The 2017 edition has a Methodology section that states:
The survey fielded from January 12th to February 6th, consistent with last year.
The initial analysis was announced today (March 22) on the blog. In addition, the team will be publishing more analysis and the anonymised results in the next few weeks.
For previous years, I believe there was an announcement on the blog under the survey tag.
The survey has been pretty consistent in running for two or three weeks at the beginning of the year. I see no reason to change that pattern (and plenty of reasons to continue it). Since Stack Overflow now has an entire team dedicated to understanding developers, I suspect the methodology will be even more rigorous and transparent in the coming years.