Selecting representative climate models for climate change impact studies: an advanced envelope‐based selection approach

Hester Biemans,Arthur F Lutz,Philippus Wester,Herbert W Ter Maat,Walter W Immerzeel,Arun B Shrestha

doi:10.1002/joc.4608

Abstract

ABSTRACTClimate change impact studies depend on projections of future climate provided by climate models. The number of climate models is large and increasing, yet limitations in computational capacity make it necessary to compromise the number of climate models that can be included in a climate change impact study. The selection of climate models is not straightforward and can be done by following different methods. Usually, the selection is either based on the entire range of changes in climatic variables as projected by the total ensemble of available climate models or on the skill of climate models to simulate past climate. The present study combines these approaches in a three‐step sequential climate model selection procedure: (1) initial selection of climate models based on the range of projected changes in climatic means, (2) refined selection based on the range of projected changes in climatic extremes and (3) final selection based on the climate model skill to simulate past climate. This procedure is illustrated for a study area covering the Indus, Ganges and Brahmaputra river basins. Subsequently, the changes in climate between 1971–2000 and 2071–2100 are analysed, showing that the future climate projections in this area are highly uncertain but that changes are imminent.

Highlights

Climate change impact studies depend on projections of future climate provided by climate models
Due to their coarse spatial resolution, outputs from general circulation models (GCMs) are usually directly downscaled to higher resolution using empirical–statistical downscaling methods or are used as boundary conditions for regional climate models (RCMs), with their outputs being subsequently downscaled to a higher resolution
Intergovernmental Panel on Climate Change’s (IPCC) fourth Assessment Report (IPCC, 2007), contains outputs from 25 different GCMs, whereas the CMIP5 archive (Taylor et al, 2012), which was used for the fifth IPCC Assessment Report (IPCC, 2013), contains outputs from 61 different GCMs

Summary

Introduction

Climate change impact studies depend on projections of future climate provided by climate models. Due to their coarse spatial resolution, outputs from general circulation models (GCMs) are usually directly downscaled to higher resolution using empirical–statistical downscaling methods or are used as boundary conditions for regional climate models (RCMs), with their outputs being subsequently downscaled to a higher resolution. The downscaled outputs are used to assess future climatic changes and to drive other sector-specific models for climate change impact studies. Outcomes from these studies are used by policymakers to support decisions on climate change adaptation measures. These GCMs often have multiple ensemble members, resulting in an even larger number of available model runs

Objectives

Methods

Results

Conclusion