Search Results
Abstract
The Weather Research and Forecasting (WRF) model version 3.0.1 is used in both short-range (days) and long-range (years) simulations to explore the California wintertime model wet bias. California is divided into four regions (the coast, central valley, mountains, and Southern California) for validation. Three sets of gridded surface observations are used to evaluate the impact of measurement uncertainty on the model wet bias. Short-range simulations are driven by the North American Regional Reanalysis (NARR) data and designed to test the sensitivity of model physics and grid resolution to the wet bias using eight winter storms chosen from four major types of large-scale conditions: the Pineapple Express, El Niño, La Niña, and synoptic cyclones. Control simulations are conducted with 12-km grid spacing (low resolution) but additional experiments are performed at 2-km (high) resolution to assess the robustness of microphysics and cumulus parameterizations to resolution changes. Additionally, long-range simulations driven by both NARR and general circulation model (GCM) data are performed at low resolution to gauge the impact of the GCM forcing on the model wet bias.
These short- and long-range simulations show that low-resolution runs tend to underpredict precipitation in the coast region and overpredict it elsewhere in California. The sensitivity test of WRF physics in short-range simulations indicates that model precipitation depends most strongly on the microphysics scheme, though convective parameterization is also important, particularly near the coast. In contrast, high-resolution (2 km) simulation increases model precipitation in all regions. As a result, it improves the forecast bias in the coast region while it downgrades the model performance in the other regions. It is also found that the choice of validation dataset has a significant impact on the model wet bias of both short- and long-range simulations. However, this impact in long-range simulations appears to be a secondary contribution as compared to its counterpart from the GCM forcing.
Abstract
The Weather Research and Forecasting (WRF) model version 3.0.1 is used in both short-range (days) and long-range (years) simulations to explore the California wintertime model wet bias. California is divided into four regions (the coast, central valley, mountains, and Southern California) for validation. Three sets of gridded surface observations are used to evaluate the impact of measurement uncertainty on the model wet bias. Short-range simulations are driven by the North American Regional Reanalysis (NARR) data and designed to test the sensitivity of model physics and grid resolution to the wet bias using eight winter storms chosen from four major types of large-scale conditions: the Pineapple Express, El Niño, La Niña, and synoptic cyclones. Control simulations are conducted with 12-km grid spacing (low resolution) but additional experiments are performed at 2-km (high) resolution to assess the robustness of microphysics and cumulus parameterizations to resolution changes. Additionally, long-range simulations driven by both NARR and general circulation model (GCM) data are performed at low resolution to gauge the impact of the GCM forcing on the model wet bias.
These short- and long-range simulations show that low-resolution runs tend to underpredict precipitation in the coast region and overpredict it elsewhere in California. The sensitivity test of WRF physics in short-range simulations indicates that model precipitation depends most strongly on the microphysics scheme, though convective parameterization is also important, particularly near the coast. In contrast, high-resolution (2 km) simulation increases model precipitation in all regions. As a result, it improves the forecast bias in the coast region while it downgrades the model performance in the other regions. It is also found that the choice of validation dataset has a significant impact on the model wet bias of both short- and long-range simulations. However, this impact in long-range simulations appears to be a secondary contribution as compared to its counterpart from the GCM forcing.