3. Extreme events: predictability


Predictability of extreme events is generally scale-dependent and limited. For effective improvement in predictive skill, it is necessary to characterize and ascertain the hydrological, meteorological and climatological controls that occur over a wide range of scales in flash floods to droughts. Given the generally large uncertainty in predicting extreme events, it is necessary to capture and convey the predictive uncertainty via both physically-based and statistical modeling. This sub-theme presents predictability studies including those from ensemble streamflow forecasting.


Chair: Paul Dirmeyer, COLA
Keynote 1: Gianpaolo Balsamo, ECMWF: “Earth Surface Modelling Advances at ECMWF and Their Connection with Extreme Events Prediction”
Improving the realism of soil, snow, vegetation and lakes parameterisations has been subject of several recent research efforts at ECMWF. These Earth surface components work effectively as energy and water storage terms with memory considerably longer than the atmosphere counterpart.
Their role regulating land-atmosphere fluxes is particularly relevant in presence of large weather and climate anomalies and it has been assessed in dedicated land reanalyses. Improved predictions associated with land representation in the Integrated Forecasting System (IFS) is detected with the aid of research and operational observing networks. Results will be presented and discussed together with the missing slow processes and future plans.
Keynote 2: Michael Ek, NCEP: “Land data assimilation systems at NCEP:  Predicting extreme hydrometeorological events”
The NCEP North American Land Data Assimilation System (NLDAS) provides support for drought interests and water resources applications over the continental U.S., while the Global Land Data Assimilation System (GLDAS) provides initial land states in a semi-coupled mode to the NCEP Climate Forecast System (CFS) for global seasonal climate prediction.  Both systems run uncoupled land models and depend on reliable land-surface model physics and appropriate land data sets for accurate hydrometeorological simulation of surface energy and water budgets, and are forced by atmospheric analyses (or reanalyses) but rely heavily on observed precipitation.  We use NLDAS and GLDAS output to examine drought and flooding events, and associated seasonal hydrometeorological predictions from CFS.
Oral 1: Ervin Zsoter, ECMWF, UK: “Discharge Modelling Experiments with the TIGGE Archive”
Oral 2: Abdul Wahid Mohamed Rasmy, The University of Tokyo, Japan: “Application of Multi-Frequency Passive Microwave Observations and Data Assimilation Strategies for Improving Numerical Weather Forecasting in the Developing Regions”