New affirmation that local weather fashions overestimate atmospheric warming – with that?

Reposted by Dr. Judith Currys Climate Etc.

by Ross McCitrick

Two new peer-reviewed papers by independent teams confirm that climate models overestimate atmospheric warming and that the problem has gotten worse, not better, over time. The papers are Mitchell et al. (2020) “The Vertical Profile of Recent Tropical Temperature Trends: Persistent Model Biases in the Context of Internal Variability” Environmental Research Letters and McKitrick and Christy (2020) “Pervasive warming bias in CMIP6 troposphere layer” Earth and Space Science. John and I didn’t know about the work of the Mitchell team until their paper came out, nor did we know about ours.

Mitchell et al. Consider the surface, troposphere, and stratosphere over the tropics (20N to 20S). John and I look at the tropical and global lower and middle troposphere. Both papers test large samples of the latest generation of climate models (“Coupled Model Intercomparison Project Version 6” or CMIP6), ie those that will be used for the next IPCC report, and compare the model results with observations made after 1979 Examine models, while Mitchell et al. 48 models viewed. The sheer number makes one wonder why so many are needed when science is sedentary. Both papers dealt with “hindcasts,” which are reconstructions of recent historical temperatures in response to observed greenhouse gas emissions and other changes (e.g. aerosols and solar propulsion). The two studies show that the models overshadow historical warming from the near-surface to the upper troposphere, in the tropics and globally.

Mitchell et al. 2020

Mitchell et al. had investigated in a previous study whether the problem is that the models are getting too much surface heating with increasing altitude, or whether they are getting the vertical gain right but starting with too much surface heating. The short answer is both.

In this figure, the box / whiskers are model predicted warming trends in the tropics (20S to 20N) (horizontal axis) versus altitude (vertical axis). The stratosphere begins where the trend sizes cross the zero line. Red = models that internally simulate both ocean and atmosphere. Blue: Models that take the observed warming of the sea surface as given and only simulate air temperature trends. Black lines: trends observed. The blue boxes are still high compared to the observations, especially in the 100-200hPa level (upper middle troposphere).

Overall, their results are:

  • “We find significant warming distortions in the CMIP6 modeled trends and show that these distortions are associated with distortions in surface temperature (these models simulate unrealistically large global warming).”
  • “We find here that from 1998 to 2014 the CMIP5 models heat up, on average, 4 to 5 times faster than the observations, and in one model the warming is 10 times greater than the observations.”
  • “In the depths of the troposphere, not a single model implementation overlaps all observation estimates. However, there is some overlap between the RICH observations and the lowest modeled trend, which corresponds to the NorCPM1 model. “
  • “With the focus on the CMIP6 models, we have used the original results from Mitchell et al. confirmed. (2013): First, the modeled tropospheric trends in the entire troposphere (and especially in the upper troposphere, around 200 hPa) are warmly distorted and, secondly, these distortions can be associated with distortions in surface warming. Therefore we do not see any improvement between the CMIP5 and the CMIP6 models. “(Mitchell et al. 2020)

A special award goes to the Canadian model! “We draw attention to the CanESM5 model: it simulates the greatest warming in the troposphere, about 7 times greater than the observed trends.” The Canadian government relies on the CanESM models “to provide science-based quantitative information to enable in Educate Canada and internationally on adapting and mitigating climate change ”. I would be very surprised if the modelers at UVic ever put warnings on their briefings to policy makers. The sticker should read: “WARNING! This model predicts atmospheric warming approximately seven times greater than the observed trends. The use of this model for purposes other than entertainment is not recommended. “

Although the above diagram looks encouraging in the stratosphere, Mitchell et al. also found the models wrong. They predict too little cooling before 1998 and too much after, and the effects cancel each other out in a linear trend. The vertical “fingerprint” of greenhouse gases in models is a warming in the troposphere and a cooling in the stratosphere. Models predict that steady cooling in the stratosphere should have continued after the late 1990s, but observations show no such cooling in this century. The authors suggest that the problem is that the models do not properly handle the effects of ozone depletion.

The graph above focuses on the 1998-2014 period. Compare the red box / whiskers to the black lines. The red lines are the results of the climate model after the injection of observed greenhouse gases and other drives over this interval. The forecast trends do not match the observed trend profile (black line) – there is practically no overlap. They heat up too much in the troposphere and cool down too much in the stratosphere. Forcing models to use prescribed sea surface temperatures (blue), which gives the “correct” answer to the model for most of the surface, alleviates the problem in the troposphere, but not in the stratosphere.

