How the increasing resolution in dynamical downscaling using regional climate models (RCMs) benefits the understanding and impact assessment of climate change has long been a debatable topic. This study investigates such an “added value” problem of high resolution RCM by a set of long-term climate downscaling simulations using Weather Research and Forecasting Model (WRF) over South Korea at 20 (WRF20) and 5 km (WRF05). The downscaling ensemble consists of the perfect-boundary condition (PBC), three Global Climate Models (GCMs) with varying climate sensitivities, and the pseudo global warming (PGW) condition. Furthermore, the generated meteorological information is post-processed by different bias correction (BC) methods and fed into a designated hydrological model (HM) for streamflow p...[ Read more ]

How the increasing resolution in dynamical downscaling using regional climate models (RCMs) benefits the understanding and impact assessment of climate change has long been a debatable topic. This study investigates such an “added value” problem of high resolution RCM by a set of long-term climate downscaling simulations using Weather Research and Forecasting Model (WRF) over South Korea at 20 (WRF20) and 5 km (WRF05). The downscaling ensemble consists of the perfect-boundary condition (PBC), three Global Climate Models (GCMs) with varying climate sensitivities, and the pseudo global warming (PGW) condition. Furthermore, the generated meteorological information is post-processed by different bias correction (BC) methods and fed into a designated hydrological model (HM) for streamflow projections. Altogether, the comparative assessment of the resolution effect in the model chain of climate – hydrological simulations demonstrates the benefits of high resolution RCM from the perspectives of apparent, process-based, and application-based added values in the context of hydrological climate-change implications.

The current-climate simulations are first validated against the observation and consistently show an added value of WRF05 for reproducing extreme events and reflecting complex geographical features. Moreover, there is an improved physical realism in the higher resolution related to the partitioning of convective and large-scale precipitation (CP/LP) and the sensitivities of precipitation to the temperature. WRF05 captures a higher fraction of LP and a positive LP-temperature scaling in the extremes which is seen in the observation but missed in the coarse resolution simulations.

The distinct process-based added value of increasing resolution is important for evaluating future projections with no observation. The improvement of WRF05 in the extreme precipitation-temperature relation implies a higher reliability in the projected precipitation. The downscaled precipitation change from both GCM ensemble and PGW concurs that there will be intensified extreme precipitation mainly contributed by CP increase that shows higher temperature sensitivity. On the other hand, larger uncertainty and inter-model spread are found in the change of mean and moderate precipitation and under lower level warming.

Apart from analyses of the climate simulations, the hydrological simulations enable an added value assessment from the perspective of end users. The result shows that although BC is still necessary as a post-processing tool, the improved representation in WRF05, especially for spatial variability, can reduce the uncertainty brought by different BC methods in the projected streamflow change, particularly for the high flows.

In all, this study produces an ensemble of the finest-resolution long-term climate simulations for South Korea so far. The major findings can advance our understanding of the potentials and limitations of downscaled climate information in the context of hydrological implications and motivates the use of high resolution RCMs by climate modelers and end users.