Correction of the SRTMGL1 Space-Borne Dem and Flooding Dynamics On The Lower Amazon Floodplain

Abstract:

Modeling floods events across large floodplains is challenging because flows respond to dynamic hydraulic controls from several water sources, complex geomorphology, and vegetation. In the Amazon basin, the recently liberated SRTM Global 1 arc-second (SRTMGL1) remains the best topographic information in view of hydrodynamic modeling purposes. Its accuracy is however hindered by errors partly due to vegetation leading to erroneous simulations. Previous efforts to remove the vegetation signal do not account for its spatial variability or relied on upon single assumed percentage of penetration of the SRTM signal.

In this study, we propose to a systematic approach, over an Amazonian floodplain (Fig. 1), to remove the vegetation signal, which consider its heterogeneity combining estimates of vegetation height and land cover map. We improve this approach by interpolating first results with drainage network, field and altimetry data to obtain a hydrological conditioned DEM.

Averaged interferometric and vegetation bias over the forest zone were found to be -2.0 m and 7.4 m, respectively (Fig. 2). Comparing original DEM and corrected one, vertical validation against Ground Control Points shows a RMSE diminution of 64%. Floodplain inundation was simulated using the 3D model IPH-ECO model. Flood extent accuracy, controlled against ALOS-PALSAR and JERS-1 images, stresses improvements at low and high waters (Fig. 3). This study also highlights that a ground truth drainage network as unique input during the interpolation permits reaching reasonable in terms of flood extent and hydrological characteristics.