Additional Model Developments

Radiative Kernels

Radiative kernels are a tool used to decompose changes in radiative fluxes into components associated with temperature, water vapor, and surface albedo. They can also be used indirectly to diagnose the contribution of changes in clouds to radiative fluxes. Previous sets of kernels for the TOA exist, but few surface kernel datasets do, and no model-based kernels had been validated prior to this study. Furthermore, a complete technical description of how to generate and use kernels had never before been published. This set of radiative kernels was generated using offline radiative transfer calculations with CAM5-PORT and the CESM Large Ensemble base state. The kernels, radiative forcing, and demo data were released on NCAR's ESG, and software to implement the kernels are provided on Github. The data and also software to implement them are available for the community.

Pendergrass, A. G., Conley, A., and Vitt, F. M., 2018: Surface and top-of-atmosphere radiative feedback kernels for CESM-CAM5, Earth System Science Data, 10, 317-324, Doi: 10.5194/essd-10-317-2018

Sea Ice Model Developments

In 2016, a group of primary developers and users of the Los Alamos Sea Ice Model (CICE) founded the interagency, international CICE Consortium, whose mission is to foster collaboration on sea ice model developments for earth system research and operational applications. The Consortium aims to accelerate sea ice model development by providing a pathway and support for community contributions to its sea ice model codes. NCAR personnel (D. Bailey and A. DuVivier) are responsible for the community aspects of the Consortium which includes documentation coordination of community announcements and forums. During the past year, Bailey and DuVivier have contributed to new sea ice code developments including the upcoming release of CICE version 6.0 (targeted for November 2018) and introduction of Icepack version 1.0, a modular software package of sea ice column physics. In addition, Bailey and DuVivier have assisted in the development and documentation of the quality control and testing procedures for incorporating community contributions into Icepack and CICE.

Land Ice Model Developments

The 2018 release of CESM2.0 was accompanied by the release of version 2.1 of the Community Ice Sheet Model (CISM; Lipscomb et al. 2018). CISM2.1 is an open-source model supported for standalone ice sheet simulations and as a coupled component of CESM. Innovations since the previous release include:

  • An efficient depth-integrated velocity solver, with accuracy comparable to the previous 3D solver at a small fraction of the cost
  • Improved parameterizations of physical processes such as basal sliding and iceberg calving
  • Support for interactively coupled simulations of the Greenland ice sheet in CESM

Recent development has focused on marine ice sheets, including Antarctica. Novel spin-up techniques and new sub-grid treatments of processes at the grounding line (the boundary between grounded and floating ice) will support projections of decade-to-century-scale Antarctic changes in a warming climate. Using CISM and CESM, members of the Land Ice Working Group (LIWG) are participating in all aspects of the Ice Sheet Model Intercomparison Project for CMIP6 (ISMIP6):

  • Model initialization studies for Greenland and Antarctica
  • Analysis of ice-sheet-relevant output from CMIP6 climate simulations
  • Standalone ice sheet experiments using CMIP-derived forcing for sea level projections
  • Coupled simulations to study climate – ice sheet feedbacks

Lipscomb, W. H., S. F. Price, M. J. Hoffman, G. R. Leguy, A. R. Bennett, S. L. Bradley, K. J. Evans, J. G. Fyke, J. H. Kennedy, M. Perego, D. M. Ranken, W. J. Sacks, A. G. Salinger, L. Vargo, and P. H. Worley (2018), Description and evaluation of the Community Ice Sheet Model (CISM) v. 2.1, Geosci. Model Dev. Discuss., In Review

Community Simple Models

The Community Earth System Model (CESM) is a comprehensive, but complex framework for undertaking earth system research. Understanding the nature of the myriad interactions and the processes involved is a significant challenge. Reducing the system down to a less complex framework can simplify the interactions relevant to a research problem and help identify key processes that can be used to understand the larger earth systems. These simpler model, idealized configurations and testing frameworks are crucial for model development, evaluation and general advancement of the science.

Although many of these capabilities existed previously, they were largely disparate, unsupported and not subject to systematic testing. Furthermore, they were generally unavailable to the broad CESM community. However, a number of these capabilities have been collated, developed and made available to the community with detailed instructions for both running the default configurations and making additional modifications (Polvani et al., 2017).  Available configurations include the aqua-planet, the dry dynamical core with the Held-Suarez configuration, the single column model and two dynamical core test cases (moist baroclinic wave with Kessler physics and the toy terminator chemistry test case).  These are documented on the simpler models' webpage
and in the CAM6 users guide

It is intended that these simpler methodologies will form a much larger role in the CESM development and evaluation process in the future, also enabling greater accessibility for University participation.

Figure: Examples of available and supported simpler models
Figure: Examples of available and supported simpler models. Left: Baroclinic wave tests showing surface pressure from CAM dynamical core comparisons. Center: Single-column (SCAM) version of CAM showing cloud ice and liquid representations for the ARM-SGP during March 2000 and Right: Hemispherically averaged precipitation showing the range of responses from different models in a simplified aqua-planet framework.
Polvani, L. M., A. C. Clement, B. Medeiros, J. J. Benedict, and I. R. Simpson (2017), When less is more: Opening the door to simpler climate models, Eos, 98, Doi: 10.1029/2017EO079417 Published on 25 September 2017.

CESM Variable Resolution Simulations of Continental US Climate

The spectral element (SE) variable-resolution (VR) mesh dynamical core has been tested in developmental versions of the Community Earth System Model version 2 (CESM2). The performance of the SE dynamical core has been assessed in simplified baroclinic wave and aqua-planet examples, as well as for full physics configurations, to evaluate variable-resolution performance against uniform high and uniform low-resolution simulations. Different physical parameterization suites, both legacy and developmental, have been evaluated to gauge their sensitivity to resolution. Dry dynamical core, variable-resolution cases compare well to high-resolution tests, which provides confidence in the use of local refinement in capturing large-scale modes of variability in an equivalent manner to global high resolution.

It is apparent that more recent versions of the atmospheric physics, including cloud schemes for CESM2, are less sensitive to changes in horizontal resolution, as demonstrated in aqua-planet configurations. Most of the sensitivity is due to sensitivity to time step and interactions between deep convection and large-scale condensation, which is expected from the closure methods. The resulting full physics SE-VR model produces a similar climate to the global low- resolution mesh and similar high-frequency statistics in the high-resolution region. The SE-VR simulations are able to reproduce uniform high-resolution results, making them an effective tool for regional climate simulations at lower computational cost. Some biases are reduced (e.g., orographic precipitation in Western United States), but biases do not necessarily abate at high resolution (e.g., summertime surface temperatures). Variable-resolution grids are a viable alternative to traditional nesting for regional climate studies and will be able in an upcoming release of CESM2.

Figure: Variable resolution mesh and JJA surface temperature biases
Figure: Left: Variable resolution mesh used for simulations refining from a global 1 deg grid (ne30) to a 0.25 deg (ne120) resolution. Right: JJA Surface temperature biases from the global uniform and variable mesh simulations.
Gettelman, A., P. Callaghan, V. E. Larson, C. M. Zarzycki, J. T. Bacmeister, P. H. Lauritzen, P. A. Bogenschutz, and R. B. Neale, 2018: Regional Climate Simulations with the Community Earth System Model. J. Adv. Model. Earth Syst., Doi: 10.1002/2017MS001227 Published on 25 September 2017.