Spatial Hydrology for the Operational monitoring of French Guiana rivers

In a context of a changing climate and an increasing anthropic pressure on natural resources, it is more than ever necessary to maintain or even improve our capacity to understand and monitor inland waters. It is particularly the case in French Guyana, where, despite its relative density, the in situ monitoring network fails at providing everyday information on rivers all over the territory. The use of free and open spatial datasets and hydrological modeling have shown great skills in complementing existing monitoring networks all over the world.  

Our work illustrates the use of a hydrological model, namely the MGB, set-up for 10 major watersheds in French Guiana (including the transboundary Maroni and Oyapock River - resp. with Suriname and Brazil- and some smaller ungauged basins) fed on a daily and near-real-time basis by IMERG-RT (Integrated Multi-satellitE Retrievals for GPM - Real Time) remote sensing precipitation products within a scheduler (namely HYFAA). In ungauged basins we used model parameters regionalisation to infer model parameters. The model performed well at inferring discharges, with KGE values higher than 0.7 when compared to gages. An extended dataset of rating curve between water surface elevation from nadir altimetry and simulated discharges is extracted using a physical-based processing of radar echoes on ESA Sentinel3 A&B and Jason3/Sentinel6 missions and also the time series available on Hydroweb website (https://hydroweb.theia-land.fr/). The quality of the rating curves confirms the skill of the model even in ungauged locations and watersheds with small contributive area. 

Thanks to this set-up, discharges and water levels are estimated daily all over the territory, and routinely corrected by the use of satellite altimetry. Using statistical rainfall predictions and watersheds concentration time, the system allows short-term forecasts of the discharge. In coordination with the in situ network operator, the critical thresholds were defined and are used to trigger  flood and droughts alerts, accessible online and received by email upon registration. As the methods used in this study have largely proved to be deployable anywhere, this simple framework draws the contour of future operational early warning systems based on space observation.