1. What is CM1?
CM1 is a three-dimensional non-hydrostatic numerical model that has been developed by George Bryan at The Pennsylvania State University (PSU) (circa 2000-2002) and at NCAR (2003-present). CM1 is designed primarily for idealized research, particularly for deep precipitating convection (i.e., thunderstorms).
2. Who can use CM1?
CM1 is freely available to anybody that wants to use it.
3. Is there support for CM1?
Support for CM1 is very limited. George Bryan maintains the code, and tries to answer as many questions as possible, but it sometimes takes him a long time to respond to some questions. Stronger support is developing gradually as more users become experts with certain aspects of the modeling system. Nevertheless, CM1 requires a user with a strong background in coding (especially of fortran-90), and good knowledge of UNIX. Experience with numerical modeling is very beneficial.
4. How many people use CM1?
The list of users that I know about is small (~20), but growing. Currently, CM1 is used by researchers at NCAR, The Pennsylvania State University, The University of Illinois at Urbana-Champaign, Central Michigan University, North Carolina State University, Colorado State University, CNR-ISAC (Italy), University of Munich (Germany), and elsewhere.
5. Does CM1 conserve mass?
No, CM1 does not precisely conserve mass. However, because the pressure equation in CM1 retains all terms, it's mass errors are several orders of magnitude smaller than those from other cloud models that integrate pressure equations (e.g., MM5, ARPS, RAMS). See Bryan and Fritsch (2002, MWR, pg 2917) for more information.
6. Does CM1 conserve total energy?
No, CM1 does not precisely conserve total energy. However, several terms in CM1's equations have been retained as it relates to conservation of total internal energy in moist environments; few numerical models retain these terms, which are important in conditions with large liquid water content (e.g., thunderstorms). CM1 also includes dissipative heating, which is neglected in most nonhydrostatic models (including, but not limited to, MM5, ARPS, WRF). Thus, CM1 conserves total energy much better than other modern cloud models. See Bryan and Fritsch (2002, MWR, pg 2917) for more information.
7. What makes CM1 different from other models
One main difference was explained in the two previous questions: CM1 conserves mass and energy better than other modern cloud models (such as ARPS and RAMS for mass and energy, and WRF for energy). Also, CM1 was designed specifically to do very-large domain simulations using high resolution; specifically, it has comparatively little memory overhead, which allows the code to be applied to very large problems (i.e., domains of order 109 grid points). Also, CM1 is rare (if not unique) in its ability to use different equation sets, for different applications; for example, the model can be run using the compressible equations (with three different solvers, depending on application and desired accuracy), but it can also be used with the anelastic equations, and even the incompressible equations. This capability allows CM1 to be used very efficiently for a broad range of problems that span many scales.
8. Can CM1 be used as a large eddy simulation (LES) model?
Yes. In fact, CM1 was originally designed to do LES of deep, precipitating convection (i.e., thunderstorms). The code is currently being configured and evaluated for LES of planetary boundary layers, including studies of dry convective boundary layers (by S.-L. Kang of NCAR and K. Davis of PSU) and of marine cumulus (by D. Kirshbaum of U. Reading).
9. Is CM1 being actively developed?
Yes, and no. YES, in the sense that bug fixes and model performance are constant being addressed. Also, if some capability is needed -- like the addition of a new microphysics scheme for a specific research project -- then the model will be modified appropriately. Work is also underway to accomodate different output file formats (e.g., netcdf and hdf), and an axisymmetric version of the model has recently been made available. However, the answer to this questions is also, NO, in the sense that CM1 is viewed (by G. Bryan) as a "frozen" code, at least in terms of the model's main solver. No major upgrades to the model are planned. This means that the model will not be adapted for real-data cases, it will not be extended to non-idealized domains by adding map projections or coordinate transforms, and no changes are planned for the model's equations, advection schemes, or solver.
10. How fast is CM1 compared to other models?
CM1 has not been rigorously benchmarked against other models. It has only been compared head-to-head on a few occasions, usually on single-processor machines. Furthermore, these tests have always been for situations that CM1 was primarily designed for (e.g., idealized cloud modeling). From these limited tests, we have concluded that CM1 is roughly 2 times faster and uses roughly half as much memory (RAM) as the Advanced Research WRF. Also, CM1 is roughly 1.5-2 times faster than ARPS (when using the same time step), and uses ~75% of the memory (RAM) required by ARPS.
Send comments and/or questions about this page to:
George H. Bryan
Mesoscale and Microscale Meteorology (MMM) Division
National Center for Atmospheric Research
3450 Mitchell Lane
Boulder, CO 80301, USA
email: gbryan at ucar dot edu
Last updated: 12 August 2008