Why We Are Bad at Predicting the Future Climate and Why It Doesn’t Matter

Different future emissions scenarios  – (IPCC, 4th AR)

In college, climate was a textbook chapter, a chain of facts I had to study. At Columbia climate became a statistic, an endless set of numbers in which I had to find and interpret patterns. Often my classmates and I did not agree on one interpretation or the patterns did not make any sense to us at all. Climate was not a certainty anymore and everything I believed I knew about the climate turned out to be a simplified approximation of what actually happens in this complex and chaotic system.  I started wondering about our ability to produce “truths” on climate change and the very nature of knowledge in general. I followed up on these thoughts throughout the program and finally turned them in a master’s thesis on uncertainty in climate modeling and its impacts on policy making.

Of course, this problem is not exclusive to climate science. For every topic, the closer we get to the margins of established knowledge, the more complicated and uncertain facts become. However, when it comes to climate science, the question of uncertainty has a particularly prominent place in the debate over the appropriate response to the human-induced climate change we are observing. The main problem of climate prediction is not so much a lack of knowledge of the workings of the climate system, nor our ability to model these processes, even though both are by no means perfect. There are two main problems, one of which is our inability to determine climate sensitivity, meaning the level of impact a certain change in greenhouse gas concentration actually has on the average global temperature. The other is the assumptions modelers have to make about the future development of society and how the amount of greenhouse gas emission might change in the coming decades, which depends on a multitude of factors like population development and possible lifestyle changes, each with their own limited predictability. If modelers assume a high level of climate sensitivity and expect a strong increase in CO2 concentrations in the next decades, then the temperature increase within this century that their model will give them will be higher than another model based on more optimistic views on emission pathways and less intense climate sensitivity .

While we can potentially strengthen our understanding of climate sensitivity and close remaining knowledge gaps of the climate system, we cannot significantly reduce the subjectivity of social assumptions, which explains why ever more research and advancements of climate models has not resulted in clearer projections.

So, how can I think that this does that not matter? In order to explain that, we have to ask why we feel the need to make climate predictions in the first place. In our knowledge-based society we do not just prefer guidelines, but actively look for legitimizing reasons for our decisions in front of voters, investors, our own moral convictions, etc. Projections and scenarios help us make decisions that will affect the future, and for that it does not actually matter how accurate the predictions are or even if they are contradictory. It only matters that they are reasonable, which often translates into being legitimate in the eyes of the audience. As long as we will be able to convince our future selves and others that we made the “right” decision based on the information that was available, climate models and other predictive tools will have fulfilled their purpose no matter the actual accuracy of their predictions.