PV on every rooftop?
It's a good question. Let's do the first order approximation.
First, how much electricity to we use? According to the DBEDT energy trends, we use about 10 terawatt-hours (TWhs) of electricity annually. In fact, 10125.94 gigawatt-hours (GWhs) in 2009, 10013.10 GWhs in 2010, and about 9985.55 GWhs in 2011. So that's our (hopefully shrinking) target.
Second, how many roofs do we have to cover? Let's just look at residential. According to the 2011 US Census, Hawaii has 519,508 housing units, 39.2% of which are multi-family. So let's just look at the single family units (we'll be more conservative here and more generous elsewhere). So that gives us 315,861 single-family home rooftops. Now let's say for each rooftop we can fit a 4 kilowatt (kW) system. This is probably being a bit generous, given the size and possible shading issues. With all those rooftops tiled with 4 kW of PV each, we have 1,263,443 kWs, or 1263 MWs of PV (which, BTW, approaches the total system capacity on Oahu).
Of course, the sun isn't always shining. In fact, for PV, the "capacity factor" is between 15% and 20%--meaning that at any given moment you will have able to produce between about 15% and 20% of the rated PV capacity. Let's use the generous 20%. For our rooftops this means (20% X 1263) 253 MW of PV capacity. Now we can look at the total production over one year (at the already "de-rated" PV installation). So 253 MW X 8760 hours in a year = 2,213,553 MWhs, or 2,214 GWhs, or 2.2 TWhs. This would provide about 22% of our overall electricity use.
This 22% is probably conservative--we ignored all of the commercial rooftops. Plus we are seeing more and more large ground-mounted PV arrays (usually in 5 MW blocks because that is the largest size before the utility needs to competitively bid). Nonetheless, it reminds us that we need a mix of renewable energy sources. And yes, we hope to shave our 10 TWhs of usage by 30% come 2030 (the HCEI target), but we're also adding a bunch of electric vehicles to the grid (which could just cancel out that efficiency gain--which is fine for the big picture).
By the way, any guess of how much all of that PV would cost? About $10 billion. It would pay for itself in about 13 years.