Search site

UK maps based on global temperature change in detail

This page gives more information about using UK maps based on global temperature change, some things to be aware of when using them and technical details about how the maps are produced.

What should I use them for?

These maps can be used to understand and present projected climate change for the UK in relation to a specified global temperature rise. They are useful to inform adaptation assessments in the UK in the context of global mitigation targets.

Things to be aware of

A single map shows the projected changes of a specified variable within the UK associated with a projected single global temperature rise and a single, user-specified, Probability level. Multiple maps at different probability levels are needed to illustrate the uncertainty across the projections.

As a general guideline it is suggested that customisable maps for the 10% to the 90% probability levels are useful to support decision-making. Outside the 10% to 90% probability range, care should be exercised when using the maps as the robustness of the projections decreases considerably.

Technical information

This section explains in more detail, the methodology used to create the UK maps based on global temperature change.

The data files on which these maps are based were generated using a corresponding annual global temperature change relative to pre-industrial times rather than selecting a specific emissions scenario. The 10,000 model variants used were selected on how well they fit the given value of global temperature change.

During the production of UKCP09, sampled values of changes in global annual mean temperature relative to a 1961-1990 baseline were provided for each of the 10,000 model variants. A linear regression, trained on results from 10,000 runs of the Simple Climate Model used in UKCP09, has been used to convert these anomalies into changes relative to pre-industrial times.

A given level of global annual mean temperature change relative to pre-industrial times was selected (such as 3ºC). The probability maps were produced by identifying the emissions scenario for which most model variants are consistent with the specified level of global temperature change. The analysis showed that these were B1, B2, A1B, and A1FI for global temperature changes of 2, 3, 4 and 5ºC, respectively. Data from the time period 2070-2099 was used, although this choice is believed to have a second order effect on the results.

A set of weights was given to the 10,000 model variants based on how well they match that level of global annual mean temperature change relative to pre-industrial times. These weights add up to 1.

The variable of interest (e.g. summer total precipitation) was extracted for all 25 km grid boxes and all 10,000 model variants at the meaning period of interest. For each grid square:

· The model variants in the variable of interest were sorted into ascending order.

· The model variant IDs were recorded for the sorted variable of interest.

· The weights were re-ordered by model variant ID to match those in the sorted variable of interest

· The re-ordered weights were used to calculate a cumulative array of weights (note these re-ordered weights still sum to 1). At this point the weights provide the y-values of a cumulative distribution function for the variable of interest

· The graph of weights vs variable of interest was then interpolated to the probability level of interest and the result stored by 25 km grid square.