Project

General

Profile

RelDiag

module load evaluation_system

analyze --tool reldiag --help

RelDiag (v1.0.0): 
    Plots reliability diagrams (by EnsDiVal), illustrating reliability, resolution and uncertainty of decadal experiments with respect to a reference (e.g., observation):
    The observed frequency given a forecast probability is plotted against the forecast probability.
    You can either use your own input and reference files or set the leadtimeselector flag to True to select model, variable, etc.

Options:
outputdir        (default: $USER_OUTPUT_DIR/$SYSTEM_DATETIME)
                 Choose the output directory.
plotdir          (default: $USER_PLOTS_DIR/$SYSTEM_DATETIME)
                 Choose the plot directory.
cache            (default: $USER_CACHE_DIR/$SYSTEM_TIMESTAMP)
                 Choose the cache directory.
cacheclear       (default: True)
                 Option switch to clear the cache.
filehead         (default: <undefined>)
                 Choose the file path/file-name head of the input data.
                 MANDATORY if the leadtimeselector is not used. All files
                 "filehead*" will be considered.
refhead          (default: <undefined>)
                 Choose the file path/file-name head of the reference data.
                 MANDATORY if the leadtimeselector is not used. If this is more
                 than one file (e.g., when selecting two lead times), all files
                 "refhead*" will be merged into one time series.
lonlatbox        (default: -180,180,-90,90)
                 Choose region to produce time series (field mean) from the
                 input and reference data.
reflonlatbox     (default: <undefined>)
                 Choose optional region for an anomaly calculation: The time
                 series (field mean) from this box will be subtracted from the
                 time series from the lonlatbox. Maybe useful to remove a
                 (global) trend, etc.
bins             (default: 11)
                 Choose number of bins for the forecast probabilities. Use 0
                 (zero) for no binning. Number of ensemble members plus one
                 might be a good choice as it provides disjunct classes.
threshold        (default: timmean IFILE)
                 Choose threshold arguments forwarded to cdo. Use the wildcard
                 strings IFILE and TSTEPS for any file in your input path and
                 number of timesteps in the files, respectively. Examples: "add
                 -timmean IFILE -timstd IFILE" to set the threshold to the mean
                 plus one standard deviation or "runpctl,50,TSTEPS IFILE" to set
                 the threshold to the median.
leadtimeselector (default: False)
                 Option switch to use the leadtimeselector.
leadtimes        (default: <undefined>)
                 Leadtimes to analyze (comma separated list possible).
variable         (default: <undefined>)
                 Choose variable to analyze (CMOR).
model            (default: <undefined>)
                 Choose model of experiment, e.g. MPI-ESM-LR.
project          (default: <undefined>)
                 Choose project, e.g. baseline0, baseline1, prototype.
institute        (default: <undefined>)
                 Choose institute of experiment, e.g. MPI-M.
product          (default: <undefined>)
                 Choose product, e.g. output1, output.
ensembles        (default: <undefined>)
                 Comma separated list of ensemble members. Use * for all
                 members.
time_frequency   (default: <undefined>)
                 Chose frequency of the input data, e.g. mon, yr, day.
decadals         (default: <undefined>)
                 Comma separated lists of decadal experiments.
observation      (default: <undefined>)
                 Observation to compare. I.e. merra or hadcrut3v. You can also
                 specify a path to your observation file.

example on command line:

analyze --tool reldiag leadtimeselector=True variable=tos leadtimes=1 model=mpi-esm-lr project=baseline0 institute=mpi-m product=output1 ensembles=r1i1p1,r2i1p1,r3i1p1 time_frequency=mon decadals=1960:2010 observation=ersst bins=4 lonlatbox=-80,0,0,60 reflonlatbox=-180,180,-60,60