Modelling and Uncertainty - Climate Sensitivity

In our second blog sharing insights from our climate change experts about uncertainty in modelling future flood risk, we discuss how different climate models can generate different outputs even when they are driven by the same input climate scenario.

Uncertainty in modelling future flood risk: climate sensitivity

Projections of future climate are associated with many uncertainties that affect our confidence in assessments of future flood risk. Although uncertainty in climate models results from many factors, one such uncertainty is the question of climate sensitivity.

Climate sensitivity is a measure of how much global warming we would expect to occur for a given increase in greenhouse gas concentrations. This has implications for all aspects of climate change, from sea-level rise and ice sheet collapse to wildfires and drought. Despite the importance of understanding how much the Earth will warm in the future, estimates of the climate’s sensitivity to greenhouse gas inputs vary considerably.

Typically, estimates of climate sensitivity are quantified by how much the global average temperature would rise following a doubling in atmospheric CO­2 concentrations. This can be estimated in several ways: by assessing how much the Earth has already warmed due to CO­2 emissions, by looking at how global temperatures have changed due to variations of atmospheric CO­2 in the past, or by simulating how temperature would change following a doubling of CO­2 using climate models.

Due to uncertainty in how we expect the Earth system to respond to increases in greenhouse gases, such as those of CO­2, there are a broad range of estimates as to what the Earth’s climate sensitivity is. For instance, the IPCC estimates that the temperature increase that would eventually result following a doubling in CO­2 concentrations likely falls anywhere in the range of 1.5°C to 4.5°C (IPCC, 2021) . Additionally, the climate sensitivity simulated by individual climate models also spans a wide range of values. This is especially true in CMIP6 models, where the climate sensitivity ranges from 1.8°C to 5.7°C.

The different climate sensitivities between models will impact estimates of future flood risk generated from their output. We explore the potential mechanisms of this further below.

From climate sensitivity to precipitation

To determine the effect of a model’s climate sensitivity on the amount of precipitation it simulates, first we must consider its effect on simulated global temperature. Several studies (e.g. Huusko et al., 2021 ) find that climate sensitivity and average global surface temperature are positively correlated in CMIP models. This means that models with a higher climate sensitivity will simulate higher global surface temperatures for a given climate change scenario. However, this varies somewhat when we look at different regions. For instance, while the temperatures in the equatorial regions and the southern tropics are strongly related to the global change, there is a weaker relationship in the North Atlantic and the European sector of the Arctic Ocean.

There is also a positive relationship between modelled changes in average global surface temperature and changes in extreme precipitation, which is particularly relevant for flood. This is governed by the Clausius–Clapeyron relationship, where the moisture-holding capacity of the atmosphere increases with air temperature at a rate of ~7% per °C of warming (see also a previous JBA blog on this). We can see this relationship illustrated in figure 1, which shows the analysis of a selected CMIP6 model.

Figure 1: Scatter plot of the change in global land median annual maximum precipitation (%) against changes in global surface air temperature (°C) for the CMIP6 simulations of the MRI-ESM2-0 climate model. The changes are calculated between two 30-year periods: 1980-2009 and a rolling 30-year window starting at 2015-2044 and ending at 2071-2100. Each dot represents the change for a different future period and the different colours show this under different SSP scenarios. The dotted line shows the Clausius-Clapeyron relationship, i.e. the 7% increase in water vapour per 1°C.

While the Clausius-Clapeyron relationship holds well at a global scale, it is modified in some regions where local processes and/or large-scale atmospheric circulation changes can either intensify or diminish the relationship. Indeed, in some regions precipitation is projected to decrease in a warming climate, with notable examples being the Mediterranean and the southwest of the African continent, both related to projected shifts in large-scale circulation patterns.

Given the relationships between climate sensitivity and temperature and then temperature and precipitation, we can infer a causal chain between climate sensitivity and changes in precipitation at the global scale. This is thanks to the fact that changes in temperature drive both of these variables.

The result is that, for a given scenario and time horizon, higher climate sensitivity models will simulate higher precipitation amounts due to greater increases in temperature, at the global scale and all else being equal. However, for a given global warming level (for instance at 2°C above pre-industrial temperatures), we do not expect global precipitation to vary much by different models and their different climate sensitivities, since all the models will show the same degree of warming.

Implications for future flood risk?

Given the relationship between temperature and precipitation, the use of higher climate sensitivity climate models could lead us to estimate a higher intensity of flood hazard compared to a lower sensitivity model, when considering global-scale impacts for a given scenario and time horizon. Depending on locations and vulnerabilities, this could translate into greater flood losses. Therefore, sampling climate models with a range of climate sensitivities is one way to sample the uncertainty in future flood intensity. This would allow for an understanding of the range of potential future flood risks that could result depending on how the climate responds to greenhouse gas emissions.


A model’s climate sensitivity will affect its projections of precipitation, and therefore the flood risk that is estimated using its output. At the global level, models with higher climate sensitivity project larger increases in precipitation extremes for a given scenario and time horizon. This is thanks to climate sensitivity modulating both temperature and precipitation amounts. Therefore, one way to sample the uncertainty in flood risk for a given scenario and time horizon is to select models that span the range of projected climate sensitivities.


Huusko, L.L. et al. 2021. Climate sensitivity indices and their relation with projected temperature change in CMIP6 models, Environmental Research Letters, 16(6), p. 064095. Available at:

IPCC. 2021. Climate Change 2021: The Physical Science Basis. Contribution of Working Group I to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change. Edited by V. Masson-Delmotte et al. Cambridge, United Kingdom and New York, NY, USA: Cambridge University Press.