Spatio-temporally Varying Manning Roughness in Rivers and Streams: A calibration approach using in-situ water level and UAS altimetry

Hydraulic roughness (expressed in terms of e.g. Manning's roughness coefficient) is an important input to hydraulic and hydrodynamic simulation models. One way to estimate roughness parameters is by hydraulic inversion, using observed water surface elevation (WSE) collected from gauging stations, satellite platforms or UAS (Unmanned Aerial System) −based altimeters. Specifically, UAS altimetry provides close to instantaneous observations of longitudinal profiles and seasonal variations of WSE for various river types, which are useful for calibrating roughness parameters. However, it is computationally expensive to run high−resolution hydrodynamic models for long simulation periods (e.g. multiple years), and thus global optimization of spatially and temporally distributed parameter sets for such models, e.g., spatio−temporally varying river roughness, is still challenging.This study presented an efficient calibration approach for hydraulic models, using a simplified steady-state hydraulic solver, UAS altimetry datasets, and in-situ observations. The calibration approach minimized the weighted sum of a misfit term, spatial smoothness penalty, and a sinusoidal a priori temporal variation constraint. The approach was first demonstrated for several synthetic calibration experiments and the results indicated that the global search algorithm accurately recovered the Manning–Strickler coefficients M for short river reaches in different seasons, and M varied significantly in time (due to the seasonal growth cycle of the aquatic vegetation) and space (due to, e.g. spatially variable vegetation density). Subsequently, the calibration approach was demonstrated for a real WSE dataset collected at a Danish test site, i.e., Vejle Å. Results indicated that spatio-temporal variation in M was required to accurately fit in-situ and UAS altimetry WSE observations. This study illustrated how UAS altimetry and hydraulic modeling can be combined to achieve improved understanding and better parameterization of small and medium-sized rivers, where conveyance is controlled by vegetation growth and other spatio-temporally variable factors.