Climate Models: What They Show Us and How They Can Be Used in Planning

Submitted by BRACC Hub | published 29th Jan 2022 | last updated 27th Jul 2022
Atmosphere, ocean and land surface processes simulated in GCMs (source: US National Climate Assessment 2014)

Atmosphere, ocean and land surface processes simulated in GCMs (source: US National Climate Assessment 2014)


The climate conditions that we experience are the result of complex interactions between processes occurring in the atmosphere and in the oceans. These processes operate at global and local scales and are influenced by other factors, including the land surface, polar ice sheets and the sun. This is why different parts of the world experience different climates. Global Climate Models (GCMs) are computer models that attempt to capture and simulate all these processes, based on our current knowledge.

This article summarizes a guide from the Future Climate For Africa programme is a primer on global climate models (GCMs) and how to use them in planning. It covers:
  • what GCMs are
  • how GCMs work
  • how we know if GCMs are reliable
  • what GCMs can tell us about future climate
  • how we can use GCMs for planning

*This weADAPT article is an abridged version of the original text, which can be downloaded from the right-hand column. Please access the original text for research purposes, full references, or to quote text. 

How do GCMs work?

Global Climate Models are run on supercomputers at a number of centres around the world, including the Max Planck Institute in Germany, the UK Met Office Hadley Centre, and the National Oceanic and Atmospheric Administration (NOAA) in the USA. The models use physical laws and mathematical equations that reflect our understanding of atmospheric and oceanic processes.
  • To test how well GCMs capture climate processes, they can be validated in various ways. 
    •  This might be done by testing based on what we know about past climate patterns, for example from recorded observations on temperature and rainfall conditions. 
    • We can also test how well the model simulates key largescale weather patterns. 
  • No predictions of the future can be made with 100% certainty. 
    • To take a the possible futures into account, scenarios – or plausible socioeconomic futures – are used.
    • Models are typically run under different scenarios to give a range of potential future climate conditions.
    • As well as the uncertainty in human activities, our understanding of the smaller-scale processes that affect climate is incomplete. 
  • GCMs model processes and interactions at global scale.
    • As this is a complex task, the models are built to emphasise particular trends (averages of weather over longer-term time periods).
    • Because they work at global scale, the resolution of GCM projections is typically coarse.
    • As computing power has increased, the detail and resolution has improved over time. 
  • Another key consideration with regard to GCMs is that temperature is easier to project than rainfall. 

How Can We Use GCMs for Planning?

When we plan for the future, we take a variety of information into account. This might include population projections, for example, or anticipated demand based on economic growth trajectories. In order to take the impact of climate change into account, we first need to understand what aspects of climate might affect planning decisions. With that understanding we can move to the second stage, which is to seek the relevant information in the projections.

If the onset of the rainy season is going to continue to be later than in the past, a 5-year agricultural plan may wish to consider a strategic move to promote switching to early maturing varieties or crops that can withstand lower water availability. Our understanding of current weather patterns can be used to contextualise future projections. Current weather patterns will vary throughout a country.

Medium-term planning can take into account the effect of likely future conditions based on projections, in conjunction with current knowledge of weather conditions and how they vary from place to place. This enables identification of robust strategies for what crops to promote and what farming techniques to investigate. The specifics of decisions – in terms of exactly what to plant and when in different places – can be part of shortterm planning (e.g. annual), and can be informed by weather information such as seasonal forecasts. However, knowing the likely longer-term situation can enable adaptive planning to meet national strategic goals, such as long-term food security and livelihood resilience.

Longer-term decisions that may result from such planning include training staff in the use of early maturing crops and the practices required to farm them optimally, informing seed-breeding activities and technology development, and setting supply chains in place.

The nature of planning decisions that we make over the medium term is different from those that we make in the short term. So what we require from climate projections is different from what we need from shorter-term weather predictions. The resolution provided by GCMs is useful to inform medium- to long-term planning decisions. Over these time frames, for example, whether the temperature will increase by 2.1°C or 2.2°C is not usually as important as knowing that there will be an increase in temperature.

The Myth of High-Resolution Climate Projections

A common myth is that high-resolution or “downscaled” climate projections are better than the coarser projections from GCMs. Climate projections take into account large-scale processes that affect weather systems. Because they are projecting far into the future, they are driven by scenarios (of greenhouse gas emissions) that are never likely to represent exactly what will unfold in the future. Although there are various methods of downscaling, they are all contingent on starting with outputs from GCMs. When model data are further manipulated in another modelling process, there is a risk that the uncertainty can increase. The outputs of downscaled models appear to be more precise as they show differences on a smaller scale – but that can give a false level of confidence. The likely future states projected by GCMs are typically sufficient to inform the kinds of planning decisions that are made over the same medium to long time frames. 

 The cascade of uncertainty in projecting future climate (source: RiskChange)