Abstract
We test the potential of two-dimensional hydraulic modeling to reproduce fish habitat in rivers analyzed at the mesoscale, i. e. the spatial scale of geomorphic units that has been shown by ecological research to represent the relevant spatial scale of fish habitat use in stream. We compare the outputs of “supervised” (classification based) and “unsupervised” (clustering) techniques to predict the space-time variability of mesohabitats (e. g. pools, riffles, glides) under non-wadable conditions, when the standard mesohabitat survey becomes mostly unfeasible. While showing potential, existing approaches based on hydraulic modelling lack universality (i. e. are riverspecific, requiring calibration and profound expert-knowledge), and have not yet been implemented and tested for complete integration within mesoscale habitat models. We explore the sensitivity of (semi-) automatic, hydraulic modelling–based mesohabitat prediction approaches to the choice of specific clustering methods referring to a wandering/braided reach of the Mareit/Mareta river in South Tyrol (NE Italy), that was recently restored to a braided morphology. We run the hydraulic simulations using a high resolution bathymetry surveyed by airborne bathymetric LIDAR, and compare the outcomes with those of ground mesohabitat surveys collected at low flow (discharge Q = 1.7 m3/s). Hydraulic model calibration is based on the spatial mapping of the hydro-morphological-units and of their measured depth and velocity distributions. Model application at non-wadable flow conditions reveal the potential of the proposed approach for a broader application of the mesohabitat concept, allowing to extend the range of validity of habitat-streamflow rating curves.