Earth System Bridge: Spanning Scientific Communities with Interoperable Modeling Frameworks
Scientific modeling helps us understand our environment. Diverse scientific communities team up to predict the environmental hazards of weather/climate, ecology, and deep Earth processes. However, technological and logistical gaps impede streamlined communication between academics whose research advances these predictions, and mission-driven federal agencies that use their tools for the benefit of society. There is also a need to integrate these modeling tools within and across academic communities. This project is developing innovative approaches that allow different Earth System modeling frameworks to work together or “interoperate,” thereby helping bridge these gaps, and enabling geoscience communities to collaborate and advance earth system science.
The Earth System Bridge (ESB) team includes leaders from several major modeling frameworks and projects - both federal and academic - including: CSDMS (Community Surface Dynamics Modeling System), ESMF (Earth System Modeling Framework), CIG/Pyre (Computational Infrastructure for Geodynamics), CUAHSI (Consortium of Universities for the Advancement of Hydrologic Science) and UCAR (University Consortium for Atmospheric Research). Leaders from the OMS (Object Modeling System) and OpenMI (Open Modeling Interface) modeling frameworks are also contributing to the project.
Designing the Bridge: Framework Definition Language (ES-FDL)
The Earth System Framework Description Language (ES-FDL, pronounced ES-fiddle) is a metadata schema that is being developed to describe modeling frameworks, much like a blueprint. It will serve as a theoretical basis for understanding and connecting frameworks such as CSDMS, ESMF, the Model Coupling Toolkit (MCT), OMS, OpenMI and the OASIS coupler. The ES-FDL activity builds on and extend a series of international workshops that brought together developers of Earth system model coupling technologies for collaborative comparison and analysis (Dunlap et al., 2013). These workshops, coordinated by the International Working Committee on Coupling Technologies (IWCCT) resulted in improved understanding of individual coupling technologies and published reviews of the state of the art (Valcke et al., 2013). An anticipated next step is cross-framework analysis of features and development of connections and collaborations. The FDL will accelerate this goal by enabling more precise classification of frameworks, and by informing the nature of the bridges that can be built between them.
Building the Bridge, Span by Span: Sharing Resources Between Frameworks
One of the technologies being used and further developed for this project is the Basic Model Interface (BMI), developed for the CSDMS modeling framework. This is a framework-agnostic model interface that is easy to implement in different computer languages but provides the caller (any modeling framework) with fine-grained control via initialize, update and finalize methods, as well as a standardized description of its input and output variables, its grid(s), time-stepping scheme, and other information that is required by the framework to enable plug-and-play model coupling. The fact that it is framework agnostic means that developers don’t need to adapt their model for any particular framework, and also means that adapters can be written that allow a BMI-augmented model to be used in multiple modeling frameworks. The CSDMS framework already supports BMI-compliant models and the Bridge team is working on adapters that connect BMI-compliant models to the ESMF/NUOPC , Pyre/St. Germain, OMS and OpenMI frameworks. A crosswalk between the CF (Climate and Forecasting) Standard Names and the CSDMS Standard Names is also being developed in support of this effort.
Crossing the Bridge: Linking Federal and Academic Models
One of the goals of the Bridge project is to use tools like ES-FDL, BMI and standardized metadata to link together multiple frameworks in order to create new modeling and predictive capabilities. The Bridge Team is partnering with the National Weather Service, CSDMS, and the WRF-Hydro project to link new models to the NOAA operational Environmental Modeling System (NEMS). There are two main parts to this effort:
- A regional atmospheric model called Nonhydrostatic Multiscale Model on the B grid (NMM-B) will be coupled to the regional Princeton Ocean Model (POM) for hurricane prediction using ESMF. This will involve testing the use of BMI within ESMF.
- The new NMMB-POM system in NEMS will be coupled to a land model that uses the hydrological components from WRF-Hydro. This effort links NOAA operational models, the NOAA Water Center, and the hydrological research community.
To demonstrate improved interoperability between modeling frameworks, the team is also linking a deep Earth process model called SNAC (a geodynamics / tectonics model, used by the CIG (Computational Infrastructure for Geodynamics) community with a surface process model called CHILD (a landscape evolution model), available within the CSDMS framework. SNAC is being augmented with a BMI interface and new CSDMS Standard Names are being developed for variables used in geodynamics models, with help from domain experts.