I've never been a fan of using "new averages". I think doing so distorts the real averages. The entire data set should be used to establish the averages....and before anyone "yells" at me, I am well aware that the official site for the weather observation(s) has changed over time.