Start Time:

2 August 2010 at 09:00

Ends On:

13 August 2010

Location:

Trieste - Italy

Venue:

AGH (Kastler Lecture Hall)

Organizer(s):

Directors: J. Shukla (GMU & COLA, USA), T. Delsole (GMU & COLA, USA) Local Organizer: Jin Ho Yoo

Description:

Current seasonal predictions in the developing countries heavily rely on the statistical methods. Even though dynamical modeling becomes a major tool of seasonal prediction, still the prediction skill of dynamical system is not much better than the statistical model which needs little computational resources. The statistical forecast also has an advantage that the knowledge obtained from data analysis can be easily applied. The statistical forecast is a good topic for effective cooperation between research and operational groups, particularly in developing countries with little resources. However, the statistical approach in the seasonal prediction has some distinctive characteristics, which require careful use of conventional statistical modeling techniques. In particular, relatively short history of observed database and the existence of long-term change in the climate system make the statistical methods unusable so that a good model to explain a strong correlation in the past (dependent) data often fails to produce good forecasts in the future (independent data). This is referred to as artificial predictability. Therefore, statistical model for seasonal prediction should be designed and assessed carefully regarding the trade-off between goodness-of-fit and stability of the model. During the last TTAs, it is observed that many of national meteorological services in the developing countries do not pay enough attention to this problem and just adopt conventional statistical methods in their seasonal prediction. In this respect, we propose the specialized training activity on the statistical method as a prediction tool. This will be a good opportunity to review the current seasonal prediction activities in the developing countries and will be an immediate guide of future development of prediction system.

The 2010 TTA aims:
1. Provide better understanding of statistical properties of climate system ;
2. Provide state-of-art knowledge on the statistical methods;
3. Highlight the limitation and cautions in the use of statistical methods in the short-term climate prediction;
4. Review the current seasonal forecast methods of participating countries.

Programme structure:

1st week: Lectures and training on the statistical properties of climate system
2nd week: Lectures and training on the statistical forecasting methods and evaluation of forecast

Material:

no notes available

• Programme (PDF )
• General_Information_Sheet
• TTA_Agenda_NEW
• TTA_Attendees_OK
• TTA_Editions