Eugenia Kalnay  - Department of Atmospheric and Oceanic Science. University of Maryland (USA)

1. Atmospheric predictability: from basic theory to forecasting practice.
2. Ensemble prediction and strategies for initialization: tangent linear and adjoint models, singular vectors, Lyapunov vectors.
3. Bred vectors: theory and applications in operational forecasting.
4. Local Ensemble Transform Kalman Filtering.
5. Review of the ensemble prediction systems used at NCEP.
6. Seminar: Comparison of 4D-Var and Ensemble Kalman Filter methods.

• lecture 1 [5.87MB]
• lecture 2 [950KB]
• lecture 3 [2.66MB]
• lecture 4 [1.86MB]
• lecture 5 [1.49MB]
• seminar [2.62MB]

Back to the program

Olivier Talagrand  - Laboratoire de Météorologie Dynamique. École Normale Supérieure (France)

1. Why probabilistic weather forecasts: theoretical fundaments of ensemble forecasting. Reminder of basics of the theory of probability and statistics.
2. Description of initialization methods for ensemble forecasting: singular modes, bred modes, ‘perturbed observations’, Ensemble Kalman Filter, Ensemble Transform Kalman Filter.
3. Theoretical aspects of objective evaluation of ensemble forecasting. Reliability and resolution. Objective scores for validation: reliability curve, Brier and Brier-like scores and their decomposition, ROC curve area, rank histograms, reduced centred random variable.
4. Advantages and limitations of ensemble predictions. Sampling effects (size of forecast ensembles, size of validating sample). Impact of observational errors.
5. Data assimilation and targeting in relation to ensemble forecasting. Economic value of ensemble forecasting.
6. Intercomparison of different initialization methods.

• lecture 1 [1.62MB]
• lecture 2 [1.31MB]
• lecture 3 [949KB]
• lecture 4 (paper) [631KB]
• lecture 5 [686KB]
• lecture 6 [351KB]

Back to the program

Thomas M. Hamill  - NOAA/ESRL, Physical Sciences Division (USA)

1. Visualization and verification of ensemble forecasts. Methods for forecasters to visualize the information. Common methods of verification.
2. Hydrologic ensemble prediction: sources of predictability. Requirements of ensemble weather forecast prediction systems for driving hydrologic models. Verification of hydrologic forecasts.
3. Limited-area ensemble forecasting. Scale interactions and effects of boundary conditions. Perturbation methods for the mesoscale. Tradeoff of domain size vs. grid spacing vs. ensemble size.
4. Severe-weather and extreme-event forecasting: predictability issues. Model errors at the mesoscale and the microscale. Review of severe-weather ensemble forecast experiments at the NOAA Storm Prediction Center.
5. Multi-model, and superensemble forecasts. Calibration techniques. Review of experiments from TIGGE multi-model ensemble data set.
6. Seminar: Use of reforecasts in probabilistic ensemble forecast calibration.

• lecture 1 [7.56MB]
• lecture 2 [9.95MB]
• lecture 3 [7.58MB]
• lecture 4 [8.57MB]
• lecture 5 [2.52MB]
• seminar [6.59MB]
• extra contents[5.38MB]

Back to the program

Tiziana Paccagnella  - Agenzia Regionale Prevenzione e Ambiente della Regione Emilia-Romagna Servizio IdroMeteorologico (Italia)

• The TIGGE-LAM project in the context of the THORPEX Interactive Grand Global Ensemble (TIGGE). Download PDF: seminar [5.34MB]

Marino Marrocu  - Centro di Ricerca Sviluppo e Studi Superiori in Sardegna (Italia)

• Skill comparison of different calibration methods for a multi-model multi-analysis limited area ensemble. Download PDF: seminar [13.0MB]

José Antonio García-Moya Zapata  - Istituto Nacional de Meteorología (España)

• Multimodel/Multinalysis Ensemble applied to the Mediterranean region.
  Download PDF: seminar [5.34MB]

Jutta Thielen  - JRC Joint Research Centre (Italia)

• Application of EPS in flood forecasting. Download PDF: seminar [5.05MB]

Back to the program