Solar pond
en.wikipedia.org
en.wikipedia.org
http://www.sciencedirect.com/science/article/pii/S0038092X10...
> Heat has generally been successfully extracted from the lower convective zone (LCZ) of solar ponds by two main methods. In the first, hot brine from the LCZ is circulated through an external heat exchanger, as tested and demonstrated in El Paso and elsewhere. In the second method, a heat transfer fluid circulates in a closed cycle through an in-pond heat exchanger, as used in the Pyramid Hill solar pond, in Victoria, Australia.
Note: I haven't tried these particular instructions myself, although I did do a similar demo around 20 years ago.
They used this, and maybe still do, at UIUC to heat the Vietnamese potbelly pig barns south of campus.
Can that really be right, it seems like a very high gradient.
I'm not sold on the numbers, but I'm also not yet convinced of the idea's physical impossibility.