Models and Software Frameworks
Because
the interactions among the Earth system’s
biological, chemical, and physical components are complex and nonlinear,
simulation using numerical models is currently the best tool
for understanding our environment.
Simulation allows scientists to perform quantitative studies, test
hypotheses, make predictions, and assess environmental mitigation
strategies.
Software frameworks help to organize complex codes, provide standard interfaces that allow model components to be combined and reused, and offer standard ways to deal with common functions such as error handling and data exchange. Frameworks reduce development and maintenance cost, allowing scientists to concentrate on their science rather than on the mechanics of software engineering.
In cooperation with many collaborators from the research community, CISL is developing a new generation of numerical models and software frameworks that will help extend understanding of the Earth and the evolution of weather and climate. These models and frameworks address complex geophysical problems, offer new functionality, take advantage of high-end computer architectures, and provide computational environments where scientists can more easily explore ideas, leverage methodologies, and share tools.
Models
HOMME: CISL is developing an accurate, efficient, and scalable general-circulation model called the High-Order Multiscale Modeling Environment (HOMME). HOMME employs advanced algorithms and computing techniques that allow it to use tens of thousands of processors effectively. The dynamical core of the model, based on the spectral-element numerical method, provides the computational foundation for solving the fluid dynamics equations necessary to study the atmosphere. CISL is now coupling the core to complex physical packages such as Emanuel and NCAR's Community Atmosphere Model, adding the ability to model moisture and its profound effects on the atmosphere. CISL is also adding parameterization techniques that improve the simulation of cloud processes.
GASpAR: Accurate and efficient simulation of turbulent flows is a challenge in many geoscience fields — for instance, meteorology, oceanography, climatology, ecology, solar-terrestrial interactions, and solar fusion. The Geophysical Astrophysical Spectral-Element Adaptive Refinement (GASpAR) code combines the efficiency of finite-element numerical methods with the accuracy of spectral methods. GASpAR uses a dynamic, adaptive-refinement technique for direct numerical simulation and large-eddy simulation of turbulent flows. The code is designed to be flexible enough for a range of geophysics and astrophysics applications where turbulence or other complex multiscale problems arise.
BOB: A dynamical core is the part of a weather or climate model that solves the primitive equations of fluid motion, which are the most computationally intensive calculations of a general-circulation model. BOB, which stands for "Built On Beowulf," is an efficient dynamical core based on the spectral-element numerical method. BOB was designed to run on inexpensive commodity clusters (called "Beowulf" systems) that are often used as an alternative to high-performance computers.
Software frameworks
ESMF: CISL is collaborating with many other agencies and institutions in a national effort to build the Earth System Modeling Framework (ESMF). ESMF is a software package that provides tools for turning model codes into components with standard interfaces and drivers. It increases scientific productivity by making components easier to build, combine, and exchange, and by enabling modelers to take full advantage of high-end computers. It also promotes new scientific opportunities and services through community building and increased interoperability of codes.
DART: CISL has developed a data assimilation facility that makes it possible for scientists to prototype the impact of data assimilation on different geophysical models and observation sets. The facility, called the Data Assimilation Research Testbed (DART), is an open-source, publicly available software environment that combines assimilation algorithms, models, and observation sets to provide a flexible, extensible framework for data assimilation research.
