Vijaywada: A 2023 report by the Intergovernmental Panel on Climate Change (IPCC), the United Nations body for assessing the science related to climate change, warned that global warming will likely exceed the limit of 1.5°C before the end of the century, against what countries had agreed at the 21st Conference of Parties in Paris (COP) in 2015. This increase in temperature will lead to more frequent and intense climate-related risks like heat waves and floods.

“Projections about the future climate can help policy makers prepare for disasters,” says Professor S. Ravichandran from the Climate Studies program at the Indian Institute of Technology in Mumbai.

Projections about the future climate are made through climate models, which aid our understanding of how the climate had changed in the past and how it is evolving. These climate models are critical for India, a country heavily affected by climate change events.

These models talk about climate, which is defined by “the long-term pattern of temperature and precipitation averages and extremes at a location,” and not about weather, which are the everyday changes in temperature, precipitation and so on, explains Ravichandran. Ravichandran studies fluidity, of gases or liquids, such as wind, on high-performing computers–an important part of climate models.

“Climate models have predicted the warming of the planet as a result of anthropogenic CO2 emissions, the resulting rise of the mean sea level, the selective warming of the Arctic, the intensification of wet and dry extremes, etc.,” says Ravichandran.

Edited excerpts from an interview with Ravichandran, on what climate models are, how they work and their limitations, and what they mean for our lives.

What is a climate model?

A global climate model is an approximate digital twin of the planet. It is digital in the sense that it lies inside a computer, and approximate because several processes in the real world are only reproduced imperfectly, if at all, on the computer. The primary function of climate models is to test how the planet might react to external changes, or "forcing". For instance, if we double the CO2 level in the atmosphere, what would happen to the temperature on the planet?

How do climate models work?

Global climate models (GCMs) have interacting components that model the behaviour of the atmosphere, the oceans, the land, the sea ice and life. These components individually model the parts of the earth system, and their interaction models the earth system as a whole. Any given forcing [stress to the system such as more CO2 added to the system] will affect all these component systems as well as their interactions, and GCMs predict the results. They do this by solving the governing equations (basically, conservation laws for mass, momentum and energy) in one, two or three spatial dimensions and time.

What are the different types of climate models and how do they work?

Depending on the complexity to which each of the components is modelled, there are several kinds of climate models. The simplest of these only deal with energy balance in one dimension: energy comes in from the sun radiatively, and the earth reflects some, and absorbs and re-radiates the rest.

The most complicated models include the effects of dozens of interacting processes in three dimensions. For example the rate at which plants and trees reflect or absorb radiation, release water vapour, absorb CO2, etc would have been assumed in simpler climate models but are modelled in more recent, more sophisticated climate models.

What are the challenges in simulating climate at global and regional levels?

Given a big enough computer, all the processes that make up the earth system can be computed. But such a computer would have to be very big indeed. The size of the computer required grows with the size of the system to be computed; after all, the simplest "computer" that simulates the entire earth system is the earth itself. So the fundamental challenge of global climate modelling is to keep the components and processes that are essential and ignore the rest. This is easier said than done, and is an area of active ongoing research.

Global climate models can predict averages well. For instance, if human beings continue to dump CO2 into the atmosphere, the average surface temperature will increase. But this does not itself tell us what happens day-to-day in, for instance, Mumbai. This is the domain of regional climate modelling.

Regional climate models are typically run using inputs from the global climate models, and run on scales much smaller (1,000 km, as opposed to 40,000 km) and over much shorter time horizons (months to years, not decades). The resolution is orders of magnitude better than in global climate models and, depending on the model, more processes are modelled explicitly. The increased resolution requires a lot of computational resources.

Perhaps most importantly, both regional and global models are limited due to the internal variability of the Earth's climate system. Internal variability is the naturally occurring variations in climate from daily weather to multidecadal processes. It is made up of components that are predictable, such as the El Niño–Southern Oscillation [warming of parts of the central Pacific], and unpredictable events because of the chaotic nature of the system.

In other words, the climate system is inherently chaotic and noisy, and rare events occur naturally. Attributing these events to climate change is a statistical exercise.

How do climate modellers overcome these challenges?

Studies of processes--physical, chemical, biological--that affect the climate help improve climate models. Internal variability can be quantified and taken into account by performing ensemble simulations–many simulations of the same events with small changes in the initial conditions or parameters. Improving the resolution of the climate models can also give better results.

However, both ensemble simulations and higher resolution simulations are computationally expensive, and faster climate models that can run on bigger computers are also an avenue of research.

What are the important things climate models have taught us so far?

Climate models have predicted the warming of the planet as a result of anthropogenic CO2 emissions, the resulting rise of the mean sea level, the selective warming of the Arctic, the intensification of wet and dry extremes, etc.

How can climate models help us plan for changes in policy making?

Projections about the future climate can help policy makers prepare for disasters. Climate projections can be used to generate hazard and risk maps for vulnerable regions. For instance, knowing that wet and dry extremes will occur more frequently, cities can invest in water harvesting and flood prevention. States and countries can also plan for the internal migration that will occur as people move to avoid climate impacts.

Projections on the annual and seasonal timescales are even more impactful. Regional climate predictions on the seasonal-to-subseasonal scales--consisting of the aforementioned risk and vulnerability maps, seasonal forecasts and timely early warnings for extreme events--can empower governments and local agencies to take action saving lives and preventing economic losses (this is called the Ready-Set-Go framework).

We welcome feedback. Please write to We reserve the right to edit responses for language and grammar.