Multi-Output Variational Gaussian Process for Daily Forecasting of Hydrological Resources
Water resource forecasting plays a crucial role in managing hydrological reservoirs, supporting operational decisions ranging from the economy to energy. In recent years, machine learning-based models, including sequential models such as Long Short-Term Memory (LSTM) networks, have been widely emplo...
Main Authors: | , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
MDPI AG
2023-07-01
|
Series: | Engineering Proceedings |
Subjects: | |
Online Access: | https://www.mdpi.com/2673-4591/39/1/83 |
_version_ | 1797580313457590272 |
---|---|
author | Julián David Pastrana-Cortés David Augusto Cardenas-Peña Mauricio Holguín-Londoño Germán Castellanos-Dominguez Álvaro Angel Orozco-Gutiérrez |
author_facet | Julián David Pastrana-Cortés David Augusto Cardenas-Peña Mauricio Holguín-Londoño Germán Castellanos-Dominguez Álvaro Angel Orozco-Gutiérrez |
author_sort | Julián David Pastrana-Cortés |
collection | DOAJ |
description | Water resource forecasting plays a crucial role in managing hydrological reservoirs, supporting operational decisions ranging from the economy to energy. In recent years, machine learning-based models, including sequential models such as Long Short-Term Memory (LSTM) networks, have been widely employed to address this task. Despite the significant interest in forecasting hydrological series, weather’s nonlinear and stochastic nature hampers the development of accurate prediction models. This work proposes a Variational Gaussian Process-based forecasting methodology for multiple outputs, termed MOVGP, that provides a probabilistic framework to capture the prediction uncertainty. The case study focuses on the Useful Volume and the Streamflow Contributions from 23 reservoirs in Colombia. The results demonstrate that MOVGP models outperform classical LSTM and linear models in predicting several horizons, with the added advantage of offering a predictive distribution. |
first_indexed | 2024-03-10T22:48:21Z |
format | Article |
id | doaj.art-78d010b1bb514e659c9e06b6306bac94 |
institution | Directory Open Access Journal |
issn | 2673-4591 |
language | English |
last_indexed | 2024-03-10T22:48:21Z |
publishDate | 2023-07-01 |
publisher | MDPI AG |
record_format | Article |
series | Engineering Proceedings |
spelling | doaj.art-78d010b1bb514e659c9e06b6306bac942023-11-19T10:31:23ZengMDPI AGEngineering Proceedings2673-45912023-07-013918310.3390/engproc2023039083Multi-Output Variational Gaussian Process for Daily Forecasting of Hydrological ResourcesJulián David Pastrana-Cortés0David Augusto Cardenas-Peña1Mauricio Holguín-Londoño2Germán Castellanos-Dominguez3Álvaro Angel Orozco-Gutiérrez4Automatic Research Group, Universidad Tecnológica de Pereira, Pereira 660003, ColombiaAutomatic Research Group, Universidad Tecnológica de Pereira, Pereira 660003, ColombiaAutomatic Research Group, Universidad Tecnológica de Pereira, Pereira 660003, ColombiaSignal Processing and Recognition Group, Universidad Nacional de Colombia, Manizales 170003, ColombiaAutomatic Research Group, Universidad Tecnológica de Pereira, Pereira 660003, ColombiaWater resource forecasting plays a crucial role in managing hydrological reservoirs, supporting operational decisions ranging from the economy to energy. In recent years, machine learning-based models, including sequential models such as Long Short-Term Memory (LSTM) networks, have been widely employed to address this task. Despite the significant interest in forecasting hydrological series, weather’s nonlinear and stochastic nature hampers the development of accurate prediction models. This work proposes a Variational Gaussian Process-based forecasting methodology for multiple outputs, termed MOVGP, that provides a probabilistic framework to capture the prediction uncertainty. The case study focuses on the Useful Volume and the Streamflow Contributions from 23 reservoirs in Colombia. The results demonstrate that MOVGP models outperform classical LSTM and linear models in predicting several horizons, with the added advantage of offering a predictive distribution.https://www.mdpi.com/2673-4591/39/1/83streamflow contributionspredictive distributionforecastingGaussian processuseful volume |
spellingShingle | Julián David Pastrana-Cortés David Augusto Cardenas-Peña Mauricio Holguín-Londoño Germán Castellanos-Dominguez Álvaro Angel Orozco-Gutiérrez Multi-Output Variational Gaussian Process for Daily Forecasting of Hydrological Resources Engineering Proceedings streamflow contributions predictive distribution forecasting Gaussian process useful volume |
title | Multi-Output Variational Gaussian Process for Daily Forecasting of Hydrological Resources |
title_full | Multi-Output Variational Gaussian Process for Daily Forecasting of Hydrological Resources |
title_fullStr | Multi-Output Variational Gaussian Process for Daily Forecasting of Hydrological Resources |
title_full_unstemmed | Multi-Output Variational Gaussian Process for Daily Forecasting of Hydrological Resources |
title_short | Multi-Output Variational Gaussian Process for Daily Forecasting of Hydrological Resources |
title_sort | multi output variational gaussian process for daily forecasting of hydrological resources |
topic | streamflow contributions predictive distribution forecasting Gaussian process useful volume |
url | https://www.mdpi.com/2673-4591/39/1/83 |
work_keys_str_mv | AT juliandavidpastranacortes multioutputvariationalgaussianprocessfordailyforecastingofhydrologicalresources AT davidaugustocardenaspena multioutputvariationalgaussianprocessfordailyforecastingofhydrologicalresources AT mauricioholguinlondono multioutputvariationalgaussianprocessfordailyforecastingofhydrologicalresources AT germancastellanosdominguez multioutputvariationalgaussianprocessfordailyforecastingofhydrologicalresources AT alvaroangelorozcogutierrez multioutputvariationalgaussianprocessfordailyforecastingofhydrologicalresources |