Short-Term Streamflow Forecasting Using Hybrid Deep Learning Model Based on Grey Wolf Algorithm for Hydrological Time Series

Sustainability, Journal Year: 2022, Volume and Issue: 14(6), P. 3352 - 3352

Published: March 12, 2022

The effects of developing technology and rapid population growth on the environment have been expanding gradually. Particularly, in water consumption has revealed necessity management. In this sense, accurate flow estimation is important to Therefore, study, a grey wolf algorithm (GWO)-based gated recurrent unit (GRU) hybrid model proposed for streamflow forecasting. daily data Üçtepe Tuzla observation stations located various collection areas Seyhan basin were utilized. test training analysis models, first 75% used training, remaining 25% testing. accuracy success compared via comparison linear regression, one most basic models artificial neural networks. results analyzed using different statistical indexes. Better obtained GWO-GRU benchmark all metrics except SD at station whole station. At Üçtepe, FMS, despite RMSE MAE being 82.93 85.93 m3/s, was 124.57 it 184.06 m3/s single GRU model. We achieved around 34% 53% improvements, respectively. Additionally, R2 values FMS 0.9827 0.9558 from It observed that could be successfully forecasting studies.

Language: Английский

---

A graph neural network (GNN) approach to basin-scale river network learning: the role of physics-based connectivity and data fusion

Hydrology and earth system sciences, Journal Year: 2022, Volume and Issue: 26(19), P. 5163 - 5184

Published: Oct. 14, 2022

Abstract. Rivers and river habitats around the world are under sustained pressure from human activities changing global environment. Our ability to quantify manage states in a timely manner is critical for protecting public safety natural resources. In recent years, vector-based network models have enabled modeling of large basins at increasingly fine resolutions, but computationally demanding. This work presents multistage, physics-guided, graph neural (GNN) approach basin-scale learning streamflow forecasting. During training, we train GNN model approximate outputs high-resolution model; then fine-tune pretrained with observations. We further apply graph-based, data-fusion step correct prediction biases. The GNN-based framework first demonstrated over snow-dominated watershed western United States. A series experiments performed test different training imputation strategies. Results show that trained can effectively serve as surrogate process-based high accuracy, median Kling–Gupta efficiency (KGE) greater than 0.97. Application graph-based data fusion reduces mismatch between observations, much 50 % KGE improvement some cross-validation gages. To improve scalability, graph-coarsening procedure introduced larger basin. coarsening achieves comparable skills only fraction cost, thus providing important insights into degree physical realism needed developing large-scale models.

Language: Английский

---

Short-term forecasts of streamflow in the UK based on a novel hybrid artificial intelligence algorithm

Scientific Reports, Journal Year: 2023, Volume and Issue: 13(1)

Published: April 29, 2023

In recent years, the growing impact of climate change on surface water bodies has made analysis and forecasting streamflow rates essential for proper planning management resources. This study proposes a novel ensemble (or hybrid) model, based combination Deep Learning algorithm, Nonlinear AutoRegressive network with eXogenous inputs, two Machine algorithms, Multilayer Perceptron Random Forest, short-term forecasting, considering precipitation as only exogenous input forecast horizon up to 7 days. A large regional was performed, 18 watercourses throughout United Kingdom, characterized by different catchment areas flow regimes. particular, predictions obtained Learning-Deep model were compared ones achieved simpler models an both algorithms algorithm. The hybrid outperformed models, values R2 above 0.9 several watercourses, greatest discrepancies small basins, where high non-uniform rainfall year makes rate challenging task. Furthermore, been shown be less affected reductions in performance increases leading reliable even 7-day forecasts.

Language: Английский

---

A Comparative Study on Forecasting of Long-term Daily Streamflow using ANN, ANFIS, BiLSTM and CNN-GRU-LSTM

Water Resources Management, Journal Year: 2023, Volume and Issue: 37(12), P. 4769 - 4785

Published: Aug. 22, 2023

Language: Английский

---

Monthly runoff prediction based on a coupled VMD-SSA-BiLSTM model

Scientific Reports, Journal Year: 2023, Volume and Issue: 13(1)

Published: Aug. 12, 2023

Abstract The accurate prediction of monthly runoff in the lower reaches Yellow River is crucial for rational utilization regional water resources, optimal allocation, and flood prevention. This study proposes a VMD-SSA-BiLSTM coupled model volume prediction, which combines advantages Variational Modal Decomposition (VMD) signal decomposition preprocessing, Sparrow Search Algorithm (SSA) BiLSTM parameter optimization, Bi-directional Long Short-Term Memory Neural Network (BiLSTM) exploiting bi-directional linkage advanced characteristics process. proposed was applied to predict at GaoCun hydrological station River. results demonstrate that outperforms both VMD-BiLSTM terms accuracy during training validation periods. Root-mean-square deviation 30.6601, 242.5124 39.9835 compared respectively; mean absolute percentage error 5.6832%, 35.5937% 6.3856% other two models, 19.8992, decreased by 136.7288 25.7274 square correlation coefficient ( R 2 ) 0.93775, increases 0.53059 0.14739 Nash–Sutcliffe efficiency 0.9886, increased 0.4994 0.1122 respectively. In conclusion, model, utilizing sparrow search algorithm bidirectional long short-term memory neural network, enhances smoothness series improves point predictions. holds promise effective

