Newer
Older
#### Assessing Skill of Climate Forecasts on Seasonal-to-Decadal Timescales
The Climate Services Tools, CSTools, is an easy-to-use R package designed and built to assess and improve the quality of climate forecasts for seasonal to multi–annual scales. The package contains process-based state-of-the-art methods for forecast calibration, bias correction, statistical and stochastic downscaling, optimal forecast combination and multivariate verification, as well as basic and advanced tools to obtain tailored products.
This package was developed in the context of the ERA4CS project MEDSCOPE and the H2020 S2S4E project and includes contributions from ArticXchange project founded by EU-PolarNet 2. This GitLab project allows you to monitor its progress and to interact with other developers via the Issues section.
A scientific publication including use cases was published in the Geoscientific Model Development Journal, and it can be cited as follows:
> Pérez-Zanón, N., Caron, L.-P., Terzago, S., Van Schaeybroeck, B., Lledó, L., Manubens, N., Roulin, E., Alvarez-Castro, M. C., Batté, L., Bretonnière, P.-A., Corti, S., Delgado-Torres, C., Domínguez, M., Fabiano, F., Giuntoli, I., von Hardenberg, J., Sánchez-García, E., Torralba, V., and Verfaillie, D.: Climate Services Toolbox (CSTools) v4.0: from climate forecasts to climate forecast information, Geosci. Model Dev., 15, 6115–6142, https://doi.org/10.5194/gmd-15-6115-2022, 2022.
A part from this GitLab project, that allows you to monitor CSTools progress, to interact with other developers via the Issues section and to contribute, you can find:
- The CRAN repository [https://CRAN.R-project.org/package=CSTools](https://CRAN.R-project.org/package=CSTools) which includes the user manual and vignettes.
- Video tutorials [https://www.medscope-project.eu/products/tool-box/cstools-video-tutorials/](https://www.medscope-project.eu/products/tool-box/cstools-video-tutorials/).
- Other resources are under-development such [training material](https://earth.bsc.es/gitlab/external/cstools/-/tree/MEDCOF2022/inst/doc/MEDCOF2022) and a [full reproducible use case for forecast calibration](https://earth.bsc.es/gitlab/external/cstools/-/tree/develop-CalibrationVignette/FOCUS_7_2).
- See and run package [**use cases**](inst/doc/usecase.md)
Installation
------------
CSTools has a system dependency, the CDO libraries, for interpolation of grid data
and retrieval of metadata. Make sure you have these libraries installed in the
system or download and install from
[https://code.zmaw.de/projects/cdo](https://code.zmaw.de/projects/cdo).
You can then install the public released version of CSTools from CRAN:
```r
# install.packages("devtools")
devtools::install_git("https://earth.bsc.es/gitlab/external/cstools.git")
```
Overview
--------
The CSTools package functions can be distributed in the following methods:
- **Data retrieval and formatting:** CST_Start, CST_SaveExp, CST_MergeDims, CST_SplitDim, CST_Subset, CST_InsertDim, CST_ChangeDimNames, as.s2dv_cube and s2dv_cube.
- **Classification:** CST_MultiEOF, CST_WeatherRegimes, CST_RegimsAssign, CST_CategoricalEnsCombination, CST_EnsClustering.
- **Downscaling:** CST_Analogs, CST_RainFARM, CST_RFTemp, CST_AdamontAnalog, CST_AnalogsPredictors.
- **Correction and transformation:** CST_BiasCorrection, CST_Calibration, CST_QuantileMapping, CST_Anomaly, CST_BEI_Weighting, CST_DynBiasCorrection.
- **Assessment:** CST_MultiMetric, CST_MultivarRMSE
- **Visualization:** PlotCombinedMap, PlotForecastPDF, PlotMostLikelyQuantileMap, PlotPDFsOLE, PlotTriangles4Categories, PlotWeeklyClim.
An `s2dv_cube` is an object to store ordered multidimensional array with named dimensions, specific coordinates and stored metadata (in-memory representation of a NetCDF file). Its “methods” are the **CST** prefix functions. The basic structure of the class `s2dv_cube` is a list of lists. The first level elements are: `data`, `dims`, `coords` and `attrs`. To access any specific element it will be done using the `$` operator.
As an example, this is how an `s2dv_cube` looks like (see `lonlat_temp_st$exp`):
Data [ 279.99, 280.34, 279.45, 281.99, 280.92, ... ]
Dimensions ( dataset = 1, var = 1, member = 15, sdate = 6, ftime = 3, lat = 22, lon = 53 )
Coordinates
* dataset : dat1
* var : tas
member : 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15
* sdate : 20001101, 20011101, 20021101, 20031101, 20041101, 20051101
ftime : 1, 2, 3
* lat : 48, 47, 46, 45, 44, 43, 42, 41, 40, 39, 38, 37, 36, 35, 34, ...
* lon : 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, ...
Attributes
Dates : 2000-11-01 2001-11-01 2002-11-01 2003-11-01 2004-11-01 ...
varName : tas
metadata :
lat
units : degrees_north
long name : latitude
lon
units : degrees_east
long name : longitude
ftime
units : hours since 2000-11-01 00:00:00
tas
units : K
long name : 2 metre temperature
Datasets : dat1
when : 2023-10-02 10:11:06
source_files : "/ecmwf/system5c3s/monthly_mean/tas_f6h/tas_20001101.nc" ...
load_parameters :
( dat1 ) : dataset = dat1, var = tas, sdate = 20001101 ...
This package is designed to be compatible with other R packages such as [s2dv](https://CRAN.R-project.org/package=s2dv), [startR](https://CRAN.R-project.org/package=startR), [CSIndicators](https://CRAN.R-project.org/package=CSIndicators), [CSDownscale](https://earth.bsc.es/gitlab/es/csdownscale).
> **Note:** The current `s2dv_cube` object (CSTools version > 5.0.0) differs from the original object used in the previous versions of the packages. If you have doubts on this change you can follow some of the issues: [New s2dv_cube object discussion](https://earth.bsc.es/gitlab/external/cstools/-/issues/94), [How to deal with the compatibility break](https://earth.bsc.es/gitlab/external/cstools/-/issues/112) and [Testing issue and specifications](https://earth.bsc.es/gitlab/external/cstools/-/issues/110). More information can be found in this document: [About the new ‘s2dv_cube’](https://docs.google.com/document/d/1ko37JFl_h6mOjDKM5QSQGikfLBKZq1naL11RkJIwtMM/edit?usp=sharing).
Contribute
----------
Before adding a development, we suggest to contact the package mantainer. Details on the procedure and development guidelines can be found in [this issue](https://earth.bsc.es/gitlab/external/cstools/-/issues/3).
If you plan on contributing, you should rather clone the project on your workstation and modify it using the basic Git commands (clone, branch, add, commit, push, merge, ...).
The code of each function should live in a separate file with the .R extension under the R folder, and the documentation of each function should live in a separate file with the .Rd extension under the man folder.
For an introductory video on Git, you can have a look at [https://vimeo.com/41027679](https://vimeo.com/41027679).
You can also find all the necessary documentation on git here: [https://git-scm.com/book/en/v2](https://git-scm.com/book/en/v2). A lot of it may be a bit complicated for beginners (and not necessary for us), but the "Getting started" and "Git basics" sections are a good resources.