Crossposted from my blog. The graphs may be viewed better there. May I suggest you print this out first and then read it, to save yourself eyestrain. Thank you.
Note: Graphs and texts in italics are from the Cato Working Paper No. 35.
Back at the end of 2015, the Cato Institute published a paper titled,Climate Models and Climate Reality: A Closer Look at a Lukewarming World (Patrick J. Michaels and Paul C. Knappenberger), which assessed that the rise in global mean temperatures expected under Global Warming and forecast by the various models were much higher than those temperatures that actually transpired almost through to the end of 2015. Almost, because their paper only covered a portion of that year. Needless to say, the actual means of 2015, 2016 (and so far in 2017) appear to indicate that the paper is not holding up very well in its main historical finding: that the globe is warming at a slow trend of only about 1 degree Celsius per decade, with an apparent deceleration from 0.17 (Hadley) / 0.19 (GISS) for 1993-2015 to 0.10 (Hadley) / 0.12 (GISS) for 2006-2015.
[Figure 1. Global average mid-tropospheric temperature variations (5-year averages) for the average of 102 models runs (red line). Circles (balloons) and squares (satellites) depict the observations.]
This paper starts with the finding of University of Alabama’s John Christy, presented to the US Senate Subcommittee on Space, Science and Competiveness, that the mid-troposphere temperatures have been increasing by the more moderate trend of about 0.1 deg C per decade to 0.32 deg C in 2015, a far cry of the IPCC model runs’ predictions’ average of 0.2 deg C per decade and hitting about 0.86 deg C in 2015.
Fair enough. These temperatures are important, for they help determine what our weather is going to be like. But they also mean that what was to come weather-wise in these years wasn’t as bad as it could have been… and yet it was still bad… and weird! Like a tropical storm, for example, Epsilon, forming in the Atlantic late in 2005 and lasting into 2006. So what’s the reason for the mid-troposphere temperatures? Well, the Cato paper forthrightly states,
Rain and snow are largely dependent on the temperature difference between the surface and the mid-troposphere. When there’s little difference, air in the lower atmosphere does not rise, meaning that the vertical motion required to form a cloud is absent. When the difference is large, moisture-laden surface air is very buoyant and can result in intense rain events.
Getting the vertical difference systematically wrong in a climate model means getting the rainfall wrong, which pretty much invalidates regional temperature forecasts. A dry surface (think: desert) warms (and cools) much more rapidly than a wet one.
The paper goes on to say this about the computer models and the data:
If the computer models are somehow getting surface temperatures right that could only be a fortuitous result if the mid-tropospheric temperatures are as far off as Christy’s data shows.
WRONG. If the mid-tropospheric temperature data are as far off from the models as Christy’s data shows and the surface temperature data are matching the temperatures predicted in the models, that means the temperature differences between the surface and mid-troposphere are getting larger, meaning more intense rainfall events! Maybe more intense windstorm events, too… we have had five major hurricanes in a row in the Atlantic oceanic basin including the Caribbean and the Gulf, a record first for this ocean, and a 185 MPH monster of a hurricane, boasting the highest tropical cyclone winds ever recorded in the Atlantic. It would be better for us if the surface temperatures are increasing only as slowly as the temperatures in the mid-troposphere are increasing, for that would mean the temperature differences between the two levels are not becoming any greater, and the weather is not becoming any weirder or any worse.
Except it is.
Which brings us to those pesky surface temperatures.
Michaels and Knappenberger present their decadal trend findings using the global mean temperature compilations from five sources: the UK Hadley Centre (HadCRUT4), Cowtan and Way (2013) (C&W), the National Oceanic and Atmospheric Administration (NOAA), NASA (GISS), and Berkeley Earth (BEST). The following graph presents the observed temperature trends:
[Figure 2. The annual average global surface temperatures from 108 individual CMIP5 climate model runs forced with historical (+ RCP4.5 since 2006) forcings were obtained from the KNMI Climate Explorer website. Linear trends were computed through the global temperatures from each run, ending in 2015 and beginning each year from 1951 through 2006. The trends for each period (ranging in length from 10 to 65 years) were averaged across all model runs (black line). The range containing 95 percent (dotted black lines) of trends from the 108 model runs is indicated. The observed linear trends for the same periods were calculated from the annual average global surface temperature record compiled by several different agencies described in the legend (colored lines) (the value for 2015 was estimated from January through October, average).]
