You often hear the term “climate model” thrown around in the news or in scientific reports but what does that even mean?
Before we dive in, it’s useful to differentiate between climate and weather. Weather is all of the short-term (minutes to days) variations of the atmosphere including phenomena like wind, precipitation, cloudiness, and humidity, as well as more organized events like thunderstorms and hurricanes.
Climate, on the other hand, is the long-term view. Climate is an amalgamation of the weather we would expect to occur over years, decades, and centuries. Further, it is composed of a vast web of atmospheric, oceanic, land, and biologic processes which interact on a myriad of timescales to set the stage for the daily weather patterns we experience.
Because we can’t double the amount of greenhouse gases in our atmosphere for one day—or millennia—to see what would happen, we created a tool that simulates how all of these processes influence each other.
Since we’ve been able to describe a lot of this understanding in terms of mathematical equations, we’re able to translate it directly into lines of computer code. So, ultimately, a climate model is just a really long computer program that incorporates all of our current understanding of how the atmosphere, oceans, land, and life interact and evolve together.
Thanks to climate models, we can now investigate changes to our climate, such as how increasing levels of greenhouse gases will affect the entire system. We can also choose to model an experiment over various timescales, from decades to millennia. So an observation which would take hundreds of years to unfold in real time can be done with just the click of a button! That and a few days or weeks of processing time…
A representation of a single grid cell in a climate model.
Used with permission from Kris Karnauskas
Each climate model divides the atmosphere and the ocean into a number of blocks, also known as grid cells. These grid cells typically represent a volume of around one hundred kilometers-cubed with each grid cell possessing its own set of characteristics (think temperature and wind speed) that can change over time. The size of these grid cells varies between models and is referred to as its resolution. Similar to a digital camera, the higher the resolution, the more accurate but computationally intensive the model will be.
Climate models also have a temporal resolution known as a time step. Time steps can be any length of time from minutes, days, to years. The shorter the time step, the more detailed the results. However, a smaller time step also necessitates more computing power and time needed to process the model.
An illustration of the changes in resolution of climate models from the Intergovernmental Panel on Climate Change (IPCC) Assessment Reports from 1990, 1995, 2001 and 2007. The resolution progresses from approximately 500 kilometers in the 1990 report to approximately 110 kilometers per grid cell in the 2007 report.
Used with permission from Kris Karnauskas
The time step is one of the main differentiators between climate models and the weather forecasting models used by your local meteorologist. Forecasting models typically have time steps on the order of minutes and hours while climate models have time steps of days, weeks, or months.
Once we set the properties of each grid cell within our climate model computer program, we can use well-established physics equations to determine how each grid cell will interact with its neighboring grid cells in all three dimensions. This process will repeat itself many times depending on the length of time step as well as how far into the future the model is projecting.
The interactions between four grid cells in a climate model. Each grid cell contains various properties (temperature, humidity, etc.) which are exchanged with adjacent grid cells.
Used with permission from Kris Karnauskas
While all models subscribe to the same basic process outlined above, each model varies slightly. For instance, some models use larger grid cells while others may use a different equation to describe the effects of vegetation. In the end, all models in use are able to properly reproduce the past based on known observations.
Since each model is slightly different, the models will provide slightly different results even when coming from the same starting point and time. The variation in these results is referred to as “model spread.” While model spread may seem like an indication that our models are flawed, it often helps us come to conclusions. For instance, if 30 different climate models all arrive at the same result—such as increases in carbon dioxide leading to increased global temperatures—we can comfortably believe in the strength of the relationship.
Ultimately, computer processing power limits the climate models of today more than anything else. Advances in technology have allowed us to shrink grid cells to produce more detailed and, theoretically, more accurate projections than before. We’re also able to cut back on some of the processing shortcuts that were necessary before.
While not perfect, climate models consistently reproduce the past and agree with other climate models. They’re the best tool we have to explore the intricacies of our past, present, and future climate.
By Ryan Harp
Science Buffs 