Model data and analyses for the research community

By Brian Bevirt
05/11/2017 - 10:00am

This article is part of a CISL News series describing the many ways CISL improves modeling beyond providing supercomputing systems and facilities. These articles briefly describe CISL modeling projects and how they benefit the research community.

Two important goals of the North America Co-ordinated Regional Climate Downscaling Experiment (NA-CORDEX) project are to quantify the scientific benefits gained by increasing model resolution, and to explore uncertainty across the outputs of these models. The co-Directors of this program are Linda Mearns (CISL IMAGe) and William Gutowski (Iowa State University).

Model output comparison
Climate projections from two regional climate models at 25-km resolution driven by boundary conditions from two different global climate models.

A highlight among the many experiments in NA-CORDEX is a comparison in which two different regional climate models (RCMs) are driven by boundary conditions generated by three global climate models (GCMs) having different equilibrium climate sensitivity. Each GCM drives each RCM at 50-km resolution, then again at 25-km resolution to estimate climate change over North America for 150 years, from 1950-2100. Researchers then compare these 12 RCM outputs to quantify the scientific benefits of doubling the resolution. Further, the NA-CORDEX design allows researchers to analyze and quantify the uncertainty resulting from different modeling choices. The original outputs and analyses are being archived and published on the Earth System Grid Federation (ESGF) and used by a wide range of climate researchers and model developers. Scientists are currently preparing research papers that establish the differential credibility of the simulations.

The four plots at left show climate projections from two RCMs at 25-km resolution. The Regional Climate Model Version 4 (RegCM4) output is on the left and the Weather Research and Forecasting model (WRF) output is on the right. The upper pair of plots shows the RCMs’ projections when they are driven by boundary conditions from the Geophysical Fluid Dynamics Laboratory (GFDL) GCM, and the lower pair shows the RCMs’ projections when they are driven by the Max Planck Institute (MPI) GCM. Scientists study the differences and similarities in these projections to understand the reasons for the differences and what those differences mean to their research. Then they judge the credibility of each result. This work improves our scientific understanding of how models represent complex climate processes, and it helps developers refine the models for more accurate projections.

Two grants are funding valuable extensions of this work and provide further benefits to the research community. A U.S. Department of Energy (DOE) grant supports research in using dynamical downscaling methods to double the 25-km RCM resolution to 12-km grid spacing for local climate projections. A U.S. Department of Defense (DOD) grant from their Environmental Security Technology Certification Program (ESTCP) supports bias-correction of the output data, creation of derived data products for climate impacts users, and archiving all of these data products for use by the climate impacts community.

The FACETS project also investigates the effects of land-cover changes on the atmosphere.

The DOE-funded project is named the Framework for Assessing Climate’s Energy-Water-Land Nexus using Targeted Simulations (FACETS). Project Scientist and lead modeler Melissa Bukovsky (IMAGe) uses dynamical downscaling models to resolve higher-resolution climate conditions using boundary conditions from coarser GCMs. Three of the same GCMs downscaled in NA-CORDEX to 25km and 50km resolution will also be downscaled to 12-km resolution, thus simulating conditions closer to the local scale. Researchers are using these diverse yet carefully designed climate projections to compare the credibility of each model’s outputs using process-level analysis. This type of analysis helps investigators determine if the models are getting the right answers for the right reasons. These intermodel comparisons are particularly valuable in helping modelers refine their codes. These comparisons also advance collaborations with Brian O’Neill’s Integrated Assessment Modeling group in NCAR’s Climate and Global Dynamics Laboratory that study the effects of land use change. The intermodel comparisons are valuable for improving our understanding of the uncertainty associated with different aspects of modeling.

The DOD-funded project supports climate-impacts users, who are typically not climate modelers, with archives of data products tailored to their needs. For these users, Associate Scientist Seth McGinnis (IMAGe) produces three types of data products from NA-CORDEX simulations: bias corrections (using statistical methods that make slight revisions to data values to improve their accuracy), daily, monthly, and seasonal averages of data values, and data interpolated onto the types of grids appropriate for the end users. All of these products are then archived and published on NCAR’s Earth System Grid. Collaborators in CISL’s Software Engineering and Gateway Applications group have added capabilities to the ESG portal that facilitate searches of this archive to improve the accessibility of all its contents. These derived data products allow non-modelers to use simulation outputs more effectively so they can maintain their focus on their research questions.

This work is one part of the World Climate Research Programme (WCRP) CORDEX project, and the simulation efforts for North America alone involve too many collaborators to name here. In addition to the people noted above, the DOE FACETS project involves collaborators at four research universities and two national labs, and the DOD ESTCP project involves two other national labs. For more information, see the NA-CORDEX project website which provides clear overview information in an easy-to-navigate format.