Arguments
Climate and economic models – birds of a different feather
Posted on 19 December 2013 by John Abraham
Climate computer models (computer predictions of the Earth's climate) are one of the critical tools we use to determine the future of our Earth and the likelihood humans are having an impact on the environment.
So, what the heck is a climate model? If you are a scientist, it is a computer construct that expresses conservation of mass, energy, and momentum over a large number of grid boxes that cover the earth, including the atmosphere and the ocean. For people who aren't scientists, it is best to think of them as virtual-reality computer programs that predict the future.
Climate models are not the only reason scientists are so certain that humans are changing the climate. In fact, I argue they aren't even the best evidence. We also know humans are causing climate change because we can look at how the deep-past Earth has changed; we can extrapolate into the future. Additionally, we have recent instruments (since approximately 1880) that allow us to relate human emissions to the increase in temperatures. Climate models aren't limited to temperature. They allow prediction of species, precipitation, cloud cover, ice, and many other quantities.
The reason scientists spend so much time using models is because they allow us to make future predictions that are testable. Climate models can predict cooling of the stratosphere, they can predict loss of arctic sea ice, they can predict changes in weather patterns, they can predict acidification of the ocean; climate models can be used to study what if scenarios in a way that our other information sources cannot. In fact, computer models have an excellent track record predicting changes in the Earth's climate before they were detected by measurements. In this way, measurements have confirmed climate models.
The accuracy of a climate model is dictated in part by how many calculation points are used to cover the Earth. A fancy term for this is grid resolution. The more grid boxes used to subdivide the ocean and atmosphere, the better the results. Today's most powerful computer models are running 24-hour calculations of enormous climate models, making calculations over millions and millions of grid boxes. Even with today's powerful supercomputers, there is a limit to how large the number of calculation points can be and a limit to how much data can be stored.
Another limit to the accuracy of climate models deals with processes that do not obey basic laws of the universe (conservation of mass, momentum, and energy). Examples of processes that fall into this category are cloud formation, fluid turbulence, ocean oscillations, dispersion of particulates, reflection of sunlight by aerosols, transfer of water between oceans and the atmosphere, etc. Here, best guesses have to be used which are broadly called parameterizations. Parameterizations are equations that reflect our best understanding of the underlying physical process.
One of the nice things about climate models is that they are testable and they follow known governing laws of the universe that cannot be broken. The climate follows these laws now, has followed them in the past, and will follow them in the future.
At the recent American Geophysical Union meeting, I met a young climate modeler, Catalina Oaida. Her work deals with particulates impact in the cryosphere and her description of models in her discipline is a great view of how insiders view these computational tools.
"My research involves using what we call regional climate models to assess the impact of dust and black carbon deposited onto the mountain snowpack. Regional climate models are very similar to general climate models [GCMs - used in the IPCC]; they are based on the same universal principles of physics and conservation, but they have a higher resolution that allows us to study in greater detail, and often with better accuracy, the processes taking place at local and regional scales. What is great about physically-based regional climate models is that they allow us to study these processes as they have occurred in that past, and also allow us to assess how this human-nature system/interaction might change in the future."
My colleagues in business are not as fortunate. Worldwide and national business and economics do not follow universal governing laws. They follow human actions that are incredibly hard to predict.
To me, a climate model is the laboratory experiment of climate science.
You can grow microbes in a Petri dish and study their inner workings with relative ease; you can study fruit flies, or mice, or even primates. You can roll balls down slopes, you can smash streams of particles traveling near the speed of light into each other and see what comes out, you can mix two substances together and see what happens. And you can do these - and many other things besides - over and over and over again.
You can't go out and build even a single planet identical to Earth and run through decades of climate history in an afternoon (or over a weekend), never mind dozens.
So you need a computer-generated simulation. You need a model.
It's not perfect, but like so many things in life, it's good enough.
To me, a model is nothing more than a hypothesis where all the supporting assumptions have been made so crystal clear and explicit that the whole thing can be put in a computer which can then calculate the consequences of the hypothesis. If the results are different from the real world, you revise the hypothesis and/or supporting assumptions.
The results will always be different from the real world, so this process never ends. But that doesn't matter. At some point the results are good enough to be useful. We have been at that point for some time now.
Ah yes, the old conversation of mass issue.
I follow the climate change issue because it is one of the issues that relates directly to the many unacceptable economic developments that have occurred in the global economy because of the success of people with unacceptable attitudes, people who greedily pursue more benefit for themselves any way they can get away with.
Economic models could be more reliable if they included more rational evaluations, including the potential influence of unethical uncaring people pursuing short-term personal benefit through unsustainable activities like the burning of fossil fuels and deliberately discounting or ignoring or not caring about the harm their pursuits will create or the risk of harm of their pursuits. The economic evaluations could also be more accurate if ‘value’ was not ascribed to activities that are fundamentally of no future worth, activities that cannot be benefited from in the future. They could also be more accurate if the evaluations did not discount future risk or challenges.
Many current economic evaluations and modeling are fundamentally flawed. Some evaluations actually try to justify the more fortunate in a current generation gaining benefit from activities that cannot be benefited from by the entire population, cannot be continue to be benefited from by people in the future, and which create added challenges for others (particularly for those in the future). The evaluations compare what their creators believe is the lost opportunity for benefit by a current generation (meaning themselves), to what they believe the costs others will face will be. That is fundamentally irrational and extremely callous, basically saying it is OK to benefit in a way that others can't and that creates problems for others as long as the trouble the evaluator believes their actions create are less than the benefit they believe they get. The more irrational evaluations go one step further and ‘discount the costs others in the future face' by applying the economic principle of 'net-present-value'.
So, the real problem with economic models and evaluations is the way the 'value' of things is measured. Evaluations that excuse damage or the risk of damage and assign value to unsustainable activity are bound to be ‘inaccurate’. These evaluations have led to the development of unsustainable and damaging activity that is a ‘constant struggle to sustain, let alone grow’ with battles and tragedy inflicted by those pursuing more of the limited unsustainable benefit they want to get the most of for themselves.
The desire to ‘sustain or grow the benefit obtained from unsustainable and damaging activities’ is almost certainly behind ‘popular’ attempts to claim the climate science and its models are flawed or inaccurate or can be ignored.
Admitting the unacceptability of unsustainable or damaging activity is the first step to more rationally evaluating the economy. Admitting the unacceptability of development that has led to the current global economy is the first step in recognizing what activities need to be ‘valued’ and what activities need to become ’worthless’. Only then can there be any expectation of reliable economic model predictions and evaluations, and the development of truly sustainable economic growth.
"Another limit to the accuracy of climate models deals with processes that do not obey basic laws of the universe (conservation of mass, momentum, and energy)."
Perhaps this might be more elegantly phrased. I sincerely doubt that even in the most chaotic systems, matter, energy and momentum are getting created or destroyed, even if it seems that way in the macro perspective of a climate model.
This article does raise a question in my mind, though ... if most models are being run on (very expensive) supercomputers 24/7 based on one small grid square at a time, it seems to me that this would be an ideal candidate for massively parallel distributed computing, if you could find programmers clever enough to write the software for it. I for one would be willing to volunteer the use of any spare cycles in a good cause, and I think most of the denizens of SkS feel the same way.
Best wishes,
Mole
Old Mole - I agree it could be phrased a lot better.
As to distributed climate models, Climateprediction.net is already doing that.Sign up.