Regional-scale climate simulations: Improvement in near-surface field projections with spectral nudging
EVS researchers have established a new optimal approach for downscaling physically based global climate model projections to the regional scale. The new approach is significant because it preserves both the prescribed large-scale dynamics of the global model and the increased variability of the higher-resolution physics in a regional-scale model.
Global-scale climate models have coarse spatial resolution and cannot resolve many local and regional-scale features, such as increased precipitation due to the lake effect in the Chicago region. Specially built models that operate at a higher spatial resolution and can resolve these types of features are referred to as regional-scale climate models. These models are often employed to telescope from the global model resolutions to finer local and regional details, a process referred to as downscaling. Downscaling performed with a physically based regional-scale model is known as dynamic downscaling, a reference to the fact that the downscaling model has a full representation of the physics of climate dynamics at regional scales. (Another process, statistical downscaling, uses historical statistics of observed climate to achieve the downscaling.)
Because regional-scale climate models only cover a portion of Earth, the edges of the model (or the boundaries) need to be prescribed on the basis of observations or calculated from a global-scale model. The choices made in prescribing the conditions at the boundaries introduce significant uncertainty in generating projections. A regional climate model ideally should simulate the physics of the climate at small scales without deviating significantly from the large-scale climate features prescribed at the boundaries (the boundary conditions).
The EVS investigators used the regional-scale Nested Regional Climate Model (NRCM) with a grid spacing of 12 km over the United States (excluding Hawaii) to dynamically downscale 2.5–degree data (National Centers for Environmental Prediction–U.S. Department of Energy Reanalysis-2 observations). To preserve the large-scale climate features of the reanalysis data, we employed a process known as nudging — bringing the large-scale fields from the boundary conditions and those calculated by the regional climate model as close to the prescribed conditions as feasible. This keeps the regional-scale modeling from drifting away from the imposed large-scale climate at the boundaries, ensuring that the model is in fact downscaling the features from the larger domain to the finer domain without creating a completely self-generated large-scale forcing that is unrelated to observations.
We tested a number of different methods for applying nudging and developed an optimal approach that improved the performance of the NRCM in predicting near-surface fields by more than 30.5% compared to a case with no nudging. A manuscript has been published in the Journal of Applied Meteorology and Climatology: “Assessment of dynamical downscaling in near-surface fields with different spectral nudging approaches using the Nested Regional Climate Model (NRCM),” by J. Wang and V. Kotamarthi.
This work was supported by the Strategic Environmental Research and Development Program. The computational resources used were the DOE Argonne Leadership Computing Facility cluster and the National Energy Research Scientific Computing Center. These model calculations are among the first achieving climate downscaling at a spatial resolution of 12 km for North America. The results will be part of an international effort to generate regional-scale model products for all regions of the world, known as CORDEX (Coordinated Regional Climate Downscaling Experiment). We plan diagnostic simulations for a total of 30 years (extending 20 years into the future) for Representative Concentration Pathways scenarios and boundary conditions commissioned by the Intergovernmental Panel on Climate Change.