Early voting turnout in Louisiana, Georgia, and North Carolina is already higher than the total early vote in 2004. Georgia's turnout is up by 36%, and the numbers are huge among African-American voters.
All three of those states have large African-American populations, and, as Nate Silver notes, African-American population at the state and county level is the "key determinant of early voting" turnout in this race. The question isn't for whom those voters are voting; we know that the vast majority of African-Americans in the United States 1) are voting, and 2) are voting for Obama.
What's not clear is what the effect of such high turnout will be on the downballot races. If African-American voters have high turnout and mostly vote straight-ticket Democratic, it could have a major effect on the Senate races in Mississippi, North Carolina, and Georgia. The polls for two of those races are neck-in-neck, which is kindof unbelievable for the solid Republican South. In North Carolina, it's highly likely that Democrat Kay Hagan will take Elizabeth Dole's seat, also known as the Seat Formerly Held by Jesse Helms. The thought that high turnout among African-American voters could lead to a Democrat taking Jesse Helms' old Senate seat seems like a bit of poetic justice, n'est pas?
Another thing to look for today: the Weekly Reader poll of elementary school students. Weekly Reader usually ends up being the most accurate of all the polls; since 1956, they've only been wrong once (in 1992). I'll post when the results are released.