Bringing forecast-based real-time reservoir operation into practice using interactive Jupyter Notebooks

Abstract:

Improved skill of hydroclimatic forecasting systems has motivated a growing research effort towards demonstrating the use of forecasts in the operation of water resources. However, several barriers limit the translation of research innovations into operational practice. Among these barriers, we focus here on the disconnect between ‘research’ software developed in academia and ‘professional’ proprietary software used in the water industry to simulate water systems and support decision-making. To contribute to close this gap and accelerate the transfer of knowledge for innovation, we present iRONS, a reservoir operation toolbox that incorporate the following aspects: 1) generation of inflow forecasts from numerical weather forecasts, 2) simulation and multi-objective optimization of the operation a reservoir system, and 3) interactive visualization of the system performance, including trade-offs between different purposes and uncertainty of forecasted demand and inflows. iRONS is structured in a modular way and combines Python core functions and interactive Jupyter Notebooks that explain all the above aspects through practical examples. The use of literate programming approaches, such as Jupyter Notebooks, allows for the creation of models in single user-friendly documents where easily readable and executable code is combined with text, graphs, videos and interactive figures. This has the potential not only to facilitate the traditional academic process of model development, but also to increase research transparency and to facilitate the communication of model outputs to the professional community. We present preliminary results of a qualitative evaluation of the efficacy of iRONS to communicate complex concepts and facilitate knowledge transfer to students and practitioners.