McKitrick and Christy 2020

John Christy and I had previously compared models with observations in the tropical mid-troposphere and found evidence of warming bias in all models. This is one of several articles I’ve written on tropical tropospheric warm bias. The IPCC cites my work (and others) and accepts the results. Our new paper shows that with the latest models, the problem is not getting less, it is getting worse. The bias can be observed in the lower and middle troposphere in the tropics, but also globally.

We examined the first 38 models of the CMIP6 ensemble. As Mitchell et al. We used the first archived run of each model. Here are the warming trend coefficients from 1979-2014 (vertical axis, degrees per decade) and 95% error bars comparing models (red) with observations (blue). LT = lower troposphere, MT = middle troposphere. Each model overshoots the observed trend (horizontal dashed blue line) in each sample.

Most of the differences are significant at <5%, and the difference between the model mean (thick red) and the observed mean is very significant, which means it is not just noise or random. The models as a group are overheating throughout the global atmosphere, even over a period of time where modelers can observe both forcing and temperature.

We used 1979-2014 (as did Mitchell et al.) Because this is the maximum interval for which all models were operated with historically observed drives and all observation systems are available. Our results would be the same if we used 1979-2018, which includes scenario drives in recent years. (Mitchell et al. Report the same.)

John and I found that models with higher equilibrium climate sensitivity (> 3.4K) warm up faster (not surprisingly), but even the group with low ECS (<3.4K) show warming bias. In the bottom group, the mean ECS is 2.7 K, the combined LT / MT model warming trend average is 0.21 K / decade, and the observed counterpart is 0.15 K / decade. This figure (green circle added; see below) shows a more detailed comparison.

The horizontal axis shows the warming trend of the model and the vertical axis shows the corresponding model ECS. The red squares are in the high ECS group and the blue circles are in the low ECS group. Filled forms come from the LT level and open forms come from the MT level. The crosses indicate the mean values ​​of the four groups and the lines connect the LT (solid) and MT (dashed) layers. The arrows point to the mean observed MT (open arrow, 0.09 C / decade) and LT (closed arrow, 0.15 C / decade) trends.

While the models in the blue cluster (lower ECS) do a better job, they still have warming rates that are above observations. If we were to envision a third cluster of models with mean global tropospheric warming rates that overlap with observations, it would have to be positioned roughly in the area I have circled in green. The associated ECS would be between 1.0 and 2.0 K.

Final remarks

I understand that modeling climate is incredibly difficult and no one is blaming the scientific community for being a difficult problem to solve. But we all live with the consequences when climate modellers persistently use generation after generation of models that exhibit too much surface and tropospheric warming, in addition to greatly exaggerated propulsion scenarios (e.g. RCP8.5). As early as 2005, Karl et al. pointed to excessive warming in the tropical troposphere as a “potentially serious inconsistency”. But instead of fixing it since then, modellers have made it worse. Mitchell et al. Note that, in addition to false warming trends, the distortions themselves have wider implications since “atmospheric circulation trends depend on latitude gradients in temperature”. In other words, if the models misunderstand the tropical troposphere, it leads to possible errors in many other features of the model atmosphere. While the original problem was limited to excessive warming in the central tropical troposphere, it has now expanded to more extensive warming throughout the global troposphere.

If the discrepancies in the troposphere were evenly distributed among the models between excessive warming and cooling, we could attribute this to noise and uncertainty. But that’s not the case: it’s all overheating. CMIP5 models warmed too much above the ocean surface and too much in the tropical troposphere. Now the CMIP6 models are heating up too much in the global lower and middle troposphere. This is a bias, not an uncertainty, and until the modeling community finds a way to fix it, the business and political communities are entitled to assume that future projections of warming are overestimated, possibly significant depending on the model.


Karl, TR, SJ Hassol, CD Miller and WL Murray (2006). Lower Atmosphere Temperature Trends: Steps to Understanding and Balancing Differences. Synthesis and evaluation product. Climate Change Science Program and the Subcommittee on Global Change Research

McKitrick and Christy (2020) “Pervasive warming bias in CMIP6 Troposphere Layers” Earth and Space Science.

Mitchell et al. (2020) “The Vertical Profile of Recent Tropical Temperature Trends: Persistent Model Biases in the Context of Internal Variability” Environmental Research Letters.


Item rating

Like this:

Like Loading…

Comments are closed.