Version 14 (modified by juaco, 6 years ago) (diff)



The latest versions (stable and/or devel) can be installed directly from gitHub, but please note that the R package devtools must be installed first to ease the installation process.

First check that devtools is installed on your system, otherwise install it by typing:

if (!require("devtools")) install.packages("devtools")

Then, the loadeR.ECOMS package and associated dependencies are installed by entering the following command (it is important to preserve the ordering of the arguments):


If attempting the installation from a proxy server and getting an error, please try this.

Alternatively, you can directly download the sources of the three packages for a local installation from their respective pages:

  2. loadeR
  3. loadeR.ECOMS

Again, bear in mind that order matters, and need to be preserved during installation.


A history of all package releases until the most recent is available on the loadeR.ECOMS releases page

loadeR.ECOMS supersedes the older package ecomsUDG.Raccess, which is no longer maintained. This is an historic of old versions:

Older versions of ecomsUDG.Raccess (deprecated)

  • version 4.2-0 -- "EC-Earth" (03 Nov 2015)
    • New SMHI-EC-EARTH_EUPORIAS hindcast available
    • Minor bug fix in monthly aggregations of single-month season queries
    • Other minor bug fixes and enhancements
  • version 4.1-0 -- "Interim" (08 Oct 2015)
    • ERA Interim dataset included
    • NCEP dataset renamed to NCEP_reanalysis1
    • New vertical level variables included in S4_seasonal_15 dataset
    • Improved connection error handling
    • Minor bug fix in the variable deaccumulation of lead month 0 S4 predictions (affects radiation and precip)
    • Other minor bug fixes and enhancements
  • version 4.0-0 -- "Bolzano" (15 May 2015)
    • CFSv2 dataset redefinition
      • Improved performance of data download
      • New member definition to avoid errors due to missing runtimes in the original dataset
      • New surface variables available, including wind speed (derived from north and eastward components)
    • New time aggregation options, including:
      • greater flexibility for daily
      • new monthly aggregation feature
    • Improved authentication scheme providing greater stability for long queries
    • Added new attributes containing relevant metadata (longname, units, time aggregation details...)
    • Several minor bug fixes and enhancements
  • version 3.0-0 (29 Apr 2015)
    • New authentication scheme implemented
    • New CFSv2 dataset definition
    • Several minor bug fixes and enhancements
  • version 2.2-6 (27 Jan 2015)
    • Bug fix in the retrieval of forecast dates beyond the last year of the runtime axis (Reported by Wietse Franssen)
    • Enhancement in System4 deaccumulation of precipitation for lead month 0 queries. First day preserved for consistency with non-deaccumulated variables (Suggested by Kathryn Nicklin).
    • Documentation update
    • Other minor bug fixes and enhancements
  • version 2.2-5 -- "SPECS Workshop" (06 Sep 2014)
    • Minor bug fixes and enhancements proposed by the participants during the practical sessions of the SPECS Hands-on Training School on Seasonal Forecasting and Downscaling.
  • version 2.2-4 (06 Sep 2014)
    • Internal changes in imports for compatibility with the new downscaleR version 0.4-x
  • version 2.2-3 (18 Aug 2014)
    • Bug fix for 1-dimensional output data queries (i.e. time series at single point locations, without members)
  • version 2.2-2 (18 Aug 2014)
    • New global attributes in output (thanks to Stefan Siegert for this suggestion):
      • dataset: Name of the dataset returned (e.g. "System4_seasonal_15", "NCEP", etc.)
      • source: Name of the dataset
      • URL: URL of the data portal
    • Consistent ordering of array dimensions: The output n-dimensional array now preserves the canonical ordering of its dimensions: (member, time, level, lat, lon) (thanks to Stefan Siegert for this suggestion)
    • New daily aggregation options: minimum and maximum daily data are now returned when using the time = "DD" option for the relevant variables (e.g. "tasmin", "tasmax" ...).
  • version 2.2-1 (05 Aug 2014)
    • Minor enhancements:
      • Improved on screen error messages with clearer instructions for error fixing
      • The 'runtime' dimension in CFSv2 output data array has been changed to 'member' for better integration with other downscaleR objects and methods.
  • version 2.2-0 (16 Jul 2014)
    • New dependency on R Package downscaleR with inherited features such as:
      • Plotting mean fields
      • Fast Multi-member Interpolation/re-gridding capabilities
      • Many more coming in the next major downscaleR release (bias correction, perfect-prog downscaling methods...)
    • On screen messages from the HTTP Java authenticator have been suppressed: Only the strictly relevant information is now displayed
    • Automatically checks and warns the user about new available versions on attach
    • Other minor bug fixes and enhancements
  • version 2.1-1 (11 Jul 2014)
    • Bug fix accessing surface air temperature and derived variables in System4 with mean daily temporal aggregation (thanks to M.D. Frías for pointing to the error)
  • version 2.1-0 (8 Jul 2014)
    • New extended list of available variables
    • NCEP reanalysis included in available datasets
    • On-the-fly computation of derived variables
    • Support for variables with vertical levels and static (e.g. geopotential surface zs)
    • New dependency on R package containing Java dependencies
  • version 2.0-0 (16-jun-2014).
    • New input/output format
    • Access to a extended list of surface variables
    • New observational gridded dataset WFDEI
    • On-the-fly time filtering/aggregation capabilities
  • version 1.0-0 (17-feb-2014). Access to a limited list of surface variables for System4 and CFSv2 datasets.

Development version

The development version is available at the 'devel' branch of the ​loadeR.ECOMS gitHub repository, but please note that the development version is unstable and may not be always functional