Large scale inundation simulations integrating floodplain terrain and 2D hydraulic modelling with remote sensing and crowdsourced data

Large-scale flood models are affected by significant uncertainties for the need of defining several input hydrologic and hydraulic parameters and the lack of calibration and validation data. Remote sensing (Earth Observation or EO) and crowdsourced data (CD) provide unique opportunities for improving the accuracy and performance of inundation models, in both gauged and ungauged basins. This research investigates on the implementation of a floodplain Digital Terrain Model (DTMs) processing technique and a Quasi-2D hydraulic model for developing a large scale inundation simulation characterized by improved performances thanks to the use of EO and citizen driven data. A Data Assimilation framework is introduced, incorporating a geomorphic floodplain terrain modelling tool, together with a coarse resolution (hydraulic model topographic DTM interpolated from 100 to 500 m grid size) Quasi-2D flood routing model, for optimizing the information and manage the errors of the diverse available data sources (both standard river flow gauges as well as novel EO and CD data). The downstream 120 km river reach of the Tiber river basin (Italy), is selected as study basin and both real and synthetic events are tested for evaluating the performances of the presented approach in providing consistent simulations of the flood wave propagation dynamics and extent.