Weather Forecast Verification Using Matt Pocernich Research Application Laboratory National Center for Atmospheric Research September 29th, 2006
DWFE Predicted Reflectivity, t=+48h Facets of Research Applications Lab Decision Support Societal Impacts * Aviation Applications * Hydrometeorological Applications * National Security Applications Weather Systems and Assessment Forecast Verification and Statistics Pronóstico Observado Falsas alarmas Ayudas Ausentes DWFE Predicted Reflectivity, t=+48h WRF-NMM 10:00 Z, 8 Mar 05
Why Verify? Quantify forecast improvements Identify forecast limitations Compare models Provide diagnostics to model developers and forecasters
Types of forecasts Binary or multiple categories forecasts Point forecasts of continuous variables Probabilistic forecasts Chance of an event Full probability distribution Ensemble forecasts Grid-base spatial forecasts
Active area of research Biased data - spatially, temporally, as a function of intensity Spatial forecasts - issues are complicated by alignment, geometry , intensity and varying scale. Spatial probabilistic ensemble forecasts Developing scale statistics incorporating values of different users.
Why R? R is Open Source! R is free Runs on all operating systems All code is visible Over 1,000 packages donated. Very large, active user base.
Origins of R Influenced by two existing languages: Becker, Chambers & Wilks' S (ATT) Sussman's Scheme Initially written by Ross Ihaka and Robert Gentleman R-0.49 23-Apr-1997 05:53 959k R-2.4 October 5, 2006 ~ 14Meg
The R Community Developers Mailing lists R Core Group (17 members), only 2 have left since 1997 Major update in April/October (freeze dates, beta versions, bug tracking, ...) Mailing lists Help list ~ 100 messages/day, archived, searchable. Packages – more than 1000 packages (9/2006)
r-project.org Contains everything Source code Documentation Newsletter Mailing list Packages
Exercise 1 (Open ex1.r script) Exploring the R environment Sources of help Importing/ exporting data Data frames and classes of objects Simple plots