There has been an increasing demand for accurate rainfall forecast in urban areas from the water industry. Current forecasting systems provided mainly by meteorological offices are based on large-scale prediction and are not well suited for this application. In order to devise a system especially designed for the dynamic management of a sewerage system the “RADAR” project was launched. The idea of this project was to provide a short-term small-scale prediction of rain based on radar images. The prediction methodology combines two methods. An extrapolation method based on a sophisticated cross correlation of images is optimised by a neural network technique. Three different application sites in Europe have been used to validate the system.

This content is only available as a PDF.