Comprehensive bias correction of regional climate model boundary conditions for simulation of hydrologic extremes.

Download files
Access & Terms of Use
open access
Copyright: Kim, Youngil
Altmetric
Abstract
High-impact extreme weather and climate events result that threaten society and ecosystems worldwide, from multiple interactions of atmospheric variables linked in dynamic ways. Ongoing global warming necessitates new ways of assessing how extreme events may change in the future. While global climate models (GCMs) have been utilised to assess the simulation of extreme events, the coarse spatial and temporal scales limit their effectiveness at regional or hydrological catchment scales. Regional climate models (RCMs) that use GCM datasets as input boundary conditions are commonly used to improve model predictability for extreme events. Although analyses of extreme events at the regional scale have evolved, systematic bias still exists and is passed onto the RCM simulation through biased boundary conditions simulated using coarser scale GCMs. Despite using various bias correction alternatives to address biases, these approaches often assume that inter-variable bias is not of key importance and that diurnal patterns are properly simulated by the GCM. However, such assumptions can result in substantial anomalies in the simulation of extreme events. Thus, this thesis investigates the impact that several bias correction alternatives can have on RCM boundary conditions with a focus on (1) Precipitation extremes; (2) Spatial, temporal, and multivariate aspects; (3) Multivariate relationships for extreme events; (4) Compound events; (5) Diurnal precipitation cycle; and develops a (6) Software tool for bias correction. The univariate techniques show improvement in precipitation extremes, but the discrepancies in inter-variable relationships are not adequately reduced through RCM boundaries. To address this issue, this study corrects the cross-dependence attributes of these fields, leading to substantial improvements in the statistics used. This study also shows that multivariate bias correction broadly represents the frequency of compound events better. The method is further developed to provide sub-daily corrections that are shown to improve the diurnal cycle of precipitation. Finally, a Python package has been developed as a software tool that simplifies the correction of systematic bias in RCM input boundary conditions. In conclusion, the work in this thesis demonstrates a significant improvement in the regional climate model simulation capacity, thereby enhancing water security and enabling more accurate forecasting of drought and flood events under climate change.
Persistent link to this record
Link to Publisher Version
Link to Open Access Version
Additional Link
Creator(s)
Editor(s)
Translator(s)
Curator(s)
Designer(s)
Arranger(s)
Composer(s)
Recordist(s)
Conference Proceedings Editor(s)
Other Contributor(s)
Corporate/Industry Contributor(s)
Publication Year
2023
Resource Type
Thesis
Degree Type
PhD Doctorate
UNSW Faculty
Files
download public version.pdf 44.21 MB Adobe Portable Document Format
Related dataset(s)