As you can see, the paper shows a leveling off of the decadal trends at trend-length of 41 years (1976-2015, inclusive) onward at roughly on average about 0.19 deg C / decade until the trend-length of 23 years (1993-2015) where it then decelerates unevenly to a rough average of 0.11 deg C / decade at trend-length of 10 years (2006-2015). The decadal trends indicated by the Hadley Centre and the NASA GISS data decelerate from 0.17 to 0.10 deg C / decade and from 0.19 to 0.12 deg C / decade, respectively. The observed temperature trends fall below the multiple model-run mean (MMM) and for several trend-lengths they lie very close to or even below the 2.5th percentile margin for all the model runs.
But before going on, I have an explanation of this graph to describe.
The forcings used for the model predictions are not consistent throughout—the IPCC included the historical forcings up until 2006 but after that year used the four various RCP forcings to predict “future” warming rates under their respective scenario. For the purpose of this study Michaels and Knappenberger picked the RCP 4.5. I was hoping the authors’ technical help would have reprogrammed the models to use the historical forcings to as recently as possible for greater accuracy, and to really see if the models are way off even with historical forcings throughout, but I suppose that in the end they were stuck with one of the four forcings. But then it doesn’t matter which one is picked, for all four of them are virtually identical in forcing predictions under “business as usual” until the year 2020.
Also please note that the lower the trend for past years, the higher the past year’s global mean temperatures is in respect to the 2015 means, when compared to adjacent years of higher trend value. Conversely, higher trends are indicative of the respective past year’s lower temperatures when compared with its neighbors. This is important.
Now I have a bone to pick over this graph.
If you notice, the trend peak that is roundabout the 25 years’ trend length (1991-2015) is located at 23 years (1993-2015). This is true for both the model runs (except for the 2.5th percentile plot line which is at 22 years) and all the observed temperature trends. Unfortunately, these are all a year off toward 2015; because, since Mount Pinatubo blew up toward the end of 1991, the observed temperature records show in reality a low point for the global means in 1992. I have seen other plots of the IPCC-CMIP5 and they show the models’ low point in 1992 also. The above discrepancy is a gross error and should not have been allowed to be presented to the public uncorrected.
Now we go on to the next graph.
The reason for this next graph is important, because Cowtan, K. et al, 2015 (“Robust comparison of climate models with observations using blended land air and ocean sea surface temperatures.” Geophysical Research Letters, 42, 6526-6534, doi:10.1002/2015GL064888) noted an inconsistency between the IPCC models and the observed data, the same discrepancy the Cato Institute noted nine months before and it was that
observed compilations combine air temperature measurements over the land with sea surface temperatures into a global average, while climate model compilations use air temperatures over both land and oceans.
This variance was compounded by another inconsistency the authose of the two papers found, which was that
observed temperature compilations include regions of missing data (i.e., incomplete geographic data coverage) while climate models include the entire surface
meaning that the model probably has a slight warming bias compared to the observed data, or that the observed data are underestimating the actual warming! So a new model dataset (http://www-users.york.ac.uk/~kdc3/papers/robust2015/methods.html) was developed and made available that took into account the different temperature data takings between land and sea—and it shows a slower global warming mean in the model, one that is roughly consistent with the observed data until the 41-year trend length (1975-2014 [see below]) at which point the two lines start to diverge from one another, showing a warming bias still in the adjusted models.
