Impact of downscaled rainfall biases on projected runoff changes

Abstract

Realistic projections of changes to daily rainfall frequency and magnitude, at catchment scales, are required to assess the potential impacts of climate change on regional water supply. We show that quantile-quantile matched (QQM) bias-corrected daily rainfall from dynamically downscaled WRF simulations of current climate produce biased hydrological 10 simulations, in a case study for the State of Victoria, Australia (237,629 km 2 ). While the QQM bias correction can remove bias in daily rainfall distributions at each 10 km 2 grid point across Victoria, the GR4J rainfall-runoff model underestimates runoff when driven with QQM bias-corrected daily rainfall. We compare simulated runoff differences using bias-corrected and empirically scaled rainfall for several key water supply catchments across Victoria and discuss the implications for confidence in the magnitude of projected changes for mid-century. Our results highlight the imperative for methods that can 15 correct for temporal and spatial biases in dynamically downscaled daily rainfall if they are to be suitable for hydrological projection.

Climate change impact on extreme precipitation and peak flood magnitude and frequency: observations from CMIP6 and hydrological models

“…The uncertainty in future climate flood hazard assessment arises from the uncertainties in the different component elements of the modeling chain used to calculate streamflow from model-derived estimates of future rainfall. These include uncertainties in the model-estimated parameters of future rainfall and temperature, climate downscaling approaches for calibrating meteorological conditions at a particular basin, the structure and parameters of the hydrological model of streamflow, and the flood frequency model used to assess flood recurrence periods (Bastola et al 2011;Charles et al 2019;Meresa 2020;Lawrence 2020;Meresa et al 2021). Therefore, it is crucial to estimate and understand the complex and uncertain impact and challenging to identify the primary sources of uncertainty in future hazard estimation.…”

“…In the other hand, different climate projects (SRES, CORDEX, CMIP) have employed different ensembles of regional climate model (RCMs) and global climate models (RCMs) with different atmosphere-land modules to derive a wide range of possible future climate conditions. Therefore, the locally projected hydrological extremes are highly dependent on the spread (or range) of forcing conditions from the ensemble of the RCM/GCM estimates employed in the local impact study area (Woldemeskel et al 2014;Hattermann et al 2018;Charles et al 2019;Meresa 2020;Lawrence 2020). The information from the spreads of multiple GCMs is considered as a band of impact uncertainty and quantified using the variance of the climate models Lawrence 2020).…”

“…The information from the spreads of multiple GCMs is considered as a band of impact uncertainty and quantified using the variance of the climate models Lawrence 2020). Bias correction methods of GCMs output are also essential and significantly impact the projected hydrological extremes magnitude and direction (Pierce et al 2015;Hattermann et al 2018;Charles et al 2019). Similarly, the biases correction of GCMs output was applied on a grid commensurate with daily climate model output.…”