On March 31, 2017 on our twitter feed we received this comment and marked up graph from peaceful data boy @jamjoumie I wish you were applied the same skepticism of methodology to the appallingly misleading visualisations you cite
Since twitter is an unlikely forum for full explanations, we offer this blog post written up by our resident climate model and graph expert, Ken Gregory, B. AppSc. Ken has assessed models for the past decade and provides these insights on the Canadian Climate Model.
Response to Tweeted Graph with Comments
Contributed by Ken Gregory © 2017
Source of the graph is: “JH Christy”, John H. Christy from the University of Alabama in Huntsville.
About the notes on the graph below;
1) [Data improperly aligned to visually exaggerate the difference.] The box at the bottom left shows that the linear trend of all times series intersect at zero at 1979, which is the first year of the satellite data. This is the best and fairest way to present the model and data comparisons. The data sets all aligned together at the start, or left side, so you can see how separated they are at the end, or right side. This is like most races, where the athletes start at the same point and you see how separated they are at the finish line.
2) [No uncertainty ranges shown.] It is impossible to estimate uncertainty ranges for a model run, or an average of many model runs. Uncertainty can be calculated when it is due to random events. Climate model trends are dependent on the choices made by climate modelers, like assuming upper atmosphere relative humidity stays constant with warming, or more water vapour and warming causes less clouds, or that urban warming has zero effect of weather stations, all of which are contradicted by data. The choices are greatly influenced by the desire to meet the IPCC and government funding agency expectations, the need to show a catastrophic projection to ensure continued funding, modeler biases, peer pressure, etc. Uncertainty can be assigned to data measurement, but only to the random error component. Data is also subject to systemic errors, unknown errors, and biases of those interpreting and correcting the raw data. These errors are impossible to quantify. The range of all the model trends is not a measure of uncertainty. The average of the models is the consensus of the AGW theory which is used for climate policy.
3) [Averages together datasets, hiding that they aren’t in close agreement.] The UAH and RSS satellite datasets trends are both much less than the multi-model average trend, which is about 0.24 C/decade. HERE is a graph of the UAH and RSS mid-troposphere datasets. The trends of UAH and RSS are 0.085 and 0.140 C/decade, respectively. RRS had recently increase the trend from 0.094 in version 3.3 to 0.140 C/decade in version 4. NOAA also publishes a satellite analysis.
4) [Doesn’t include other groups work estimating greater warming.] There are only 3 groups publishing satellite temperature data.
5) [We don’t live on Mount Everest or in airplanes.] Nobody claimed this so the comment is irrelevant. The graph is presented as a test of the AGW theory embedded in the climate models. A graph comparing surface temperature trends would not be a proper test as the climate modelers adjust various parameters, especially aerosols, to roughly match the surface temperature trend to the published surface datasets. The greenhouse effect operates primarily in the upper atmosphere, ie, a change in the amount of a greenhouse gas at 11 km altitude has 80 times as much effect as the same change near the surface, so the mid-troposphere is the proper region to test the theory.
People often like to say Canadians cause global warming because of our oil and gas industries, but Ken shows in this video that Canadians only cause noticeable global warming in simulations – and then only because these rather exaggerated predictions are the ones favoured by groups in power of climate policy reports…