Will,

Where did you get your US electrical data? The BP spreadsheet shows 2005 US electricity generation at 4,257 TWh per year.

Using this number, I come up with 6,949 square miles (7,130 square miles using 2007 data of 4,368 TWh).

Yes, my first comment above was before my first cup of tea this morning and reflects the TOTAL energy consumption of the US, not just electricity. I'll refrain from posting before morning tea in the future :-) , though it doesn't hurt show what a complete replacement of US energy by solar might be as a brainstorming exercise.

I found 4,064,702 GWh/yr of 2006 US electricity at http://www.eia.doe.gov/cneaf/electricity/epa/epates.html , so definitely replace my previous number. Your 7130 square miles sounds accurate, and it's only about 1/15th of the size of Arizona (114006 square miles).

Other comments below about reducing energy consumption are absolutely valid, and I consider these to be steps everyone should be taking, (myself included).

You have to be very careful about those total energy consumption numbers, because large fractions of raw energy inputs represent heat lost in conversion to electricity or fail to reflect efficiencies which would become standard if the input was electric (e.g. heat pumps for both space heat and DHW).

You are correct, that is for total energy. As an exercise, I went through the BP spreadsheet, and calculated fossil replacement for each fossil fuel:

Total Energy: 28,332 TWh = 46,446 SqM = 0.41 Arizonas

Oil: 11,316 TWh = 18,551 SqM = 0.16 Arizonas

Nat Gas: 7,152 TWh = 11,725 SqM = 0.10 Arizonas

Coal: 6,888 TWh = 11,292 SqM = 0.10 Arizonas

The next logical step is to calculate how much you would need to replace oil and nat gas to get to 2030, when the ELM says we'll be SOL:

That's about 18,000 TWh. Which means, that we would need to build 782 TWh per year, for 23 years. This essentially means, we would have to build one solar plant, 18% larger than the one described, every DAY for the next 23 years. Time to get out the shovels.

When looked at as a percentage of total land area of the USA the amount of area is 1/8th of one percent. Subtracting Alaska and figuring in the low insolation factors of some regions then around 1/2 of a percent of any region's surface area could provide all of our energy needs. That is 3.2 acres per sq mile leaving 636.8 acres for other uses.

The solar energy arriving per 1 year on 1 km² desert is on average 2.2 Terawatt hours (TWh) (Desertec pp 17)

World annual energy consumption 2007 was 11099.3 million TOE. (BP statistics)
The conversion factor used by the IEA for electricity is: 1 TWh = 0.086 Mtoe.
World annual energy consumption 2007 was 11099.3 million TOE / 0.086 = 129,062 Twh

So, 129,062 / 2.2 = 58,665 km², or a square area 245 km on a side. or 100 areas 24.5 km on a side.
given a 50% area efficiency, then use
So, 129,062 / 2.2 x 2 = 58,665 km² x 2 = 120,000 km², or a square area 347 km on a side. or 200 areas 24.5 km on a side.
assuming North America = 25% of all world energy 2007, then
So, 129,062 / 2.2 x 2 X .25 = 30,000 km², or a square area 180 km (110 miles) on a side. or 100 areas 17.5 km (10 miles) on a side.
(NB: occasionally rounded)

Edit - Error above, sorry
Failed to provide for only 15% insolation to electricity conversion. Corrected numbers are:

2.2 Twh x 15% station effic. x 40% land effic. = 0.132 Twh per km²

So, 129,062 / .132 = 955,000 km², or a square area 1000 km on a side. or 100 areas 100 km on a side.
assuming North America = 25% of all world energy 2007, then
So, 129,062 / .132 x .25 = 250,000 km², or a square area 500 km (310 miles) on a side. or 100 areas 50 km (31 miles) on a side.