As evidence of climate change began accumulating in the mid-20th century, computers were becoming increasingly commonplace. Scientists were faced with the uncertainty of a transforming climate with no way to observe its effects until it was too late. They began developing computer programs that would simulate these possible changes, and it quickly became clear just how ambitious this project would become. While the most basic simulation can run on a spreadsheet, the most complex requires massive, tennis-field sized supercomputers running 14 000 trillion calculations every second. This begs the question: is it worth it to dedicate all these resources to such a task? Can the chaotic and convoluted processes that influence our climate really be reduced to formulas on a computer?
Translating the World into Math
Although climate models can be programmed to simulate small parts of the Earth, they can also be designed to simulate the entire planet. Hundreds of scientists collaborate constantly to develop this software, translating every physical principle that affects the climate into equations, which can then be written as code and simulated. These include greenhouse gas emissions, rain, clouds, volcanic activity, and vegetation. With the effects of human development further increasing the necessary parameters, creating this software becomes a daunting task for scientists.
For the simulation to function in a reasonable amount of time, it divides the Earth’s surface into a grid of squares and then simulates the conditions of each square, combining the results at the end. Dividing the world up into smaller squares yields more accurate data, but takes a disproportionately longer time to run, so scientists must always find the right balance between time and precision when running their experiments.
The Past Precipitates the Present
With the environmental processes that affect our climate being as elaborate as they are, is it really possible for a computer program to simulate them accurately? This is a constant concern for scientists, but the evidence indicates that climate models can actually be extremely effective.
One test that scientists conduct is called a ‘hindcast,’ which is a simulation of the climate conditions of a period of time in the past using certain parameters and then comparing those results with observed values from that time period. If the simulated and observed values match, this indicates that the program could be useful for predicting future changes. Scientists have also compared the results of old simulations from between 1970 and 2007 to observations from 2017 and found that most of the models were accurate. Models today are significantly more complex and powerful, so scientists are confident in their reliability.
Tempestuous Troubles
Climate simulations are an incredibly impressive achievement, but there are still many improvements that could be made. One of the biggest challenges that scientists face is the simulation of clouds, which can behave erratically and have dramatically different effects depending on various other factors. Some simulations can yield emergent factors which have not been programmed directly into the system and are accounted for as a result of interactions between other processes, but many other mechanisms still need to be implemented or improved. Despite their limitations, climate models stand as a testament to what hundreds of scientists can accomplish through hard work and collaboration, and provide us with the knowledge needed to prepare for our future.
References
https://www.carbonbrief.org/qa-how-do-climate-models-work
https://www.carbonbrief.org/in-depth-scientists-discuss-how-to-improve-climate-models
