Closer to option 1, but I'm a meteorology major, so that's to be probably to be expected.
Not really.
American Meteorological Society, 2003: "The nature of science is such that there is rarely total agreement among scientists. Individual scientific statements and papers—the validity of some of which has yet to be assessed adequately—can be exploited in the policy debate and can leave the impression that the scientific community is sharply divided on issues where there is, in reality, a strong scientific consensus.... IPCC assessment reports are prepared at approximately five-year intervals by a large international group of experts who represent the broad range of expertise and perspectives relevant to the issues. The reports strive to reflect a consensus evaluation of the results of the full body of peer-reviewed research.... They provide an analysis of what is known and not known, the degree of consensus, and some indication of the degree of confidence that can be placed on the various statements and conclusions."
I don't now if referencing a 2003 claim of consensus is really valid considering recent revelations.
2005 maybe?
http://nationalacademies.org/onpi/06072005.pdf
How about 2007?
http://www.wmo.ch/pages/themes/wmoprod/documents/iwtc_statement.pdf
AMA didn't go all out and say global warming is definitely causing an increase in hurricaines, only that studies seem to preliminarily indicate some convergence. They sure as heck didn't switch their position about global warming however.
Climategate began in 2009, though some of the fishy climate science was known before then. Just read over Snowguy's climate thread if you haven't already.
But really the whole consensus thing is political BS - if there was a consensus I don't think there would be this much debate going on involving real scientists on both sides. Besides, even if there was a consensus that does not imply correctness. Many times in history the scientific community at large has held ideas that were incorrect but widely accepted only to have it changed by new data.