Multi-step-ahead water level forecasting for operating sluice gates in Hai Duong, Vietnam

Environmental Monitoring and Assessment, Год журнала: 2022, Номер 194(6)

Опубликована: Май 21, 2022

Язык: Английский

---

Smart Climate Hydropower Tool: A Machine-Learning Seasonal Forecasting Climate Service to Support Cost–Benefit Analysis of Reservoir Management

Atmosphere, Год журнала: 2020, Номер 11(12), С. 1305 - 1305

Опубликована: Дек. 1, 2020

This study proposes a climate service named Smart Climate Hydropower Tool (SCHT) and designed as hybrid forecast system for supporting decision-making in context of hydropower production. SCHT is technically to make use information from state-of-art seasonal forecasts provided by the Copernicus Data Store (CDS) combined with range different machine learning algorithms perform accumulated inflow discharges reservoir plants. The considered include support vector regression, Gaussian processes, long short-term memory, non-linear autoregressive neural networks exogenous inputs, deep-learning model. Each model trained over past decades datasets recorded data, performances are validated evaluated using separate test sets reference historical average discharge values simpler multiparametric regressions. Final results presented users through user-friendly web interface developed tied connection end-users an effective co-design process. Methods tested forecasting river up six months advance two catchments Colombia, South America. Results indicate that complex and/or recurrent architecture can better simulate temporal dynamic behaviour both case reservoirs, thus rendering useful tool providing water resource managers planning allocation resources plant when negotiating power purchase contracts competitive energy markets.

Язык: Английский

---

An optimized long short-term memory (LSTM)-based approach applied to early warning and forecasting of ponding in the urban drainage system

Hydrology and earth system sciences, Год журнала: 2023, Номер 27(10), С. 2035 - 2050

Опубликована: Май 26, 2023

Abstract. In this study, we propose an optimized long short-term memory (LSTM)-based approach which is applied to early warning and forecasting of ponding in the urban drainage system. This can quickly identify locate with relatively high accuracy. Based on approach, a model developed, constructed by two tandem processes utilizes multi-task learning mechanism. The superiority developed was demonstrated comparing widely used neural networks (LSTM convolutional networks). Then, further revised available monitoring data study area achieve higher We also discussed how number selected points influenced performance corrected model. over 15 000 designed rainfall events were for training, covering various extreme weather conditions.

Язык: Английский

---

Fully distributed rainfall-runoff modeling using spatial-temporal graph neural network

EarthArXiv (California Digital Library), Год журнала: 2022, Номер unknown

Опубликована: Янв. 15, 2022

Recent studies using latest deep learning algorithms such as LSTM (Long Short-Term Memory) have shown great promise in time-series modeling. There are many focusing on the watershed-scale rainfall-runoff modeling or streamflow forecasting, often considering a single watershed with limited generalization capabilities. To improve model performance, several explored an integrated approach by decomposing large into multiple sub-watersheds semi-distributed structure. In this study, we propose innovative physics-informed fully-distributed model, NRM-Graph (Neural Runoff Model-Graph), Graph Neural Networks (GNN) to make full use of spatial information including flow direction and geographic data. Specifically, applied each grid cell for its runoff production. The output is then aggregated GNN final at outlet. case study shows that our based successfully represents predictions. network has less over-fitting significant improvement performance compared baselines information. Our research further confirms importance spatially distributed hydrological learning, encourage researchers incorporate more domain knowledge

Язык: Английский

---

Rain-Code Fusion: Code-to-Code ConvLSTM Forecasting Spatiotemporal Precipitation

Lecture notes in computer science, Год журнала: 2021, Номер unknown, С. 20 - 34

Опубликована: Янв. 1, 2021

Язык: Английский

---