Monday, 23 January 2017: 4:15 PM
Conference Center: Chelan 5 (Washington State Convention Center )
Python plays an increasingly important role during the model development process at NOAA-GFDL. Python-based analyses and workflow infrastructure enhancements enable rapid analysis and evaluation of model simulations and are central to the support and development of GFDL’s next-generation coupled climate model, CM4. The open-source nature of Python, coupled with more advanced computing capability and robust plotting and mapping packages, have made it an attractive tool for analysis. Here we present a broad overview of ways that Python has become part of the model development process at GFDL. GFDL’s Diagnostics and Evaluation Team assembled a custom collection of Python Common Gateway Interface (CGI) scripts that enable real-time, web-based image generation and analysis. Python serves as a link between GFDL model data and community analysis efforts, such as the PCMDI Metrics Package (PMP) and the Earth System Model eValuation Tool (ESMValTool). GFDL’s Ocean Model Working Group developed a collection of Python-based analysis and benchmarking scripts that are part of version 6 of the Modular Ocean Model (MOM6). Jupyter notebooks are used increasingly among GFDL scientists to conduct and share analyses. The notebooks, along with JupyterHub, are being tested as a primary method of documenting CMIP6 models through the Earth System Documentation (ES-DOC) effort. GFDL’s Modeling Systems group developed a Python API that allows scientists to interface with GFDL’s “Curator” database, which stores metadata for most of GFDL’s simulations, supports model development efforts, facilitates quality control efforts of publically-available model results, and provides access to additional analysis capabilities. The use of Python has sparked interest among a broad range of scientists at GFDL and is evident through ad-hoc training sessions and support groups that have grown organically within the lab. The flexibility and interoperability of Python allows new analysis efforts to integrate well with existing analysis capabilities at GFDL and will help to expand the diversity and number of ways GFDL’s climate models are evaluated.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner