A source tracking framework for attribution of reservoir fluxes in multi-reservoir systems

Abstract

Reservoir water balance models are required to analyze the impact of alternative designs and operations on the performance of water infrastructure projects spanning multiple basins. However, the impact of such projects on instream ecology and water quality of participating basins remains relatively less explored quantitatively from a design perspective. This is due to the requirement of the complex systems model to simulate processes, which require additional data for parameterization while also increasing simulation time, which affects the performance of optimization algorithms. Here, we propose a conceptual framework to quantify the contribution of water from different basins to various reservoir-related fluxes in a multi-reservoir system. These include water released for demand satisfaction, maintaining minimum environmental flows, and preventing dam failure. For a given reservoir, we quantify the time-varying proportions of water supplied from other reservoirs and from the reservoir’s inflows for each release. We apply this framework to a proposed water transfer project in southern India that caters to the diminishing water supply of the Nagarjuna Sagar (NS) reservoir in the Krishna river basin by transferring water from the Godavari river basin. Our results show that the transferred water majorly caters to the additional demands on the NS reservoir to transfer water further south. We also find that, in general, the transferred water has little to no contribution to satisfying environmental flows downstream of the recipient. However, we do note a very high (up to 50%) contribution of transferred water to environmental flows at the NS reservoir for a few selected realizations of inflows in the monsoon season. Based on prior literature, these releases are concerning given the substantial differences in the ecological composition of fish species in both basins and may lead to invasive species downstream of the NS reservoir. We show how the proposed framework can be used to attribute recipient reservoir releases to donor and recipient inflows and shed light on failure dynamics by identifying fluxes that contribute to failures.