Web processing services for hydrological modeling

Hydrological models simulate river flow based on meteorological inputs, land surface use and soil conditions. They are used to map flood risks, manage dams and hydropower production, allocate water for competing uses, design infrastructures and evaluate fish habitat among many other applications. Such hydrological analysis usually requires a trained hydrologist to perform the model calibration and simulation, make sure the results are credible and interpreted adequately given models’ limitations.

To streamline the work of hydrologists and grant scientists from other disciplines easier access to hydrological modeling tools, prof. Richard Arsenault and François Brissette (École de Technologie Supérieure, ÉTS) are working in collaboration with University of Waterloo (prof. James Craig and Bryan Tolson), Ouranos and CRIM to develop a WPS server dedicated to hydrological modeling. Services will include a set of lumped hydrological models, calibration algorithms, watershed delineation services, pre-calibrated model parameters for watersheds in Canada and the U.S and processes to compute streamflow indices. For casual users, this means lower data storage requirements, no need to install models locally, and access to remote computing resources. Other services will speed-up the creation of multi-model hydrological ensembles in future climate conditions, both for gauged and ungauged watersheds.

This work funded by CANARIE is based on the birdhouse architecture and the Raven hydrological framework, as well as research projects conducted at ÉTS. Started in July 2018, the development phase will end in July 2019 and be followed by a maintenance period of 15 months. The server will be hosted at Ouranos on the PAVICS platform, and the code base is entirely open-source and available on Github.