A starting point is the future emissions scenarios. What would you like to explore? Business-as-usual or a moderate path with reduced emissions? Scenarios for the evolution of greenhouse gas emissions are provided by Representative Concentration Pathways (RCPs) or Shared Socioeconomic Pathways (SSPs), as prepared for the IPCC-assessments.
Global Climate Models (GCMs) are then applied to study how these emissions affect the climate. The various GCMs differ in how they simulate processes of the Earth’s climate, and thus also in their sensitivity to the greenhouse gas forcing, i.e. how the GCM’s climate respond to changes in emissions. Each GCM has its own strengths and weaknesses.
Future climate change cannot be supported by evidence from a future that does not exist yet, therefore the evidence in support of a hypothesis about the future comes in the form of model projections. In climate science, projections are regarded as well-supported if a sufficiently diverse set of models agrees on the results. Ensemble statistics, e.g. the average or median result from of a number of models, are therefore considered more reliable and robust than using a single model projection.
GCMs have a typical horizontal resolution of several hundred kilometers, and since climate impact processes occur at much smaller scales, a further downscaling of the GCM information is useful. This is performed by Regional Climate Models (RCMs), which nest into the GCM and provide finer scale information of 50 km or less. CMIP (Coupled Model Intercomparison Project) and CORDEX (Coordinated Regional climate Downscaling EXperiment) are two global collaboration programs under WCRP (the World Climate Research Program) of WMO (World Meteorological Organization) that coordinate GCM and RCM ensembles respectively.
Hydrological and other impact modelling applications are sensitive to climate model bias. Bias is a systematic deviation from observed statistics, e.g. consistently too wet or too warm climate in a certain region of the world. It is common practice to apply bias adjustment, a statistical method that removes various errors from the climate model output so that the model data become more similar to observed data (locally gridded or point data from gauges).
Climate indicators of weather and water variables are produced at the end of the computation chain, to show how the climate changes between different time periods and for a given emission scenario. Since climate is an average of the weather over a long period of time, indicators are often produced as an average over thirty years.
Each model step in the production chain includes uncertainties, with the main uncertainty within each emission scenario arising from difference between the climate models. Therefore, an ensemble of projections (multiple RCMs and/or GCMs) is used to account for the spread of possible climate impacts in the future. The uncertainties inform about the reliability of the climate projection. A result is more robust when many models agree on the sign and the magnitude of climate change. The exact future still remains unknown but the model results indicates tendencies and future risks generated by climate change.
