PREDICTION OF MAXIMUM STREAMFLOWS OF THE FUNIL RESERVOIR (MG) VIA A BAYESIAN APPROACH

Predicting maximum streamflows is important for reservoir management. Accordingly, this study aimed to model maximum streamflows at the reservoir of the Funil Hydropower Plant (Funil HPP) using the Gumbel distribution within a Bayesian framework and to assess the influence of different prior specifications on return level estimation. Daily mean streamflow data from 2003 to 2025 were obtained from the HidroWeb system of the Brazilian National Water and Sanitation Agency, from which annual maxima were extracted separately for the wet and dry periods. Informative priors were elicited from the Camargos HPP time series. Records up to 2017 were used for model selection, with the Funil data from 2018 to 2025 held out for performance evaluation. Models were compared using the deviance information criterion (DIC), mean prediction error (MPE), and mean width of the 95% highest posterior density (HPD) intervals (MHPDW). Compared with the noninformative specification, the informative prior with the smallest variance reduced MPE from 30.93% to 26.85% in the wet period and from 27.22% to 26.31% in the dry period. This specification also reduced MHPDW for the 2, 4, 6, and 8-year return levels by 62.86% and 61.50%, respectively. In the dry period, however, the lowest MPE was obtained with the prior with twice the baseline variance, reaching 26.14%. The results indicated that the main contribution of the information from Camargos was a reduction in posterior uncertainty in the return levels, whereas the numerical reductions in MPE were more modest, particularly in the dry period.

Authors

Institutions

Publication Details

Journal
Revista Multidisciplinar do Nordeste Mineiro
Published
2026-11-17
DOI
https://doi.org/10.66104/5yj0r293
Primary Topic
Hydrology and Watershed Management Studies
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

PREDICTION OF MAXIMUM STREAMFLOWS OF THE FUNIL RESERVOIR (MG) VIA A BAYESIAN APPROACH

Luiz Alberto Beijo, Carlos José dos Reis
Revista Multidisciplinar do Nordeste Mineiro
Hydrology and Watershed Management Studies
article

PREDICTION OF MAXIMUM STREAMFLOWS OF THE FUNIL RESERVOIR (MG) VIA A BAYESIAN APPROACH

Luiz Alberto Beijo, Carlos José dos Reis
article en

Abstract

Predicting maximum streamflows is important for reservoir management. Accordingly, this study aimed to model maximum streamflows at the reservoir of the Funil Hydropower Plant (Funil HPP) using the Gumbel distribution within a Bayesian framework and to assess the influence of different prior specifications on return level estimation. Daily mean streamflow data from 2003 to 2025 were obtained from the HidroWeb system of the Brazilian National Water and Sanitation Agency, from which annual maxima were extracted separately for the wet and dry periods. Informative priors were elicited from the Camargos HPP time series. Records up to 2017 were used for model selection, with the Funil data from 2018 to 2025 held out for performance evaluation. Models were compared using the deviance information criterion (DIC), mean prediction error (MPE), and mean width of the 95% highest posterior density (HPD) intervals (MHPDW). Compared with the noninformative specification, the informative prior with the smallest variance reduced MPE from 30.93% to 26.85% in the wet period and from 27.22% to 26.31% in the dry period. This specification also reduced MHPDW for the 2, 4, 6, and 8-year return levels by 62.86% and 61.50%, respectively. In the dry period, however, the lowest MPE was obtained with the prior with twice the baseline variance, reaching 26.14%. The results indicated that the main contribution of the information from Camargos was a reduction in posterior uncertainty in the return levels, whereas the numerical reductions in MPE were more modest, particularly in the dry period.

Revista Multidisciplinar do Nordeste MineiroVol. 13(24)
Universidade Federal de Alfenas (BR)
Hydrology and Watershed Management Studies
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon