The Role of Reservoir Reoperation to Mitigate Climate Change Impacts on Hydropower and Environmental Water Demands

Dynamic Adaptive Environmental Flows (DAE‐Flows) to Reconcile Long‐Term Ecosystem Demands With Hydropower Objectives

“…Results in Dalcin et al. (2023) and supported by previous results in Oliveira et al. (2015) and Suzuki et al.…”

“…This model is able to track fish abundance response for different flow regimes, as presented (training and validation) in Dalcin et al. (2023). $$${\boldsymbol{x}}_{i}=f\left({\boldsymbol{r}}_{y,t}\right)$$$ $$${\boldsymbol{g}}_{y}={\boldsymbol{\phi }}_{o}\left\{{\boldsymbol{\beta }}_{k}+\sum\limits _{j}{\boldsymbol{w}}_{j}{\boldsymbol{\phi }}_{h}\left({\boldsymbol{\beta }}_{j}+\sum\limits _{i}{\boldsymbol{w}}_{i,j}{\boldsymbol{x}}_{i}\right)\right\}$$$ where, variables in bold represent the vector of environmental sites, the index $$j$$ represents the hidden neurons, $$i$$ represents the input neurons, and $$k$$ represents the output neurons.…”

“…The second and third scenarios considered hydroclimatic projections spanning from 2021 to 2065 from two climate models, Eta‐MIROC5 and Eta‐HADGEM2‐ES, under RCP 4.5 and 8.5 emission scenarios, respectively. While the hydrological results conditioned to the climatic variables of Eta‐MIROC5 under the RCP 4.5 scenario presented the lowest mean flow reduction in relation to the historical period (−3%) for the study area, the Eta‐HADGEM2‐ES under the RCP 8.5 scenario showed the largest mean flow change (−59%) (Dalcin et al., 2023). These scenarios were called minor and major climate change scenarios.…”

“…The data‐driven fish‐flow model was previously trained based on the observed time‐series of daily streamflow/level and fish sampling data during the 1992 to 1994 and 2000 to 2019 campaigns (see Dalcin et al. (2023) for further details on training). The migratory fish abundance (ANN output) was represented by the sum of five long‐distance migratory fish species Brycon orbignyanus (Valenciennes 1850) , Pseudoplatystoma corruscans (Spix & Agassiz 1829) , Pterodoras granulosus (Valenciennes 1821) , Prochilodus lineatus (Valenciennes 1837) , Salminus brasiliensis (Cuvier 1816), which are among the most abundant long‐distance migratory fish in the study area (Agostinho et al., 2007; A. G. Oliveira et al., 2015).…”

“…The data-driven fish-flow model was previously trained based on the observed time-series of daily streamflow/ level and fish sampling data during the 1992 to 1994 and 2000 to 2019 campaigns (see Dalcin et al (2023) The flow regime was characterized by nine indices (ANN inputs), at least one index for each flow regime component (duration, magnitude, timing, frequency, and rate of change), chosen according to their relevance to trigger the ecosystem function of interest. For the study area, this function was the fish migration, reproduction and initial growth, which depends on floodplain lateral and longitudinal connectivity dynamics over a water year.…”