The uncertainty range is largely derived from looking at how well using the same months of data in previous years would have performed and then selecting the worst case. (carbonbrief.org)
In 2016 I only had a few months of data to work with, so small fluctuations would cause my projected 12 - month number to jump around. (themoneycommando.com)
I think this number will start to stabilize now that I have 12 full months of data to use in the calculations. (themoneycommando.com)