Boundaries of Climate Models

In the last two articles, we discussed the hierarchy of various climate models, starting from simple zero-dimensional Energy Balance Models to the complex General Circulation Models (GCMs) and Earth System Models (ESMs). However, constructing models, running simulations, and then interpreting the results is far from simple. Further, each model and its simulations lead to different pathways. In climate models, one is dealing with many probability distributions, and climate change is a change in these distributions. Finally, the global climate model is just that: a simplified representation of the real world in the form of thousands of equations and millions of variables. The reality may be quite different. Let us discuss these issues in more detail here.

The Basics of Climate Model Building

Most climate models are based on the fundamental laws of physics and chemistry and simulate the process of transfer of energy, gases, and matter through various parts of the climate system. These fundamental laws include the laws of conservation of energy and conservation of mass, Newton’s second law of motion, and the ideal gas law (which defines the relationship between pressure, volume, and temperature of a gas). These models divide the Earth into 3D grids, with each grid having specified variables such as temperature, pressure, density, wind speed, etc. Based on the fundamental laws of physics, each grid is characterised by a set of equations, which tell us how the variables change over a period of time. This period of time is called a ‘time step’ and can vary from a few seconds to hundreds of years, but is typically 10 minutes, which means that the model produces calculations for each grid every 10 minutes. Inter-grid exchange of matter and energy takes place in the model to approximate real-world processes. Each grid is typically 100 km by 100 km, though smaller grids are being built with a rise in computational power.

The climate model solves the equations for each grid based on today’s values. This is repeated for the next time period, and these iterations continue until you reach a point in the future that you want to analyse and predict. GCMs and ESMs solve millions of such equations to predict the climate over large periods (100 years or up to the year 2100). A climate model is run and tested through a process called ‘hindcasting’. This requires the specification of initial conditions (for example, many models specify the pre-industrial conditions of 1850 as a starting point to measure the effects of greenhouse gases on the rise in temperature over the past centuries). The model run gives us outputs of various important variables, such as changes in temperature or sea levels, etc. Various simulations can also be run with the model to help guide decision-makers in policy-making.

It is important to note here that the time step is an important constraint while running models. Many GCMs use a 10-minute time step to enable them to deal with high-resolution detail such as local weather (e.g., hurricanes). What is gained in high-resolution detail is lost in predictions that are not very long into the future. Ultimately, a balance has to be struck between geographical resolution, timeframe, and computing resources.

Climate models are, of course, tools in a climate scientist’s armoury and are used to study the climate system and climate change. They are, however, only the first step. There are still many issues that need to be resolved, such as greenhouse gases, clouds, rainfall, and thunderstorms, which are not captured by the equations in each grid box. The way to get around these is through ‘parameterisation’, which we will discuss below. There is also the issue of modelling the oceans, their surface temperature, salinity, and ocean currents.

It may be noted that the climate system and climate change are very different: the climate system is the climate of all the interconnected parts, such as the Earth, atmosphere, land, oceans, and ice sheets, as it exists. As David Stainforth has suggested in his book Predicting Our Climate Future, climate change has many aspects, ranging from climate prediction to extrapolation, probabilities of outcomes (as he says: “Climate change is a one-shot bet”), and non-linearity (similar to sensitivity to initial conditions, or the butterfly effect). Let us discuss these issues in more detail in Part II.