As part of my presentation to the Friends of Science on January 19 about the absence of a climate emergency, I also examined the surface temperature in Canada to determine how much it warmed up there compared to climate models.
Canada has an enormous temperature variation from year to year due to its strong continental climate. To study how observed surface temperature trends compare to climate model simulations, you need many of these simulations, each with its own large variability.
I examined the last 30 year period (1991-2020) with a total of 108 CMIP5 simulations from about 20 different climate models and calculated land surface trends over the latitudes from 51 N to 70 N and longitudes from 60 W to 130 W., which is approximately Canada covers. For observations, I used the same Lat / Lon limits and the CRUTem5 dataset that the United Nations IPCC and world governments rely heavily on. All data was downloaded from the KNMI Climate Explorer.
Let us first examine the annual mean temperature deviations from the 1981-2010 mean for the mean of the 108 model simulations compared to the observations. We see that Canada has only warmed by 50% from the average for the CMIP5 models;; The linear trends are +0.23 C / decade and +0.49 C / decade. Note that for 7 of the last 8 years, the observations were below the average of the models.
Next, I show the trends of each model versus the observed trends with a histogram of the rankings from least to maximum warming (1991-2020).
Note that 93.5% of the model simulations show warmer temperature trends than the observations.
These results from Canada are generally consistent with the results I found in the American Midwest over the summer, where the CMIP5 models heat up, on average, four times faster than the observations (since 1970) and, on average, six times faster newer CMIP6 model simulations.
The Paris Agreement, along with other national and international efforts to reduce greenhouse gas emissions, is based on warming estimates that are roughly the average of the various climate models. Thus, these results have a direct impact on the proposed energy policy decisions.
As you may know, proponents of these climate models often emphasize the general agreement between the models and observations over a long period of time, for example since 1900.
But that is misleading.
We would expect that little anthropogenic signal on global warming will emerge from the noise of natural climate variability until (approximately) the 1980s. There are two reasons for this: Little CO2 was emitted in the 1970s, and even as emissions increased after the 1940s, the cooling effect of anthropogenic SO2 emissions offset much of that warming. This is also widely accepted by climate modelers.
Therefore, to really get a good signal for global warming – both in observations and in models – we should study temperature trends since about the 1980s. That is, only in the decades since the 1980s should we see a robust signal of anthropogenic warming against the backdrop of natural variability and without the confusion (and uncertainty) of large SO2 emissions in the mid-20th century.
And every year the warming signal should get a little stronger.
I go on to claim that climate models are now generating at least twice as much warming as they should, probably due to an equilibrium climate sensitivity that is about twice as high in the climate models. Given that the average CMIP6 climate sensitivity is even greater than that of CMIP5 – close to 4 degrees. C – It will be interesting to see if the divergence between models and observations (which began around the turn of the century) will persist in the future.