Future developments in surge forecast: probabilistic forecast and future surge statistic

This research investigated how the surge forecast can be improved, moving from a single-deterministic to a probabilistic forecast. Moreover a future storm surge scenario is estimated using new meteorological data: sea level (SL) forecast for the city of Venice and future changes of storm surge regime due to climate changes are of paramount importance for the management and maintenance of this historical city and for operating the movable barriers that are presently being built for its protection. An Ensemble Prediction System (EPS) for operational forecasting of storm surge in the northern Adriatic Sea is presented. EPS is meant to complement the existing SL forecast system by providing a probabilistic forecast and information on uncertainty of SL prediction. Ten relatively high storm surge events in the period 2009-2010 are simulated producing for each of them an ensemble of 50 simulations, using the meteorological data input of the European Centre for Medium-Range Weather Forecasts (ECMWF) as input to a shallow water hydrodynamic model “Hydrostatic Padua Surface Elevation Model” (HYPSE), which computes sea level and barotropic currents in the Adriatic Sea . It is shown that EPS slightly increases the accuracy of SL prediction with respect to the deterministic forecast (DF) and it is more reliable than it. It is shown that the SL peaks correspond to maxima of uncertainty ( as described by the spread of the EPS members) and the values of these maxima increase linearly with the forecast range. Uncertainty on sea level is caused by the uncertainty of the forcing meteorological fields and the quasi linear dynamics of the storm surges plays a minor role on its evolution except it produces a modulation of the uncertainty after the SL peak with period corresponding to that of the main Adriatic seiche. Finally, the error of the EPS mean is correlated with the EPS spread. The second part of the research focus on the future storm surge scenario, that is estimated using new high resolution data recently produced by EC-Earth, an Earth System Model based on the operational seasonal forecast system of ECMWF. The study considers an ensemble of six 5-year long simulations of the rcp45 scenario (Hazeleger et al. 2006) and compares the 2094-2098 to the 2004-2008 period. EC-Earth sea level pressure and surface wind fields are used as input to HYPSE. The results show that high resolution of wind fields are essential for producing realistic values of storm surge statistic. However, results confirm previous studies in that they show little sensitivity of storm surge levels to climate change.

Il presente lavoro è incentrato sulla previsione dell’acqua alta a Venezia, con particolare riferimento alla marea meteorologica. Un miglioramento di tale previsione si può ottenere affiancando, alla singola previsione deterministica ottenuta utilizzando campi meteo ad alta risoluzione, una previsione probabilistica generata da campi meteo di tipo ensemble. Verrà inoltre studiato l’impatto dei cambiamenti climatici sulla distribuzione statistica dei contributi meteorologici nell’alto Adriatico, facendo uso di nuovi campi meteo ad alta risoluzione per confrontare le statistiche presenti con quelle di fine secolo. Si presenta un metodo operativo di previsione della marea meteorologica basato sul metodo dell’Ensemble Prediction System (EPS), in modo da aggiungere un’informazione probabilistica alla previsione stessa. Sono stati analizzati dieci eventi significativi del 2009 e del 2010, generando per ciascuno di essi 50 differenti previsioni di marea con il modello idrodinamico HYPSE, utilizzando come input 50 differenti campi meteo ensemble forniti dl centro meteorologico europeo ECMWF. I risultati mostrano un miglioramento della previsione, ottenuta mediando le 50 corse ensemble, rispetto a quella singola ad alta risoluzione, e ad un irrobustimento della stessa tramite l’aggiunta dell’informazione probabilistica. Si evidenzia infine una correlazione tra l’errore e la varianza delle 50 differenti corse. La seconda parte della ricerca si focalizza sulla simulazione di uno scenario futuro tramite l’utilizzo dello scenario rcp45 ottenuto dal modello EC-Earth. Si analizzano sei serie ensemble lunghe 5 anni riferite al presente (2004-2008) e sei riferite al futuro (2094-2098). Per quanto riguarda le statistiche dei livelli di marea non si riscontrano tuttavia significativi cambiamenti.