Input

The [input] section of a configuration file contains options used to specify the location of the “main” simulation and some settings for how these data will be read in:

[input]
## options related to reading in the results to be analyzed

# directory containing model results
baseDirectory = /dir/for/model/output

# Note: an absolute path can be supplied for any of these subdirectories.
# A relative path is assumed to be relative to baseDirectory.
# By default, results are assumed to be directly in baseDirectory,
# i.e. <baseDirecory>/./

# subdirectory containing restart files
runSubdirectory = .
# subdirectory for ocean history files
oceanHistorySubdirectory = .
# subdirectory for sea ice history files
seaIceHistorySubdirectory = .

# names of namelist and streams files, either a path relative to baseDirectory
# or an absolute path.
oceanNamelistFileName = mpaso_in
oceanStreamsFileName = streams.ocean
seaIceNamelistFileName = mpassi_in
seaIceStreamsFileName = streams.seaice

# name of the ocean and sea-ice mesh (e.g. EC30to60E2r2, WC14to60E2r3,
# ECwISC30to60E2r1, SOwISC12to60E2r4, oQU240, etc.)
mpasMeshName = mesh

# Large datasets can encounter a memory error.  Specification of a maximum
# chunk size `maxChunkSize` can be helpful to prevent the memory error.  The
# current maximum chunk size assumes approximately 64GB of ram and large files
# with a single time slice.
maxChunkSize = 10000


# Whether missing input data should produce an error.  If not, the user gets
# a warning and the time bounds are adjusted to the beginning and end of the
# available data
errorOnMissing = False

Input Directories

The input directories are specified through a base directory and optionally separate subdirectories for the run (continaing namelist, streams and and one or more restart files for each MPAS component) and for each component’s simulation output (“history” files). You will always need to set baseDirectory in a custom config file.

By default, all files are assumed to be located in the base directory. However, E3SM supports short-term archiving of component output to separate directories. If this feature was enabled for the E3SM run to be analyzed, the configuration should be:

baseDirectory = /dir/for/model/output
runSubdirectory = run
oceanHistorySubdirectory = archive/ocn/hist
seaIceHistorySubdirectory = archive/ice/hist

where /dir/for/model/output is replaced with the directory containing the run and archive directories.

Namelist Files

MPAS components are configured with a namelist file containing a very large number of parameters and other configuration settings. Part of the strength of MPAS-Analysis is that it is aware of these namelists and can automatically disable analysis tasks that are not supported under a given configuration. By default, the namelist files for ocean and sea ice components are:

oceanNamelistFileName = mpaso_in
seaIceNamelistFileName = mpassi_in

For older E3SM (v1 alpha and beta) runs, a different naming convention was used and these options will need to be updated to:

oceanNamelistFileName = mpas-o_in
seaIceNamelistFileName = mpas-cice_in

The location of the namelist files is relative to the baseDirectory, so if they are located within a run directory inside the base directory, they should instead be specified as:

oceanNamelistFileName = run/mpaso_in
seaIceNamelistFileName = run/mpassi_in

Streams Files

Streams files are used to configure input and output from MPAS components. MPAS-Analysis can parse these files to determine the locations of output files (using the *HistorySubdirectory to find them if they have been moved by short-term archiving). Similarly to namelist files, there are default names:

oceanStreamsFileName = streams.ocean
seaIceStreamsFileName = streams.seaice

alterations appropriate for E3SM v1 alpha and beta runs:

oceanStreamsFileName = streams.ocean
seaIceStreamsFileName = streams.cice

and the addition of the run/ subdirectory if analyzing a run that used short-term archiving:

oceanStreamsFileName = run/streams.ocean
seaIceStreamsFileName = run/streams.seaice

Mesh Name

The MPAS-Ocean and MPAS-Seaice run on the same mesh. There are a number of standard E3SM ocean/sea ice meshes at various resolutions. The meshes currently supported by the public release of MPAS-Analysis include:

oEC60to30v3: An Eddy-Closure (EC) mesh with 30-km resolution at the poles and equator and 60-km resolution at mid latitudes,

oRRS30to10v3: A Rossby-Radius-Scaled (RRS) mesh with 10-km resolution at the poles and 30-km resolution at the equator,

oRRS18to6: An RRS mesh with 6-km resolution at the poles and 18-km resolution at the equator.

Mapping files (see Mapping Files below) and region mask files (see Region Groups) are provided from the E3SM public data repository for these meshes. For assistance with other mesh resolutions, please contact the MPAS-Analysis developers.

Xarray and Dask

MPAS-Analysis makes extensive use of the xarray package, which uses the dask package internally to perform operations that are too large to fit in memory. While most tasks in MPAS-Analysis have moved away from opening multi-file data sets using xarray in favor of concatinating these data sets together using NCO tools, there are some legacy options that users can modify if they experience errors related to dask:

maxChunkSize = 10000

If an out of memory error occurs, it may first be worth reducing the number of parallel tasks running (see Execute) but if the error is clearly related to dask (which might be the case, for example, if the error occrus in the streamfunctionMOC task), you may wish to reduce the maxChunkSize. This will make tasks using dask slower but will reduce their memory usage.

Errors on Missing Data

if errorOnMissing = False, the time ranges (startYear and endYear) in climatology, timeSeries, and index will be clipped to the range of the available data. If this option is set to True, an error will be produced. A value of end can be used for endYear to indicate that the full range of the available data should be used.