The 0-700m data seems to be less "noisy" from about 2010 onwards compared with before 2010. Is that statistically significant and if so then why?