commonly used rated output. A post-3.11 Asahi Shinbun article (2011/04/22) used 24% for average wind power output, while METI used 85% for average nuclear power output in its most recent energy outlook. If we round those numbers out to 25% and 75% respectively, we would get annual outputs that put the 0.41 TW in wind power capacity potential at a par with 0.14 TW in nuclear power capacity. Too put it another way, wind power, under the MOE best-case scenario, could provide more or less all the electricity that Japan consumes.

So far, things look good for wind power. However, there are complications. First, there are enormous variations in wind power output, some of it predictable (ex.
seasonal cycles, morning and evening calms), some of it less so (ex. typhoons). That means that a massive network of storage systems (0.41 TW minus some measure of consumption would be a conceptual starting point for calculating necessary capacity) will be necessary to make use of all the output. Needless to say, that cost is not accounted for in the MOE report. We could avoid this problem for the most part by keeping wind power capacity somewhere below maximum demand. I don't have enough information to put a number on that capacity, but assuming that gas turbine power plants can be turned on and off seamlessly within a nationwide smart grid, I suspect that it would be more in the neighborhood of current nuclear capacity than that of the 0.41 TW wind power potential, with a corresponding output of roughly 1/10 of total electricity output.

Second, the potential capacity on land exists mainly in Hokkaido and, to a lesser extent, Northeast Honshu. The distribution is less skewed at sea, but the more productive areas lie overwhelmingly in offshore Hokkaido. Exploiting these resources to their full potential would require significant investment for high-voltage transmission systems and 50/60 Hz conversion systems in order to match supply and demand.
That's another cost not accounted for in the MOE report. However, this problem should also be largely, possibly completely, resolved by keeping wind power capacity down at the level suggested above.

I mostly follow the politics of electricity, and it has taken me too much time just to read the MOE report, apply what knowledge I had or was readily available to me, and lay out my thoughts in a way that hopefully illustrates what I meant when I sought "numbers." I'll leave the meaning of global exajoules for someone else to consider. I also did not address the possibility of wind power "exports," since I could not imagine a scenario in which neighboring countries would willingly subsidize Japanese wind power generation. In the meantime, though, I'd like to sign off with the following paradox:

"The longer the nuclear shutdown continues, the less the wind and solar power installed over the coming decades will be."