EPS Training Module 1: Introduction Richard Verret (Normand Gagnon) Meteorological Service of Canada The illusion of determinism…

EPS Training

Module 1: Introduction

Richard Verret (Normand Gagnon)Meteorological Service of Canada

The illusion of determinism…

Module 1: Introduction – Page 2R. Verret –

Welcome to the EPS Training session

Module 1: Introduction – Page 5R. Verret –

Context

• Ensemble prediction system (EPS) have been around for a long time:

– An Ensemble forecast can be thought of as a collection of two or more Numerical Weather Prediction (NWP) model forecasts verifying at the same time:

▪ Helps to gain a feel for possibilities of pattern evolution.

▪ Helps to partially gage confidence in a particular model solution.

Module 1: Introduction – Page 7R. Verret –

48-h GEM regional 48-h GEM global 48-h NCEP-GFS

48-h NCEP-NAM 48-h UKMetO 48-h ECMWF

Context

M.-F. Turcotte, CMCMean sea level pressure – all valid at 12 UTC February 15 2007

Module 1: Introduction – Page 8R. Verret –

Context

• Forecasts are usually generated within the deterministic paradigm – one scenario is selected.

• Determinism is favoured by:– Heritage of the past.– Improvements of high resolution Numerical Weather Prediction (NWP)

models.– Satellite and radar (and other) remote sensing technologies.

>>> Over-confidence in NWP models <<<

• Clients are not necessarily prepared to use probabilistic forecasts or measures of forecast uncertainty:

– Training is required both on the users side and on the forecasters’ side.

Module 1: Introduction – Page 11R. Verret –

Models look quite realistic these days… HRDPS at 2.5 km

• http://iweb.cmc.ec.gc.ca/~afsgfau/LAM2.5km/NationalDomain/loop20_N1.gif

• http://iweb.cmc.ec.gc.ca/~afsgfau/LAM2.5km/NationalDomain/loop20_N2.gif

M. Faucher, CMC/CMDN

Module 1: Introduction – Page 12R. Verret –

T. Robinson, CMC

~10 years ~10 years

48-h gain in predictability in ~ 20 years

NWP is improving!

Module 1: Introduction – Page 13R. Verret –

Context

120-h integration – mean sea level pressureTwo integrations done with identical NWP models but on different computers

M. Lajoie, CMC

Module 1: Introduction – Page 14R. Verret –

Context

Precipitation amountdifference > 45mm

MSL pressure

Bit flipping experiment

N. Gagnon, CMC

240-h forecast

Module 1: Introduction – Page 15R. Verret –

Context

• Ensemble forecasts have evolved significantly over the past years:

– Systematic approach to model uncertainty.– Perturbations as simulation of uncertainty.– Better simulation of uncertainties in forecast processes.– Increasing number of members.– Increasing resolution of members.

• With Ensemble forecast, it is possible to evaluate, express and forecast uncertainty.

Module 1: Introduction – Page 16R. Verret –

Context

• An Ensemble Prediction System is a set of integrations of one or several NWP models that differ in their initial states (and sometimes in their configurations and boundary conditions).

• Ensemble prediction is an attempt to estimate the non-linear time evolution of the forecast error probability distribution function.

• Ensemble prediction is a potential method of estimating forecast predictability beyond the range in which error growth can be described by linearized dynamics.

Module 1: Introduction – Page 17R. Verret –

ContextInitial states Final states

True initial stateTrue final state

Climatology

Ensemble mean

Analysis

Deterministicforecast

Uncertainty oninitial state

R. Verret, N. Gagnon, CMC

Module 1: Introduction – Page 18R. Verret –

Context

• Common usages of Ensemble forecasts:– Ensemble mean as a substitute for a single deterministic

forecast.

– Clustering to produce a small set of forecast states characterized with the cluster mean.

– A priori prediction of forecast skill.

– Ensemble probability distribution function.

– Measure of uncertainty.

– Extension of forecast range.

Module 1: Introduction – Page 19R. Verret –

Context

• There is an important research effort on EPS around the world:

– Research done at most EPS producing Centers.– THORPEX (THe Observing system Research and Predictability EXperiment).– NAEFS (North American Ensemble Forecast System).

• There is an important effort devoted to the usage of EPS:

– At each EPS producing Centers.– NAEFS.

Module 1: Introduction – Page 24R. Verret –

Expected results

• Shift from a deterministic paradigm toward one where uncertainty is part of forecasts:

– EPS can provide flow-dependent predictive probability distribution for future weather quantities or events.

– Probabilistic forecasts allow one to quantify weather-related risks and show greater economic value than deterministic forecasts.

– Ensemble forecasts are not meant to be a consensus technique.

Module 1: Introduction – Page 26R. Verret –

Expected results

• Overall result:– Develop a motivation to use EPS outputs/products.

• Module 2 – probabilistic forecasts:– Understanding of basic concepts in probability.

• Module 3 – EPS basic concepts:– Understanding of basic concepts in Ensemble forecasting.

• Module 4 – EPS construction:– Basic understanding of how EPS are constructed.

• Module 5 – EPS products and usage:

– Know how to access and use EPS products.

• Module 6 – EPS application:– Know how to apply EPS in the forecast process.

Module 1: Introduction – Page 27R. Verret –

Training proposed schedule8:15 – 8:45 Module 1 - Introduction

8:45 – 10:00 Module 2 – Probabilistic Forecasts

10:00 – 10:30 Break

10:30 - 11:30 Module 3 – EPS Basic Concepts

11:30 – 12:30 Lunch

12:30 – 13:30 Module 4 – EPS Construction

13:30 – 13:45 Break

13:45 – 14:45 Module 5 – EPS Products

14:45 - 15:00 Break

15:00 – 16:00 Module 6 – Application, Web Sites

16:00 - 16:15 Module 7 – Future / Conclusion

Module 1: Introduction – Page 35R. Verret –

Conclusions

Uncertainty is the only certainty.

Module 1: Introduction – Page 36R. Verret –

Questions?