Skip to content

Home

Jump to bottom
Rasmus E. Benestad edited this page Sep 10, 2024 · 27 revisions

Welcome to the esd wiki!

The esd R package or library is made freely available by the Norwegian Meteorological Institute (MET Norway) and is for use by the climate community (see documentation and 'Heavy MET talk' on YouTube). It was primarily built for statistical downscaling of climate variables and parameters form global climate models (e.g. ENSEMBLEs, CMIP3/5/6 model results), and has recently been extended to deal with data I/O (e.g. global climate datasets), statistical analysis, and visualization. It also incorporates various aspects of machine learning (ML) techniques, and we hope it will provide a toolbox that is a step towards a solution to the problem described in the opinion piece by Greene and Thirumalai (2019).

Once the esd R package is installed, you can get access to a compendium on empirical-statistical downscaling and the manual cited above (R will download a PDF-document)

ABC4ESD()
manual()

If you use esd and want to register, please use our form. If we know that there are many who use it, then it will give us stronger justification to spend our effort on further developement. We also have a Facebook page where we announce changes etc, and there is also information about the package on https://sites.google.com/met.no/r-esd.

Please cite the esd package as following: Benestad et al.. (2015). esd V1.0. Zenodo. 10.5281/zenodo.29385 (http://dx.doi.org/10.5281/zenodo.29385). You can also view the contents of these reports on http://metno.github.io/esd/.

The esd-package is a new generation of analysis and downscaling tool that originated from clim.pact, and the compendium on downscaling still applies to esd in general, but the examples are old and specific to clim.pact. Its principles are similar to those of Tidyverse.

It consists in i) retrieving and manipulating large samples of climate and model data from various sources, ii) computing Empirical Orthogonal Functions (EOFs), Principal Component Analysis (PCA), and Multi-Variate Regressions (MVRs) using very simple R commands and procedures (See Examples).

The esd library has been built with the emphasis of traceability, compatibility, and transparency of the data, methods, procedures, and results.

As the library has been built on R system, it inherits from the large number of R built-in functions and procedures.

The esd library can be easily tailored to various downscaling projects with few adaptations such as the few examples shown here. Further examples are presented as Rmarkdown files and Rshiny apps also available on github.com.

The main functionalities of the esd are

Data handling

  • Formatted datasets (data(package='esd'))
  • Select weather station from meta data (stationmeta) based on a list of selection criteria (select.station())
  • Retrieving data from observational weather stations included in the package (station())
  • Retrieving field data from a netcdf file (retrieve())
  • Regridding (regrid())
  • Subsetting (subset())
  • Aggregating - Spatial and temporal aggregation (aggregate())

Data processing

  • Empirical Orthogonal Functions for fields' objects (EOF())
  • Principal component Analysis for stations' objects (PCA())

Downscaling

  • Downscale an ensemble of stations (DS())
  • Downscale an ensemble of stations from an ensemble of gcms (DSE()).

Visualizations

  • Plots for esd objects (plot())
  • Maps for esd objects (map())

Presentations

Copyright of MET Norway 2022

Clone this wiki locally