When timing matters - misdesigned dam filling impacts hydropower sustainability

Abstract

Decades of sustainable dam planning efforts have focused on containing dam impacts in regime conditions, when the dam is fully filled and operational, overlooking potential disputes raised by the filling phase. Here, we argue that filling timing and operations can catalyze most of the conflicts associated to a dam’s lifetime, which can be mitigated by adaptive solutions that respond to medium-to-long term hydroclimatic fluctuations. Our retrospective analysis of the contested recent filling of Gibe III in the Omo-Turkana basin provides quantitative evidence of the benefits generated by adaptive filling strategies, attaining levels of hydropower production comparable with the historical ones while curtailing the negative impacts to downstream users. Our results can inform a more sustainable filling of the new megadam currently under construction downstream Gibe III, and are generalizable to the almost 500 planned dams worldwide in regions influenced by climate feedbacks, thus representing a significant scope to reduce the societal and environmental impacts of a large number of new hydropower reservoirs.

Policy Representation Learning for Multiobjective Reservoir Policy Design With Different Objective Dynamics

“…Additionally, NEMODPS contains original strategies to address the specific complexities of multiobjective optimization problems, which make NEMODPS the first multiobjective NE algorithm. An in‐depth explanation of NEMODPS can be found in Zaniolo, Giuliani, & Castelletti (2021), but here we discuss the main algorithmic components.…”

“…NEMODPS will be briefly introduced in Section 2.2 of the Methods. The reader is referred to Zaniolo, Giuliani, Sinclair, et al (2021) for a more detailed analysis of NEMODPS, and its benchmarking against traditional DPS in terms of performance and computational costs.…”

“…Popular NE algorithms (e.g., NEAT Stanley & Miikkulainen, 2002) are, however, strictly applicable to single-objectives problems. The here employed Neu-roEvolutionary Multi-Objective Direct Policy Search (NEMODPS), is the first NE routine specifically designed to solve MO problems in one algorithmic iteration (Zaniolo, Giuliani, Sinclair, et al, 2021). NEMODPS is here employed for the first time to jointly evolve policies with different feature representations.…”

“…In the water resources literature, few studies have tackled the issue of the optimal selection of streamflow forecast horizon for a single‐objective reservoir operated for water supply (Anghileri et al., 2016), hydropower (Hamlet et al., 2002; Block, 2011; Xu et al., 2014), or for a generic concave objective function (Zhao et al., 2014, 2019). Additionally, the breath of information sources that was demonstrated to be valuable to inform reservoir operations is by no means limited to streamflow forecasts, but includes the previous period's inflow (Gal, 1979; Maidment & Chow, 1981), available hydrological observations (Denaro et al., 2017), traditional (Hejazi & Cai, 2011) or basin‐specific (Zaniolo et al., 2018, 2019) drought indexes, measures of snow abundance (Desreumaux et al., 2014; Giuliani, Castelletti, Fedorov, & Fraternali, 2016), shifts in hydrological regimes (Turner & Galelli, 2016), teleconnection indices (Libisch‐Lehner et al., 2019), and sea surface temperature measured in appropriate locations (Giuliani et al., 2019; Zaniolo, Giuliani, & Castelletti, 2021).…”

“…The method is applicable to large and heterogeneous datasets of candidate policy inputs, from meteorological and hydrological forecasts with disparate horizons, to observational data. The framework, namely SINEPS, Selection of Information for NeuroEvolutionary Policy Search, combines automatic feature selection with NEMODPS (NeuroEvolutionary Multi-Objective Direct Policy Search, Zaniolo, Giuliani, Sinclair, et al, 2021), a policy search routine that can accommodate changes in the policy input set. SINEPS starts with a simple operating policy and a minimal policy representation and gradually includes new inputs to the policy representations while automatically adjusting the policy processing capacity.…”