Delft FEWS: A proven infrastructure to bring data, sensors and models together

Abstract

Delft FEWS is a proven real time software infrastructure for operational water management and forecasting, having been applied in several operational forecasting centres all over the world. The system has contributed to changing the paradigm of flood forecasting systems from a model centric approach to a data centric approach. The system is a time series oriented ETL-infrastructure (Extract, Transfer, Load) with the ultimate aim to provide the forecaster with relevant information in support of the operational process of flood forecasting and warning. Key to its success is the design choice for separation of data communication protocols from the content, as well as its openness to integration of other applications and data. While keeping its focus on its core tasks, the design concept has allowed the system to evolve from a stand alone application, through a client-server concept into a (web) service oriented architecture. Continued research is ongoing to explore new technologies and turn these into operational features available to operational forecasters. This paper will highlight some of our experiences in system design to combine real time data with models.