[Figure 3. The annual average global surface temperatures, derived from a similar methodology used by the UK’s Hadley Centre in compiling temperature observations, from 109 individual CMIP5 climate model runs forced with historical (+ RCP4.5 since 2006) radiative changes. These were obtained from the University of York website (http://www-users.york.ac.uk/~kdc3/papers/robust2015/index.html), see Cowtan et al., 2015 for more details. Linear trends were computed through the global temperatures from each run, ending in 2014 and beginning each year from 1951 through 2005. The trends for each period (ranging in length from 10 to 64 years) were averaged across all model runs (black line). The range containing 95 percent (dotted black lines) of trends from the 109 model runs is indicated. The observed linear trends for the same periods were calculated from the annual average global surface temperature record compiled by the UK’s Hadley Centre (red line).]
Now the above graph appears to have some really gross errors introduced into it; for what reason I cannot say.
First, the Hadley trend-line is at wide variance from that shown in Figure 2. Either they ran the observed temperature trends with reference to 2014, or they really put some really fat-fingered errors in it, to wit: The plot line of trends has been bent or otherwise “adjusted” at the 18-year trend length and the 35-year trend length so that it slopes further down for the smaller trend lengths, compared to those of the same line in Figure 2. Even the slopes of the line between the 18-years’ length and 10-years’ length is noticeably different to the naked eye. And of course, this reflects variances in the trend values from those in Figure 2, especially so for the shorter trend lengths. This in my book is a gigantic mistake: either Michaels and Knappenberger computed the trends from the data sets with the reference years at a one-year variance between Figure 2 and Figure 3 (i.e., 2015 and 2014), or they “adjusted” the data from the one figure to the next. This last is inexcusable.
Second, the model runs in Figure 3 are different (reference year 2014) from those in Figure 2 (reference year 2015). This to me is just as big a glitch as the second error I discovered, for we are no longer comparing apples to apples, but apples to oranges.
I cannot say this emphatically enough; the model runs and the Hadley Centre’s observed data plots ought to be identical in both Figures 2 and 3! There should be no room for such gross discrepancies as this.
Before I verify or debunk the these findings, allow me to take an excursis back to the mid-tropospheric temperature trends as the Cato authors do in their working paper.
[Figure 4. The annual average global mid-tropospheric temperatures derived from 102 individual CMIP5 climate model runs forced with historical (+ RCP4.5 since 2006) forcings were obtained from John Christy (personal communications). Linear trends were computed through the global temperatures from each run, ending in 2015 and beginning each year from 1975 through 2006. The trends for each period (ranging in length from 10 to 40 years) were averaged across all model runs (black line). The range containing 95 percent (dotted black lines) and the minimum (dashed black line) of trends from the 102 model runs are indicated. The observed linear trends for the same periods were calculated from the annual average global mid-tropospheric temperature record compiled by several different agencies (and include compilations derived from satellite observations as well as weather balloon observations) described in the legend (colored lines) (the value for 2015 was estimated from January through October, average (data provided by John Christy).]
The above shows that the observed mid-tropospheric temperatures ran far below what the IPCC model runs predicted. So bad, in fact, that some of the observed data trend-lines run below the minimum model run line and they all basically run below the 2.5th percentile line for a good portion of their length before rising to a mean average of about 0.16 deg C / decade at the 10-year trend length, with Christy’s data trend, UAHv6.0, coming out at 0.18 deg C / decade for the ten-years’ trend length. Still, the general rough average trend for all years is about 0.1 deg C / decade, with obvious variations, of course.
And I notice that the longest trend length is 40 years, with the data starting in 1975 according to the caption and ending in 2015. Yet 40 years inclusive from 1975 would bring us to 2014. And if the ten-year trend length is set to begin in 2006, the 1979 data would appear in the 37 years’ trend length, which it does! For proof, see the Christy / Univ. of Alabama in Huntsville (orange) line above and his data in Figure 1 at the top. And the orange line appears to show a higher trend when compared with Christy’s data in the top figure—because the above line goes by the annual averages and the Figure 1 data are annual five-year averages. Ugh!
I’ve run through 2300 words now so the rest I’ll have to include in later parts.