Title: Tools for running limited area ensemble meteorological forecast in a
GRID and HPC environment

Authors: Marino Marrocu, Gabriella Pusceddu, Elisaveta Peneva

At present there is an ever increasing demand for reliable probabilistic short
range weather forecast. At regional scale these kinds of forecasts are even more valuable 
especially in cases of high impact weather such as violent extra tropical storms which 
may result in loss of lives and properties due to wide-spread flooding and gale force 
winds.  The Multi-Analysis Multi-Model  (MAMM) approach is a relatively new method of 
Ensemble Prediction System (EPS)  in which the single deterministic forecasts are 
combined using specific statistical techniques to estimate the probability  of a certain 
event to develop. MAMM simulations are very demanding in terms of computer resources, 
requiring CPU time and storage capacity which a single organization is hardly to provide. 
Moreover operational short range weather prediction at regional scale is a time critical 
task wich has to be completed in less than a couple of hours to be really usable.
In this presentation we will discuss the results of two project (GRIDA3: grida3.crs4.it, 
CYBERSAR: www.cybersar.com) financed by the Italian Ministry of Research (MIUR) in which 
we have implemented a MAMM EPS based on 3 limited area models (BOLAM, MM5 and WRF) fed 
with two different sets of Initial and Boundary Conditions (NCEP and ECMWF).
The system has been ported on a PC cluster and made accessible by a specifically designed
web portal (grida3.crs4.it/enginframe/premiagrid). This allows an authorized user to
use the remote HPC resources behind the web portal, and to run the service (named
PREMIAGRID) on demand setting only three parameter: the place in the world, the initial date
and the time of integration.  Results obtained with the porting of the PREMIAGRID
service in the virtual grid  of the CyberSAR project using the gLite middleware, will be also
discussed.
                                                                               
2)
Title: Simulations of wind and precipitation variability in Sardinia using
REGCM3
Authors: Elisaveta Peneva, Marino Marroccu, Gabriella Pusceddu

The seasonal and interannual variability of atmospheric circulation and
precipitation is very important to study in this part of the world as the
water is the most significant climate resource for households and industry.
The climate in the island is entirely influenced by the seasonal development
of the Mediterranean Low pressure system, thus the precipitations occur mainly in
the cold part of the year.  The wind is also of great interest in view of the
future plans to increase the share of renewable energy sources in this region.
Taking into account the relatively small area of Sardinia (approximately
300x100 km) and the landscape diversity a model with finer resolution than the
global circulation models is needed to account for the orographic effects.
Long-term variability of precipitation is analyzed using CRU analysis of daily
precipitation in Europe with resolution of 0.25 degree in latitudinal and
longitudinal direction. The representative years for "dry" and "wet" climate
conditions are identified and the regional climate model RegCM3 is run for
those years. The model horizontal resolution is 20 km and the initial and
boundary conditions are taken from NCEP 6-hourly reanalysis. The results are
verified with the CRU analysis data. Conclusions regarding the adequacy of the
results and the model skill to represent the regional  climate characteristics
are given. The relation between circulation and precipitation during the
simulated years is discussed.