Have GCM’s been proven able to predict changes in the Global-Mean Annual-Mean Temperatures? That is the question.
I’m pursuing the answer and beginning by examining the predictions contained in Hansen et. al 1988,
This paper contains two distinction types of runs, each of which could be said to contain a numerical experiments of sorts; in Hansen’s paper, these are divided into ‘the control run’ and ‘transient simulations’. In this post, I discuss the control runs.
The Control Run
The control run describes model computations holding the level of Green House Gases (GHGs) in the atmosphere at 1958 levels. The motivation for computing these is likely to show the Mean Global Temperature does not increase indefinitely when CO2 is head constant. Also, these runs substituted a simplified model for ocean transport. Nevertheless, if the runs were considered good enough to include in the paper, the runs might also be considered to represent numerical experiments that permit some sort of answer to these questions:
- Would GISS GCMs have predicted warming from 1958-1987 had GHG’s been kept at 1958 levels?
- Would they have predicted continued warming from 1988-2006? and
- WBased on this GCM model, what sort of temperature excursions might we expect when there are no aerosols causing dimming, no additional injections of GHG’s, and no volcanic eruptions?
Here is Hansen et al. figure 1, with a few additions:
Discussion of 1958 atmosphere numerical experiment.
- Years: Hansen et al. does not describe how the initial conditions were set and I don’t see a citations suggesting this is discussed elsewhere. So, I have assumed the computations were initialized using weather data from 1957. If so, then years 0-100 correspond to numerical experiments predicting what would have happened from 1958-2057 had the atmosphere remained unchanged after 1958.
- Trendlines: Eyeballing the graph, it appeared that the model predictions indicated that global average temperature would experience a distinct up trend from 1958-2008, rising roughly 0.35 C in 5 decades. It would then plunge even more rapidly from 2008-2030 and rise from 2030-2057. Interestingly, despite these wild oscillations, temperature in 2057 and 1958 were nearly equal.
So, it appears this GCM control run predicted the world would experience a noticeable heat wave after 1958, even with no rise in CO2 or GHGs. Moreover, the temperature was expected to rise both from 1958-1988, when Hansen et al. 1988 was published, and continue to rise until a few years after 2006, when the data comparison paper was published.
It also appears that, when nothing interesting happens (no volcanic eruptions, no increasing GHGs, no addition or subtraction of aerosols) the GCM control run predicts that Global Average Temperature experiences swings of as much as 0.5C during a century, with a precipitous drop of 0.5C happening in only 25 years.
Preliminary answers to two interesting questions:
Would GISS GCMs have predicted warming from 1958-1987 had GHG’s been kept at 1958 levels? Yes. It we might have expected measurable warming from 1958-1987 even if GHG’s were kept at 1958 levels.
- Would GISS GCMs have have predicted continued warming from 1988-2006?
Yes. It appears we might have expected measurable warming from 1958 through next year even if GHG’s were kept at 1958 levels.
- Would the control run have predicted continued warming from 1958-2006?
Yes. It appears that we would have expected somewhere between 0.3 C and 0.4 C or warming during this period.
- Based on this control run, what sort of temperature excursions might we expect when there are no aerosols causing dimming, no additional injections of GHG’s, and no volcanic eruptions?
Because only once 100 year experiment was done, it would be best to assume that the results are more likely normal than exceptional. If so, one might expect swings of 0.5 C during a decade to be normal. If we consider each of the four 25 year periods within the 100 years to be independent realizations, we might suggest that drops of 0.5 C in 25 years happen once every 5 groups of 25 years. No uncertainty bounds on these claims can be provided.
What do we might we conclude by treating the control run as a numerical experiment?
We may learn many things. Among them, it appear thats, the world might have been set to warm 0.3 – 0.4 C since 1958 even without changes in CO2 or GHG.
Or? Or, the Hansen et al. 1988 omits details that might explain why the ‘1958 atmosphere’ control run does not represent any sort of numerical experiment for climate change from 1958-2058.
My career has been associated with mathematical modeling and understanding of the transient response of human-made engineered systems to upset conditions. Conditions both inside and outside the range of normal operating conditions. These systems and conditions are not natural systems such as the climatic response of the planet to imposed boundary conditions. There are arguments floating around that ‘natural systems’ are somehow different from human-made engineered systems, but I don’t want to get into those discussions.
But the systems on which I work are much like the planet’s systems in that an imposed energy source is present and the internal states of the systems are determined by the hydrodynamic and thermodynamic phenomena and processes occurring within the systems. That is, the internal states are determined by the transport and storage of the imposed energy source by means of the usual heat transfer and fluid flow processes and the thermodynamic processes that the working fluids undergo due to interaction with the energy source and the equipment subsystems.
When the models/codes/applications are Validated by comparisons with experimental data, or when projections such as those shown in the Figure of this post are made, all aspects of the transient response of the system must be fully understood with respect to, and reconciled with, the underlining basic physical phenomena and processes (mass and energy storage and distributions, for examples), plus these as captured by the continuous model equations, and also the possible effects of numerical solution methods. And by all aspects I mean both the general trends of the calculated results over longer time scales as well as those that occur over smaller time scales. The smaller-scale responses include things like the time-rate of change of the response and changes in the sign of the time-rate of the response in addition to the actual magnitude of the quantity of interest. This is especially true whenever measured data and calculated values are different in these respects. All must be eventually reconciled and the root-cause of the differences unambiguously identified.
The Figure in this post is useful as a basis for discussions of these aspects of Validation exercises as follows. I think it is a true statement that the smaller-scale aspects of the response of The Global Average Temperature have never been subjected to any such investigations and reconciliations. More importantly, even the longer-time scale response has not been subjected to detailed root-cause level understanding. As to the former, the many changes over shorter-range time periods in the time-rate-of change and the sign of the time-rate-of-change in the Figure are simply ignored as if they do not exist. But here is an important point, I think. No explanations for the changes in the sign of the time-rate-of-change at about 50 and 75 years after 10s of decades of the general trend are offered. And I’m using the Figure in this post as only an example of almost all the similar plots that I’ve seen in the Climate Change literature.
Without a root-cause-level basis for understanding of the changes in the general trend I consider the calculated numbers to very likely be more associated with non-physical phenomena and processes than with actual physical phenomena and processes. I tried to write aphysical, for non-physical but my dictionary doesn’t know that word. ‘Physics by Fortran’ would also work.
The changes in the energy transport and storage, and the thermodynamic phenomena and processes that can make such a fundamental change in the long-range response of the GAT are very likely very large. So large that they should be easily identified in the calculated output of the codes. Following specification of the cause of the changes after decades on a constant general trend, we can next ask about the causes of the changes that occur on smaller-range time scales. If an explanation based on the fundamentals of mass and energy conservation concepts is not offered, why should the displayed results be taken to be valid representations of the response of the planet?
All corrections to incorrectos are appreciated.