Hydraulic Model Calibration Using Water Levels Derived from Time Series High-Resolution SAR Images

Hydraulic models require water extents (WE) and water levels (WL) for their calibration and validation. The synthetic aperture radar (SAR) has been readily used in the past to delineate water bodies and extract WL from the digital terrain model (DTM). However, studies using SAR data to calibrate hydraulic models have been carried out with a limited number of images. This study aims to use WL derived from a time series of high-resolution (5 m) Radarsat-2 SAR images to calibrate a one-dimensional model (HEC-RAS) using an automated algorithm on a 40-km reach of the Athabasca River in Alberta, Canada. Eighteen images, spanning 2012-2016, were processed to extract WL using a high-resolution (2 m) DTM that combined light detection and ranging (LiDAR) data and bathymetry acquired by boat surveying. The impact of the number of images used for the calibration has been assessed. The best root mean square error in validation between the SAR-derived and simulated WL was 0.28 m using seven images. Finally, a critical success index (CSI) was performed to compare the SAR-derived and simulated WE. No significant change in CSI was observed because of riverbank steepness.