30GW of power from solar ? Hmm. Considering that a reasonable estimate of energy density for solar is 5 W/m^2, we would need (30*10^9)/(5*10^6) = 6000 sq km  of land dedicated for power generation.

This is no big amount for a country with 3,287,590 sq km of land. But when you consider that most of this land is already occupied for food production and forestry, you will be hard pressed to find good locations for these solar power plants.

For a tropical country with a very sensitive ecosystem, 6000 sq km is a very tall order.

When you get ambitious and want to produce 100 GW instead of 30 GW, you will need 20,000 sq km of land instead.

Readers are welcome to point out on a map of India where they will keep their choice solar power installations. One good choice is using the Thar desert in the north west, probably for concentrated solar power generation. It has 200,000 sq km of area, and we might cover a portion of this with concentrated solar power installations. Any idea how much this would cost  (including the cost of laying down transmission cables to ship this power to cities such as Mumbai and Delhi) ?

Any other alternatives ?

Average daily solar radiation in India is about 4-7kwh/m2, not 5w/m2.  There's 5,000 trillion kwh annually over India's entire landmass during an average of 300 sunny days per year (http://mnes.nic.in/faq-solarpower-map.htm). That's plenty of sun.

The key economic input to solar energy is not sun but land (and fabrication costs).

Sun costs nothing. But land is very valuable.

Why don't you give the requisite figures based on land occupancy ?

Imagine nuclear energy costs being given in terms of Uranium fuel costs, but the plant construction and decommission being ignored. How would be your outrage ? It is equally outrageous ignoring the demand on land for solar energy production.

There are efficiency losses with each different method when sunshine is converted to electricity. By far, the best method we have is solar PV and its current efficiency is around 20%. New research advances might make it 40% efficient, but these are not available in the market yet.

I borrowed the analysis of energy density from the physicist Prof David Mackay for England.

His reasoning :

The power of raw sunshine at midday during a cloudless day is 1000 W/m^2. It is fair to say it is midday for about 1/4 of the time, and dark the rest 3/4. So we have a daylight factor of 1/4. We also lose power due to cloud cover. We should include a sunniness factor of 1/3. So we get 1/4*1/3*1000 = 80 W/m^2 of solar power hitting south faced roofs. Using 20% efficient solar panels, we have an energy density of about 16 W/m^2.

From the numbers you quoted, average solar irradiance in India is 5.5 KWH/m^2 per day. That is a capacity of 5.5 * 1000/24 = 229 W/m^2.

So India apparently gets 2-3 times the amount of sunshine as England. But I don't know how much of this is true !

With 20% efficient solar PV, we get an energy density of 45 W/m^2

With 10% efficient solar PV, we get an energy density of 23 W/m^2

As the article (and previous articles such http://gristmill.grist.org/story/2008/4/9/162115/5465) argue, because the national transmision system is so rickety and unreliable, many in India have moved to using their own local power sources. The article then goes on to say this is bad thing. In fact its not at all. The problem is what is being used to provide that power - nasty dirty diesel generators in the cities and wood burning in the villages. But in fact if those nasty generators could be replaced with small scale hydro or wind or solar generators, suddenly you've solved two problems at once: the problem of a national government that moves like molasses, and a rickety transmission system as well as loosening the grip on power that comes whenever anything is decentralized. the fact that many Indians are already used to the idea of decentralization anyway should make this a much easier proposition than say here in the US. Why aren't env groups pursuing this option?