The future of climate modelling

As faster supercomputers with more processing power are developed, harnessing this power and speed for the benefit of improving climate projections is the dream of climate scientists.
The reality is there will never be enough speed or capability to infinitely improve climate models in all aspects, so trade-offs are inevitable and frequent. Here we explore the different elements of climate modelling and how each can be best utilised for the most effective outcome. 


The three axes of climate modelling

Three key elements of climate models are a focus for scientists:

  • Increasing the spatial resolution of models to resolve more detail;
  • Increasing the realism of the climate system within models to include more of earth’s processes (which makes models more complex);
  • Increasing the number of individual model runs (each with a slightly different starting set-up) to help understand uncertainty, especially around those extreme events which only happen infrequently.

Development in any one of these three directions implies an increase in the computational cost of the models. Even with the next generation of high-performance supercomputers, climate scientists will need to balance these three aspects.

 

Increasing model resolution

Each climate model covers the earth with a digital mesh. The finer the grid squares of the mesh, the finer the resolution.

Higher-resolution models (with resolution down to the scale of a single kilometre) provide the ability to model the earth in unrivalled detail.

This increased resolution can capture the effect of physical features of the earth’s surface or vital processes which happen at smaller spatial scales, like the formation of thunderstorms or clouds. In coarser-resolution models these processes are not resolved and the 'bulked-up' effects must be represented in a simplified way. 

However, running climate models of this resolution is computationally expensive, especially when compared with coarser models.
 

Increasing the range of processes represented within computer models

Computer models of the earth system try to capture as much realism as possible, including:

  • important aspects of the carbon cycle;
  • how ice sheets respond to a warming world;
  • and the microphysics of clouds.

Increasing the number and detail of these processes within a climate model leads to more robust projections of our future climate and allows us to answer more detailed questions about possible outcomes, such as different pathways to net zero or possible tipping points within the climate system. but also makes the models more complex.

Increasing the complexity of climate models increases costs, but this cost is relatively much lower than the costs of increasing model resolution.

Increasing the number of ensemble members

The weather and climate we experience is just one outcome of a huge range of possibilities. This means if you run a climate model once, you only see one potential outcome and we don't know how likely that outcome is.

By running a model many times, each time with a slightly different starting set-up, you can sample a broader range of the possible future outcomes.

This is really important for understanding how climate change might influence the likelihood of extreme events – which are relatively rare by their nature.

Running more individual model runs is computationally expensive so the processing time can be far longer, especially when combined with high-resolution models.

What is the future of climate modelling?

The Met Office Research and Innovation Strategy has an ambition to expand on all three axes of climate modelling. Expanding all three axes will provide benefits that cannot be realised by focusing on the development of one or two axes alone.

For example:

  • increased resolution will bring more confident projections of regional changes to allow societies to build resilience to the changes to come;
  • more comprehensive representation of the carbon cycle and short-lived greenhouse gases, such as methane, will allow the assessment of alternative pathways to a net zero future;
  • and larger ensembles of model runs will give a clearer idea of the future risks from extreme weather.  

Climate scientists will need to assess the relative merits of each and establish the most effective and efficient trade offs to answer the key questions to build a climate-resilient future, guided by the needs of decision makers.