Weather Radar and raingauge data fusion for rainfall estimation: The Livorno 2017 case

A severe weather event mainly interesting the coastal area of the Livorno and Pisa provinces occurred during the night between the 9th and 10th of September 2017. The event was rather limited in terms both of size of the involved area and of temporal duration, but the exceptional amount of rainfall caused the flooding of three torrents and landslides in the urban area of Livorno, and the loss of life of 8 people and extensive damages undergone by the entire urban area. Both the regional raingauge network and the operational weather radars correctly detected such a heavy rainfall event. The quantities measured by the rain gauges recorded peak values above 230 mm in three hours, but unfortunately the distance between the measuring points did not allow the event to be perfectly reconstructed. On the contrary, the radar and satellite systems were able to follow the event development and highlighted the most intense areas, although with little accuracy for what concerns the quantitative values. This work presents a data fusion technique for the real time estimation of the time-continuous 2D cumulative rainfall fields, based on radar and raingauge data. The technique has been applied to the dataset relative to the Livorno event. The results show the benefits of this data fusion scheme in terms of better localization and monitoring of dangerous rainfall patterns, in a perspective of early warning detection.