Premium
Validation of the Flood‐PROOFS probabilistic forecasting system
Author(s) -
Laiolo P.,
Gabellani S.,
Rebora N.,
Rudari R.,
Ferraris L.,
Ratto S.,
Stevenin H.,
Cauduro M.
Publication year - 2013
Publication title -
hydrological processes
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.222
H-Index - 161
eISSN - 1099-1085
pISSN - 0885-6087
DOI - 10.1002/hyp.9888
Subject(s) - hydrometeorology , probabilistic logic , flood forecasting , computer science , flood myth , probabilistic forecasting , operations research , statistical model , meteorology , environmental science , machine learning , engineering , artificial intelligence , geography , precipitation , archaeology
Probabilistic hydrometeorological forecasting systems are becoming more and more an operational tool used by civil protection centres for issuing flood alerts. One of the most important requests of decision makers is related to the reliability of such systems and to the validation of their predictive performances. For these reasons, this work is devoted to the validation of a probabilistic flood forecasting system called Flood‐PRObabilistic Operational Forecasting System (Flood‐PROOFS). The system is operational in real time, since 2008, in Valle d'Aosta, an alpine Region of northern Italy. It is used by the Civil Protection regional service to issue warnings and by the local water company to protect its facilities. The system manages and uses both real‐time meteorological and satellite data and real‐time data on the operation of the control structures in dam and river, managed by the water company. It has proven a useful tool for flood forecasting and for managing complex situations, facilitating the dialogue between civil protection and the water company during crisis periods. The system uses both a limited area model forecast and a forecast issued by regional expert meteorologists. The main outputs are deterministic and probabilistic discharge forecasts in different outlet areas of the river network. The performance of the system has been evaluated on a 25 months period with different statistical methods such as Brier score and Rank histograms. The results highlight good performances of the system as support system for emitting warnings, but there is a lack of statistics especially for huge discharge events. Copyright © 2013 John Wiley & Sons, Ltd.