Correcting systematic bias in derived hydrologic simulations – Implications for climate change assessments

Abstract

Abstract Quantifying climate change impact on water resources systems at regional or catchment scales is important in water resources planning and management. General circulation models (GCMs) represent our main source of knowledge about future climate change. However, several key limitations restrict the direct use of GCM simulations for water resource assessments. In particular, the presence of systematic bias and the need for its correction is an essential pre-processing step that improves the quality of GCM simulations, making climate change impact assessments more robust and believable. What exactly is systematic bias? Can systematic bias be quantified if the model is asynchronous with observations or other model simulations? Should model bias be sub-categorized to focus on individual attributes of interest or aggregated to focus on lower moments alone? How would one address bias in multiple attributes without making the correction model complex? How could one be confident that corrected simulations for the yet-to-be-seen future bear a closer resemblance to the truth? How can one meaningfully extrapolate correction to multiple dimensions, without being impacted by the ‘Curse of Dimensionality’? These are some of the questions we attempt to address in the paper.