metrics.msss

This class is for calculating the MSSS (Mean Squared Skill Score) following the paper of Goddard et al. (2012)

author:Sebastian Illing contact:sebastian.illing@met.fu-berlin.de

MSSS Module

class metrics.msss.Msss(output='/tmp/msss/output/', output_plots='/tmp/msss/output/', decadals='1960, 1965, 1970, 1975, 1980, 1985, 1990, 1995, 2000', variable='tas', project1='baseline1', product1='output', institute1='mpi-m', model1='mpi-esm-lr', experiment1='decs4e', ensemblemembers1='*', project2='baseline0', product2='output1', institute2='mpi-m', model2='mpi-esm-lr', experiment2='decadal', ensemblemembers2='*', leadtimes='1, 2-9', observation='HadCrut', observation_ensemble='*', maskMissingValues=True, result_grid=None, cache='/tmp/msss/cache/', baseDir='..', timeFreq='mon', basic_output=True, bootstrap=None, bootstrap2=None, obsRemapped=None, leadtimes_mode='yearly', level=None, lonlatbox=None, fieldmean=False, zonalmean=False, colormap='RedBlu', observation_type='REANALYSIS', reanpath='', months=None)

Class to calculate the MSSS. Call method “analyze” to get results

Parameters:

• output – Path where the results are saved
• output_plots – Path where the plots are saved
• decadals – a list with decadal experiments, i.e. [1960,1960,...,1995]
• variable – CMOR-Name of variable like ‘tas’ (Near Surface Temperature)
• project1 – CMOR-Name of the project like CMIP5 or BASELINE1
• product1 – CMOR-Parameter
• institute1 – CMOR-Name of the institute like MPI-M
• model1 – CMOR-Name of the used model like MPI-ESM-LR
• experiment1 – Experiment name like decadal or historical
• ensemblemembers1 – Comma separated string of esemble members like ‘r1i1p1,r2i1p1,...’
• project2 – CMOR-Name of the project like CMIP5 or BASELINE1
• product2 – CMOR-Parameter
• institute2 – CMOR-Name of the institute like MPI-M
• model2 – CMOR-Name of the used model like MPI-ESM-LR
• experiment2 – Experiment name like decs4e or historical
• ensemblemembers2 – Comma separated string of esemble members like ‘r1i1p1,r2i1p1,...’
• leadtimes – Leadtimes to analyze, ie. 1,2-9
• observation – Observation or Reanalysis Experiment, HadCrut or ERA-Int
• observation_ensemble – If your observation has multiple ensemble members. Not necesary if you use an observation file
• maskMissingValues – Boolean
• result_grid – Griddescription of resultgrid like r72x36
• cache – Path for cachedir during analysis
• baseDir – Path of Class to find default files
• timeFreq – Timefrequency of the files, ie ‘mon’ or ‘day’
• basic_output – Boolean - If true only basic out is produced (Correlation, MSESS, and Conditional Bias).
• bootstrap –
• bootstrap2 –
• obsRemapped –
• level – For 3D-Files –> Select a single level
• lonlatbox – If you want to select a specific lonlat-Box
• fieldmean – Boolean - If you want to calculate field means
• zonalmean – Boolean - If you want to calculate zonal means
• colormap – Matplotlib Colormap used for ploting
• observation_type – NOT USED ANYMORE!!!
• reanpath – NOT USED ANYMORE!!!
• months – Usefull if your analysing monthly and your experiment name has a month included

analyze()

Main method to calculate the MSSS. Following steps are performed:

1. Searching for file using solr_search
2. Remapping Files to coarser grid
3. Calculating Ensemble Mean
4. Calculating cross-validated mean
5. Calculating anomalies for both models
6. Analyzing different timeranges

analyzeYearRange(input1Anomalies, observationRemapped, input2Anomalies, startYear, endYear)

Calculate the MSSS for given timerange. The following steps are perfomed in this method:

1. temporal smoothing
2. calculation of correlation, bias, and msss
3. calculation of model1 vs. model2
4. plotting of the 9 fields

Todo :

maybe this could be done in multiprocessing –> so different timeranges at once

Parameters:

• input1Anomalies – dict with anomalies of model1
• observationRemapped – dict with anomalies of observations
• input2Anomalies – dict with anomalies of model2
• startYear –
• endYear –

getInitVsUninit(initResults, uninitResults, flag, outputDir)

Calculates comparing values between two models/modelversions i.e. initialized vs. uninitialized. Following variables are computed:

• correlation: corr1 - corr2
• conditional bias: abs(bias1) - abs(bias2)
• msss: (msss1-msss2)/(1-msss2)

Parameters:

• initResults – dict with “corr”, “bias”, and “msss” of first model
• uninitResults – dict with “corr”, “bias”, and “msss” of second model
• tempRange – start- and endyear –> used for filenames

Returns:

msss filename

getCorrelationAndVariancesAndMSSS(hindcast, observation, flag, outputDir)

Calculates the following quantities:

• correlation bewtween observation and hindcast
• temporal variance of hindcast
• temporal variance of observation
• conditional bias
• msss
• biasSlope so/sh * c
• std ratio so/sh

Parameters:

• hindcast – dict of hindcasts
• observation – dict of observations
• flag – additional string for filenames

Returns:

dict with “msss”, “bias”, “corr”, “biasSlope, “stdRatio”, “biasSlopeMinusOne”

getMSE(anomalies, obsAnom)

Method calculates the Mean Squared Error.

Note :

method ist not used at the moment

Parameters:

• anomalies – dictionary with anomalies
• obsAnom – dict with observation anomalies

Returns:

file with mse field

getRMSS(MSSS, flag, outputDir)

Calculates the RMSSS using the MSSS

Parameters:

• MSSS – netcdf file
• flag – name flag
• outputDir – output path

Returns:

file with RMSSS field