The hands-on training workshop on "seasonal forecasting and downscaling" was organized in Santander (Spain), 8-12 September 2014, by SPECS (WP 5.2 and RT 6) with the collaboration of EUPORIAS.
This workshop is designed specifically for the stakeholders and impacts communities of the ECOMS projects and the main goal is to train attendees in the use of different R-based tools to access, calibrate and downscale seasonal forecasting data. These tools are being developed as part of the SPECS and EUPORIAS activities, and include the ECOMS user data gateway (providing hindcast data from operational seasonal forecasting systems such as System4) and different R-packages for statistical downscaling. It will be also shown how to download and access the new SPECS seasonal forecasting simulations stored in BADC. In order to allow non-experienced R users to follow this workshop, it has been organized in two parts:
1) Atwo-days pre-workshop "an Introduction to R" (8-9 September) for trainees with no previous experience in R, and
2) a three-day workshop (10-12) including lectures on the fundamentals of bias correction and downscaling in the field of seasonal forecasting and hands-on training sessions using R.
This work will allow obtaining the regional and local seasonal predictions necessary for other tasks of the projects. Please note that the workshop is intended for the people who will actually work with the data.
Format and participation:
The workshop format will consist of lectures (30%) and hands-on training (70%) using the R software environment. The pre-workshop ("An introduction to R for climate data analysis") will only consist of hands-on training activities. The program will be shaped to enable group discussions and exchanges. The number of seats is limited to 30 and the hands-on sessions will be assisted by three trainers in order to guarantee convenient support.
The demos and hands-on sessions in this workshop are based on three R packages (accesible from GitHub):
ecomsUDG.Raccess. R package for accessing data (seasonal forecasts: System4, reanalysis: NCEP, etc.) from the ECOMS User Data Gateway.
downscaleR. An R package for statistical bias correction and downscaling.
esd. Climate analysis and empirical-statistical downscaling R package.
8-9 September 2014 (An introduction to R for climate data analysis)
J Bedia, S Herrera, MD Frias, D SanMartín, M Tuni
Quick R start: data structures, basic operations, control structures.
PRACTICE: using a climate data object
DEMO: Introduction to R objects [OO-Reference]
PRACTICE: Weather generators
10 September 2014
[09:00-09:15] JM Gutiérrez. Welcome and presentation
[09:15-10:00] JM Gutiérrez. Basic concepts on S2D forecasting
[10:00-10:30] M De Felice. Impact models and data I
[10:30-11:00] K Nicklin. Impact models and data II
[11:30-12:00] PA Bretonniere. SPECS experiments and access
[12:00-12:30] JM Gutiérrez. ECOMS User Data Gateway
[12:30-13:00] AS Cofiño, J Bedia. Demo: ECOMS-UDG [DEMO1: Accessing seasonal forecast data using R] [DEMO2: Validating and visualizing tercile-based probabilistic predictions]
11 September 2014
[09:00-09:30] J Fernández. Introduction to downscaling
[09:30-10:00] S Herrera. Introduction to MOS & bias correction methods
[10:00-10:30] R Wilke. Bias correction methods I
[10:30-11:00] J Bhend. Bias correction methods II
[11:30-12:15] J Bedia. Demo: The 'downscaleR' package. [downscaleR wiki]
[12:15-13:00] S Herrera, J Bedia. Hands-on training: Bias correction with 'downscaleR' [DEMO: Bias correction]
[14:30-17:30] S Herrera, J Bedia. Hands-on training: Bias correction with 'downscaleR' [PRACTICE: Applying Bias Correction Techniques to System4 hindcast]
12 September 2014
[09:00-10:00] JM Gutiérrez. Perfect prog downscaling methods
[10:00-10:30] N Cortesi. Analogues and weather typing
[10:30-11:00] R Benestad. Downscaling PDFs/parameters
[11:30-13:00] [90'] J Bedia. Demo and hands-on training with 'downscaleR' package [DEMO: Perfect prog downscaling with analogs] [PRACTICE: Downscaling System4]
[14:30-17:30] R Benestad, A. Mezghani. Demo and hands-on training: Statistical downscaling with 'esd' package. [DEMO] [PRACTICE]