There are many packages out there related to computing metrics on initialized geoscience predictions. However, we didn’t find any one package that unified all our needs.
Output from earth system prediction hindcast (also called re-forecast) experiments is difficult to work with. A typical output file could contain the dimensions initialization, lead time, ensemble member, latitude, longitude, depth. climpred leverages the labeled dimensions of xarray to handle the headache of bookkeeping for you. We offer HindcastEnsemble and PerfectModelEnsemble objects that carry products to verify against (e.g., control runs, reconstructions, uninitialized ensembles) along with your initialized prediction output.
When computing lead-dependent skill scores, climpred handles all of the init+lead-valid_time-matching for you, properly aligning the multiple time dimensions between the hindcast and verification datasets. We offer a suite of vectorized deterministic and probabilistic metrics that can be applied to time series and grids. It’s as easy as concatenating your initialized prediction output into one xarray.Dataset and running the HindcastEnsemble.verify() command:
HindcastEnsemble.verify(
metric="rmse", comparison="e2o", dim="init", alignment="maximize"
)
- Calculate skill for NWP model GEFS for 6-hourly global forecasts
- Calculate skill of a MJO Index of SubX model GEOS_V2p1 as function of daily lead time
- Calculate skill of a MJO Index of S2S models as function of daily lead time
- Calculate skill of a MJO Index of SubX model GEOS_V2p1 as function of weekly lead time
- Calculate skill of S2S model ECMWF for daily global reforecasts
- Calculate ENSO Skill of NMME model NCEP-CFSv2 as Function of Initial Month vs. Lead Time
- Calculate Seasonal ENSO Skill of the NMME model NCEP-CFSv2
- Demo of Perfect Model Predictability Functions
- Hindcast Predictions of Equatorial Pacific SSTs
- Diagnosing Potential Predictability
- Significance Testing
- Using dask with climpred
- climpred on CPU vs GPU
- Setting up your own output
- intake-esm for cmorized output
You can install the latest release of climpred using pip or conda:
pip install climpred[complete]conda install -c conda-forege climpredSee the "climpred" for detailed instructions
The detailed application of the technique can be checked through the link of each technique.
- Calculate skill of a MJO Index of SubX model GEOS_V2p1 as function of daily lead time
- Calculate skill of a MJO Index of S2S models as function of daily lead time
- Calculate skill of a MJO Index of SubX model GEOS_V2p1 as function of weekly lead time
- Calculate skill of S2S model ECMWF for daily global reforecasts
- Calculate ENSO Skill of NMME model NCEP-CFSv2 as Function of Initial Month vs. Lead Time
- Calculate Seasonal ENSO Skill of the NMME model NCEP-CFSv2
(Detailed guide to run program codes will be described here!)
example)
MJODiagnostics(Parameter,...)The latest releases of climpred can be found on climpred's github.
xskillscore is an open source project and Python package that provides verification metrics of deterministic (and probabilistic from properscoring) forecasts with xarray.
- Aaron Spring [github]: https://github.com/aaronspring/
- Riley X. Brady [github]: https://github.com/bradyrx/
- Andrew Huang [github]: https://github.com/ahuang11/
- Kathy Pegion [github]: https://github.com/kpegion/
- Anderson Banihirwe [github]: https://github.com/andersy005/
- Ray Bell [github]: https://github.com/raybellwaves/
(Contributors and relevant Acknowledgements for program codes will be listed here!)
example)
The MJOWG wishes to acknowledge and thank U.S. CLIVAR and International CLIVAR for supporting this working group and its activities
by MJO Simulation Diagnostics
See the "climpred" for detailed instructions
(Citation for program codes will be listed here!)
example)
National Center for Atmospheric Research Staff (Eds). Last modified 08 Oct 2013. "The Climate Data Guide: MJO: Madden-Julian Oscillation Diagnostics."
Any claims against the Institute stemming from the use of any GitHub-related project will be governed.