Improved hydrodynamic modelling with UAV-derived topography and bathymetry

Maria Peppa (NU), Vassilis Glenis (NU), Jon Mills (NU)


Apply for this PhD here using application studentship code FLOOD242. Please contact Caspar Hewett ( if you have any questions about the application process. 




Flooding has been categorised as the second highest priority risk in the UK after a pandemic [1]. The National Infrastructure (NI) Assessment recently recommended prioritizing the identification of impermeable surface expansion and areas at higher risk as key strategies to mitigate surface water flooding at household-levels [2]. Hydrodynamic modelling is a well-established approach to support flood risk management at a local-scale. To achieve flood outputs at such scales, detailed mapping of urban features (e.g., inlets, gardens etc.), 3D floodplain, and river geometry are essential, but are not always feasible with traditional topographic and hydrographic surveys due to high operational costs. Furthermore, simplified river shapes that lack riverbed roughness details and bathymetry can result in unreliable flood predictions [3]. With the advance of Unmanned Aerial Vehicle (UAV)-based passive and active sensors, in tandem with geospatial AI, it is now feasible to generate cm-scale digital datasets of urban fabric, and river geometry with bathymetry up to c. 2 Secchi depths. As a contribution to NI priorities, the aim of the proposed project is to introduce a step change in local-scale hydrodynamic modelling outputs through the incorporation of highly detailed, UAV-derived, geospatial datasets.



In year 1, the successful candidate will firstly focus on structuring research objectives/methodology and secondly gaining the necessary knowledge and skills in (i) UAV-based sensors (including RGB, multi/hyper-spectral, LiDAR and thermal) and associated geospatial data processing workflows, and (ii) hydrodynamic modelling. By the end of Year 1 the candidate will be able to generate UAV-based digital surface models, understand and apply modelling using the in-house software CityCAT. During Years 2-3, the candidate will exploit geospatial AI analytics to the multimodal datasets. The student will be able to derive local scale urban land cover assets (e.g., green spaces, impermeable surfaces, inlets etc.) and river geomorphology (e.g., water surface roughness, riverbed geometry, flow resistance etc.) and incorporate information into the flood models. To simultaneously map riverbed depth and inundated areas, the student will also investigate novel approaches to refraction correction. Different scenarios of rainfall duration, depth and spatial domain coverage will be considered based on past events during CityCAT experimentation.


Newcastle University