83rd Annual

Monday, 10 February 2003
Internet technologies for enabling flood emergency early warning systems: the Arno river basin case
Paolo Mazzetti, University of Florence, Florence, Italy; and S. Nativi, L. Bigagli, and M. Mancini
Poster PDF (1.0 MB)
A flood emergency early warning system, implemented using common Internet technology is presented. The provision of early warning requires the integration of measured and forecasted rainfall rate over the river basin area according to a reliable hydrological model, and the delivery of the result to operators connected to the system through several communication infrastructures. Operators connected to the emergency management Intranet need also an effective visualization tool to inspect current and foreseen situation. Real-time data collection and fusion, network data transfer, and result presentation issues are addressed resorting to open Internet technologies. The adoption of Java platform on both processing servers and graphical clients allows easy multi-platform deployment and code delivery. XML/DOM technology allows the transport of structured information about status over common HTTP transfer protocol. According to a first experimentation, the system has proven to be reliable and scalable in operational setting.

The architecture described has been developed to implement a system for flood emergency early warning regarding an important Italian basin: the Arno river basin. The Arno river runs through Florence and its basin covers the northern part of the Tuscany region. The following main requirements were addressed: * To interface existing legacy systems which generate measured and forecasted rainfall datasets; * To feed, with the rainfall datasets, a server running an hydrological model in order to generate warning situations at the main sub-basins level. * To work out nearly real-time processing; * To provide the infrastructure to distribute results in Internet/Intranet environment; * To develop a specific web-based graphical client for decision makers; * To achieve a scalable architecture; * To allow warning management extensions.

We developed the following components: * the Data Server: to wrap and pre-process rainfall datasets resources; * the Hydro Model Server: to implement and run the hydrological model; * the Web-Based Graphical Client suited for decision makers.

--The Hydrological Model

The Department of Hydraulics of the Politecnico di Milano has designed an operative methodology to evaluate the flood warning rainfall threshold on a basin scale.

--The Measured Data System

A data collection system is available for the Tuscany raingauge network. It provides data readings of all network sensors and to collect them in a legacy format file. Update time is 30 minutes, but it can decrease to 15 minutes if a critical event is in progress.

--The Forecasted Data System

The Tuscany Regional Laboratory for Applied Meteorology provides rainfall forecasting maps for the Tuscany Region, resorting to RAMS Local Area Model. Spatial resolution is 4x4 km and the system works out one map per hour within 24 hours starting from the current time.

--Technology view

All developed components (data server, hydro server and client/GUI) are entirely written in Sun Java 2 language which allows an easy multi-platform deployment. Java WebStart technology was used as reliable code delivery technology -Internet/Intranet. Database access is performed with JDBC/ODBC technology. Dataset encoding and transfer is performed with XML over HTTP.

The described solution can be easily customized to address other similar problems of real-time data retrieval, integration, processing and delivering.

Supplementary URL: http://www.pin.unifi.it/