Improving hydrological climate impact assessments using multirealizations from a global climate model

Frederiek Sperna Weiland*, Dana Stuparu, Renske de Winter, Marjolijn Haasnoot

*Corresponding author for this work

Research output: Contribution to journalArticleAcademicpeer-review

Abstract

Many flood risk assessments account for uncertainties in future anthropogenic emissions by considering multiple representative concentration pathways (RCPs). The imperfect knowledge and representation of the climate system is considered with multiple global climate models (GCMs). Yet, uncertainty introduced by incomplete representation of natural variability is also relevant but not always accounted for. A set of realizations provides improved insights in natural variability presented by the GCM. This study explores the potential of using a set of realizations from a single GCM-RCP combination instead of single realizations. We use (subsets of) 16 realizations from EC-Earth for RCP8.5 and focus on three locations along the Rhine. We use a single GCM-RCP combination to avoid the interference of additional sources of uncertainty. We find that projected changes in future river flows highly depend on the realization chosen. Individual ensemble members provide different changes for annual mean flow, extreme flows, and regime shift. By increasing the number of realizations and combining their annual maxima in extreme value analysis, future projections of flow extremes converge. We conclude that a single ensemble realization gives overconfident and possibly erroneous projections. In climate science, this is well studied; however, in flood risk assessments, it is still often neglected.

Original languageEnglish
Article numbere12787
JournalJournal of Flood Risk Management
Volume15
Issue number2
DOIs
Publication statusPublished - Jun 2022

Keywords

  • climate change
  • flood risk
  • GCM
  • natural climate variability
  • Rhine

Fingerprint

Dive into the research topics of 'Improving hydrological climate impact assessments using multirealizations from a global climate model'. Together they form a unique fingerprint.

Cite this