Satellite multispectral data for improved floodplain roughness modelling

Riparian vegetation plays a crucial role on affecting the floodplain hydraulic roughness, which in turn significantly influences the dynamics of flood waves. This paper explores the potential accuracies of retrieving vegetation hydrodynamic parameters through satellite multispectral data. The method is focused on estimation of vegetation height (hg) and flexural rigidity (MEI) for herbaceous patterns and of plant density (M), tree height (h), stem diameter (Ds), crown base height (cbh) and crown diameter (Dc) of high-forest (hf) and coppice (cop) consociations for arboreal and shrub patterns. The method is organized in four sequential steps: (1) classification procedure of riparian corridor; (2) land cover-based Principal Component Analysis of spectral channels; (3) explorative analysis of correlation structure between principal components and biomechanical properties and (4) model identification/estimation/ validation for floodplain roughness parameterization. To capture the hydrodynamic impacts of stiff/flexible vegetation, a GIS hydrodynamic model has been coupled with a flow resistance external routine that estimates the hydraulic roughness by using simulated water stages and the remote sensing-derived hydrodynamic parameters. The procedure is tested along a 3-km reach of the Avisio river (Trentino Alto Adige, Italy) by comparing extended field surveys and a synchronous SPOT-5 multispectral image acquired on 28/08/2004. Results showed significant correlation values between spectral-derived information and hydrodynamic parameters. Predictive models provided high coefficients of determination, especially for mixed arboreal and shrub land covers. The generated structural parameter maps represent spatially explicit data layers that can be used as inputs to hydrodynamic models to analyze flow resistance effects in different submergence conditions of vegetation. The hydraulic modelling results showed that the new method is able to provide accurate hydraulic output data and to enhance the roughness estimation up to 73% with respect to a traditional look-up table approach, with higher improvements for low flow conditions and over shrub covers