110 likes | 199 Views
This study examines the heavy rainfall event in North-East Estonia in August 2003, focusing on the influence of model resolution on precipitation forecasts. Various resolution models were tested, revealing differences in precipitation amounts. The analysis questions the necessity of high-resolution modeling for accurate predictions in such cases.
E N D
Modeling a heavy rainfall case in North-East Estonia, August 2003. Influence of model resolution. Andres Luhamaa University of Tartu 2005
Overview of the event • Precipitation amount: in one day 1.5 times more than monthly mean. • High precipitation rates measured only in one (Estonian station)
Forecasts • EMHI – general storm warning for Gulf of Finland, nothing more. • FMI 22km forecast estimates high precipitation at the northern coast of Estonia.
Modelling experiment • Influence of model resolution • 22km, 11km, 5km, 3km • HH and NHH • Reference HIRLAM and non-hydrostatic • Different versions of STRACO cloud scheme • HIRLAM 6.1.0 - STR1 • 6.1.2 (last major updates) - STR2 • Research version for high resolutions (<5km) called straco031 - STR3
11km STR-1 simulation FMI 22km forecast
STR-2 NHH 11km simulation HH 11km simulation
STR-1 NHH 5km simulation HH 5km simulation
NHH STR-3 5km simulation HH STR-2 5km simulation NHH STR-2 5km simulation
3-km simulations NHH STR-3 NHH STR-1 NHH STR-2
Conclusions: • Nothing better than 22km HH forecast could be achieved ;( • Significant difference between NHH and HH precipitation amounts already at 11km resolution! • Parameterization tuning for HH->NHH modeling? • As difference is only in amounts, is NHH necessary at coarser resolutions?