What is an inflow forecasting tool?
Water is one of our most precious resources and indeed a scarce resource in some parts of the world, so basic needs can be better met if better flow forecasting techniques are used. It can and should be used continuously to achieve potential profits in energy generation, for example. Beyond that, it is most useful in the presence of extreme events, which are, by definition, rare, and should provide sound information to be acted upon, both in the case of floods and droughts. The value of inflow forecasting systems is thus often considered too late, after the event has passed and damages are done. The time to the next extreme event (normally, but not always, long) allows the occasion to be forgotten.
What is different about the Tethys forecasting system?
Using a data-driven approach and machine learning models, the system is capable of producing reliable probabilistic forecasts. The use of the term “probabilistic” implies that the forecasts attempt to characterize not only future discharges or accumulated flows, but also the uncertainty that is associated with the predictions.
The image illustrates the improvement of going from deterministic to probabilistic predictions. Shaded gray bands’ values correspond to the probability of observations being found within them.

Why choose Tethys over traditional methods?
Typical forecasting methodologies are expensive to setup, complex to prepare and maintain, or simply not reliable. Tethys is a powerful and practical method of producing reliable hydrological forecasts efficiently and effectively. Tethys is non-parametric, which means that it can be used to predict not only inflows but many other variables.
It is currently being developed for the Enguri Dam and under use at Rogun Dam and the Cahora Bassa Dam, have a look at the website here.