I'm really starting to think that firmware 7.0 has something to do with this, whereby temperature does have a strong correlation to the reported rated range.
I've put together the last 130+ data points from my 90% charges (range mode was always ON) since late-January, when VisibleTesla started working for me with the new-at-the-time experimental version, and created the graph below.
I upgraded to firmware 7.0 (2.7.056) as soon as it came out, which is on the evening of 10/14/2015. My 90% charges were still very stable at around 224 miles of rated range, while temperatures were still fairly mild in the San Francisco bay area. So there was no change between 6.2 and 7.0 in calculating rated range in mild temps.
Then in late October, temperatures dropped by at least 10ºF and have stayed fairly low. I keep track of my daily commute, and my morning temps had been in the mid- to upper-60's until 10/30, at which point they dropped to lower- to mid-50's. FYI, my service center upgraded my firmware to 2.7.77 on 11/7.
I know some have seen drops unrelated to temperature, but my pack is coming up on one year and over 22,000 miles so it should have stabilized by now (you can see that from mid-May to late-October in my data).
Which reminds me, some of you 90 kWh pack owners are probably seen early drops, which are more common at first, then degradation stabilizes to a lower rate.
But what I'm seeing from my data is highly unusual drops in reported rated range, and it's highly correlated to the recent colder temperatures in my area.