Language: Английский

---

Comparative Study for Daily Streamflow Simulation with Different Machine Learning Methods

Water, Journal Year: 2023, Volume and Issue: 15(6), P. 1179 - 1179

Published: March 18, 2023

Rainfall–runoff modeling has been of great importance for flood control and water resource management. However, the selection hydrological models is challenging to obtain superior simulation performance especially with rapid development machine learning techniques. Three under different categories methods, including support vector regression (SVR), extreme gradient boosting (XGBoost), long-short term memory neural network (LSTM), were assessed simulating daily runoff over a mountainous river catchment. The performances input scenarios compared. Additionally, joint multifractal spectra (JMS) method was implemented evaluate during wet dry seasons. results show that: (1) LSTM always obtained higher accuracy than XGBoost SVR; (2) impacts variables such as antecedent streamflow rainfall LSTM; (3) showed relatively high seasons, classification seasons improved performance, seasons; (4) JMS analysis indicated advantages hybrid model combined trained wet-season data dry-season data.

Language: Английский

---

A comparative study of Machine Learning and Deep Learning methods for flood forecasting in the Far-North region, Cameroon

Scientific African, Journal Year: 2023, Volume and Issue: 23, P. e02053 - e02053

Published: Dec. 27, 2023

Flood crises are the consequence of climate change and global warming, which lead to an increase in frequency intensity heavy rainfall. Floods are, remain, natural disasters that result huge loss lives material damage. risks threaten all countries globe general. The Far-North region Cameroon has suffered flood on several occasions, resulting significant human lives, infrastructural socio-economic damage, with destruction homes, crops grazing areas, halting economic activities. models used for forecasting this generally physical-based, produce unsatisfactory results. use artificial intelligence based methods order limit its consequences is a way be explored Cameroon. aims present research work design compare performance Machine Learning Deep such as one dimensional Convolutional Neural Network, Long Short Term Memory Multi Layer Perceptron short-term long-term designed take input temperature rainfall time series recorded region. Performance criteria evaluating Nash–SutcliffeEfficiency, Percent Bias, Coefficient Determination Root Mean Squared Error. As results comparison models, best model LSTM , still model. obtained from comparisons have satisfactory good generalization capabilities, reflected by criteria. our can implementation floods warning systems definition effective efficient risk management policies make more resilient crises.

Language: Английский

---

Long-term streamflow forecasting in data-scarce regions: Insightful investigation for leveraging satellite-derived data, Informer architecture, and concurrent fine-tuning transfer learning

Journal of Hydrology, Journal Year: 2024, Volume and Issue: 631, P. 130772 - 130772

Published: Feb. 2, 2024

Language: Английский

---

In-depth simulation of rainfall–runoff relationships using machine learning methods

Water Practice & Technology, Journal Year: 2024, Volume and Issue: 19(6), P. 2442 - 2459

Published: June 1, 2024

ABSTRACT Measurement inaccuracies and the absence of precise parameters value in conceptual analytical models pose challenges simulating rainfall–runoff modeling (RRM). Accurate prediction water resources, especially scarcity conditions, plays a distinctive pivotal role decision-making within resource management. The significance machine learning (MLMs) has become pronounced addressing these issues. In this context, forthcoming research endeavors to model RRM utilizing four MLMs: Support Vector Machine, Gene Expression Programming (GEP), Multilayer Perceptron, Multivariate Adaptive Regression Splines (MARS). simulation was conducted Malwathu Oya watershed, employing dataset comprising 4,765 daily observations spanning from July 18, 2005, September 30, 2018, gathered rainfall stations, Kappachichiya hydrometric station. Of all input combinations, incorporating Qt−1, Qt−2, R̄t identified as optimal configuration among considered alternatives. models' performance assessed through root mean square error (RMSE), average (MAE), coefficient determination (R2), developed discrepancy ratio (DDR). GEP emerged superior choice, with corresponding index values (RMSE, MAE, R2, DDRmax) (43.028, 9.991, 0.909, 0.736) during training process (40.561, 10.565, 0.832, 1.038) testing process.

Language: Английский

---

Real-time error correction of multiple-hour-ahead flash flood forecasting based on the sliding runoff-rain data and deep learning models

Journal of Hydrology, Journal Year: 2025, Volume and Issue: unknown, P. 132918 - 132918

Published: Feb. 1, 2025

Language: Английский

---