Skip to content

Hands on Training

Sixto Herrera García edited this page Aug 25, 2014 · 9 revisions

Practice 1A. Validating seasonal forecast (System4 vs ECA) focusing on a point of interest.

In this practice we will work at a point level, choosing a location of particular interest to analyze System4 seasonal forecasts. Details on different stations are given in https://github.com/SantanderMetGroup/downscaleR/blob/master/inst/datasets/observations/GSN_Iberia/stations.txt Tasks to be done: 1) choosing a target season (e.g. MAM, i.e. 3:5), 2) loading the data using loadECOMS (https://meteo.unican.es/trac/wiki/EcomsUdg/RPackage/Examples/pointSelection) and loadStations, 3) obtaining seasonal mean/aggregated values, 4) plotting the observed and predicted inter-annual series, 5) validating the results (correlation, BSS for probabilistic-based tercile prediction). Key points: How to manage different members.

Practice 1B. As practice 1A but for a region (System4 vs WFDEI) Tasks to be done: As in Task 1A, but considering also the re-gridding of predictions (System4 -> 0.75 deg) and observations (WFDEI -> 0.5 deg). https://meteo.unican.es/trac/wiki/EcomsUdg/RPackage/Examples/continentalSelection Key points: Validating spatial aggregated predictions or aggregating the valuation at a gridbox level.

Practice 2A. Applying Standard Bias Correction Techniques to Seasonal Forecast.

The objective of this practice is to work with and to analyze the different bias correction techniques included in the loadECOMS and downscaleR packages. Then, we will use the observations and forecast loaded in the previous practices in order to use them as input for the bias correction functions.

Tasks to be done: 1) loading the data using loadECOMS (https://meteo.unican.es/trac/wiki/EcomsUdg/RPackage/Examples/) and loadStations, 2) analyzing the effect of the bias correction techniques in the training period, 3) applying the different techniques to adjust the forecast in an unobserved period, and 4) comparing both corrected and uncorrected forecast with different plots (maps, scatter-plots, etc.).

Key points: Identify the effect of the different bias correction techniques and point out their main advantages and shortcomings.

Clone this wiki locally