Coupling in-situ, remote sensing and modelling tools to enhance water resources assessment over large-scale watersheds: the case of the Madeira Basin (Brazil)

Process based hydrological models have being widely used to simulate hydrology over large-scale basins, which are usually under gauged in terms of input data. The Madeira Basin is one of the major tributaries of the Amazon River, forms the south-western boundaries of the Amazon watershed and possesses a large alluvial floodplain throughout the stream continuum. The aim of the study is to propose a transposable methodology that couples in-situ, remote sensing and modelling tools to better assess the hydrological functioning of large-scale watershed; more particularly over their alluvial floodplains. We use the Soil and Water Assessment Tool (SWAT), which is a hydro-agro-environmental, semi-distributed, physically and climatically, continuous-time and processed-based model, to simulate water discharge at a daily time step and water resources for each hydrological compartments. We modify the model to integrate the floodplain and its hydrology. We use several different remote sensing products in order to calibrate both the discharge and the water balance in the different hydrological compartments: spatial altimetry (Jason 2 / Sentinel 3) for water height in streams and floodplains, L-band passive microwaves (SMOS) for floodplain delineation and soil water fraction, MODIS for ETP and sediment load, and more. We supply our methodology with the in-situ gauging stations of the HyBAm observatory over the Basin. First results show that a coupling methodology is paramount over under-gauged Basin. This methodology also allows reverse engineering where models can provide cross validation data or can complete chronicles for remote sensing and in-situ observations. As a perspective, we are investigating the coupling of remote sensing observation and modelling for suspended sediment dynamic in streams at large scale.