Climate Nonstationarity

Changes in the statistics of climate and climate variables is sometimes referred to as climate nonstationarity.  It makes climate "normals" and other statistics based on past data increasingly irrelevant as the climate changes. 

A good example is provided by the mean summer (June, July, August) temperatures in Salt Lake City. The graph below shows the mean summer temperatures in the Salt Lake City area from 1875-2021 (blue;the latter year included with a few days missing), the 30-year running mean temperature (red), and the 1991-2020 average temperature (orange).  

Data source: http://xmacis.rcc-acis.org/

Prior to approximately 1965, temperatures in the Salt Lake City area fluctuated without a major trend, but after that time, temperatures began to climb.  The average summer temperature for the 30-year period from 1936-1965 was 72.9˚F, but for the 30-year period from 1991-2020 it was 76.4˚F, an increase of 3.5˚F.  

The 30-year averages are sometimes referred to as "climate normals."  However, when the climate is nonstationary and warming rapidly, they really represent the climate of the past not the climate of the present.  From 2006-2020, the latter half of the most recent 30-year climate normal period, 12 of the 15 years had average temperatures above the 30-year average and the average temperature for that period was 77.6˚F.  

One challenge, however, in using a 15-year period is that shorter periods are more strongly influenced by climate variability.  One can see how such variability caused some periods or relatively warmer or cooler climate prior to 1965 in the graph above.  This is a significant issue in the fall, winter, and spring when weather variability in northern Utah is quite large.  It's smaller during the summer months (especially July), but still a factor to consider.  

This is an issue that we will be facing in the coming decades as climate change continues and accelerates.  Temperature is an important variable, but precipitation extremes will be even more challenging to anticipate and plan for and will require better modeling and tools than we have today to provide guidance for future infrastructure and adaptation efforts.  

Note from Jim:

In preparing this post, I noticed that the 30-year averages I was obtaining both on the xmacis site and using summer averaged temperature data from that site were about 1˚F lower than the National Center for Environmental Information (NCEI) climate normals.  I have not been able to reconcile these differences in the short time that I have, but opted to not use the latter in this analysis so that the graph above was based on a consistent dataset.