CAM Use Case
Curator Archival and Query Tool Client Use Cases
Archive meta-data from a CAM model run.
Jane modeler interfaces with the Curator front-end web application to store and archive metadata about an experiment studying the impact of large-scale wind power on global climate.
Actors
o Jane Modeler
o Curator Query and Archival Tool Web Client
Summary
Jane runs an NCAR CAM model experiment to study the effects of large-scale wind power on global climate. She wishes to document and archive her study so that it may be shared with the scientific community. In addition, she wants to record her scientific experiment in detail so that it may be reproduced and verified at a later time.
Trigger
Jane has completed a CAM model run that she would like to document and share with the climate science community.
Preconditions
Jane has a CAM model experiment that she would like to document. She knows how to run the CAM model. She has an account on a supercomputer that can execute CAM. She has a user name and password on the Curator system.
Goals and Post-conditions
Jane successfully stores metadata about her model run into Curator so that she or other Curator system users may retrieve those metadata at a later time.
Main Success Scenario
Jane is implementing a simulation of large-scale wind power and its impact on global climate. She chooses to use the NCAR CAM model (http://www.ccsm.ucar.edu/models/atm-cam/) for her simulation.
o Jane downloads the CAM model from http://www.ccsm.ucar.edu/models/atm-cam/download/index.html and unpacks it in her account on an IBM SP platform at NCAR.
o Jane sets up the UNIX environment variables that are required to configure and run CAM.
o Jane runs the CAM “configure” utility to configure various compile-time options. [There are many configuration options.] The CAM documentation states:
configure produces the configuration build files Filepath, misc.h, params.h, preproc.h, and Makefile. Each of these files specify compile-time parameters and settings needed to build the model. In addition, a configuration cache file (config_cache.xml by default) is written which may be used in a subsequent invocation of configure to exactly reproduce the configuration files.
o Jane compiles the model by running gmake.
o Jane runs the build-namelist utility “that builds namelists which specify run-time details for CAM.”
o Jane runs her CAM model on an NCAR IBM SP platform.
The model yields a netCDF file that can be subsequently interpreted in various software tools such as GrADS. Finding the results of her experiment noteworthy, Jane wants to document and share the results with her peers.
o Jane points her browser to http://www.esmf-curator.org.
o She logs into the system with her username and password.
o She navigates through the web site to the archival tool. This tool provides various options to collect metadata from the several components of the CAM model used for the wind power study as well as the output data.
o Jane points Curator to config_cache.xml file. From this file, Curator is able to harvest meta-data about Jane’s model run. In particular, Curator is able to obtain information about:
- Model resolution.
- Time period covered by model.
- Dynamical core (Eulerian Spectral, Semi-Lagrangian Spectral, or Finite-Volume).
- Type of parallelism to employ (shared-memory and/or distributed memory).
- Number of constituents.
- Number of vertical levels.
- [CAM experts: What else?]
o Jane points Curator to the namelist input files produced by the build-namelist utility. From these file, Curator is able to harvest meta-data about Jane’s model run. In particular, Curator obtains pointers to the following input datasets:
- Vegetation type.
- Soil texture.
- Urban area.
- Etc.
o Jane stores the netCDF output files on the NCAR mass storage system. Jane points Curator to the location of those files.
o Curator must now obtain manually-input information from Jane:
- Funding source.
- Institution where the model was executed.
- Name of modeler and collaborators.
- Reference to peer reviewed journal where the study was (eventually) published.
Implications
o Who is responsible for maintaining the model metadata? Is it guaranteed to be there forever?
o Each model run ultimately needs to be associated with a unique identifier. Where is it going to come from?
Obstacles and Caveats
o Jane may see no point in storing her metadata in Curator if this task requires a lot of work, and “manual” intervention.
o What if not all relevant information from the CAM model experiment can be obtained from config_cache.xml, the name list files, and the output files?
o How will Curator obtain the UNIX environment variables needed to run the CAM model?
o What are (if any) the security implications and considerations?
Link to this Page