Water Level Prediction Research Articles

Efficient river hydrodynamics modelling in realistic river systems using a Fourier neural operator-based network

Effective river management relies heavily on the accurate simulation of river flow dynamics to develop scenario-based strategies and inform decision making. Deep learning (DL) models, as alternatives to conventional simulation techniques, offer a compelling blend of precision and efficiency. This study introduces FNOCL (Fourier Neural Operator with ConvLSTM units), a novel DL neural network for modelling unsteady natural river flows. The network is adept at handling the complexities of spatial and temporal variations. FNOCL builds upon the Fourier neural operator (FNO) by incorporating convolutional long short-term memory (ConvLSTM) layers, combining robust performance with high proficiency in terms of capturing spatiotemporal patterns. Notably, FNOCL accommodates complex river cross-section profiles, thus enabling the accurate modelling of flow dynamics within intricate river morphologies. Our study applies FNOCL to two distinct river networks with varying morphology complexity levels. With reference to the HEC-RAS model, FNOCL showcases its ability to attain high accuracy in model prediction tasks and superior generalization with regard to addressing the uncertainty associated with inflow variations and cross-sectional changes. In comparative analyses with benchmark models (i.e., FNO and ConvLSTM), FNOCL performs most efficiently during the training process and obtains the most accurate water level and river flow predictions in the spatiotemporal domain. Furthermore, FNOCL exhibits a substantial speedup over HEC-RAS, notably reaching up to 50 times in a complex river network. These results highlight FNOCL as an efficient surrogate for conventional simulators, facilitating precise analyses across various scenarios. It provides efficient computations for system responses within complex river flow networks, thereby enhancing applications in river management, such as flood risk assessment, flood control, and the planning and design of hydraulic control structures.

Double decomposition with enhanced least-squares support vector machine to predict water level

ABSTRACTAs global climates undergo changes, the frequency of water-related disasters rises, leading to significant economic losses and safety hazards. During flood events, river water levels exhibit unpredictable fluctuations, introducing considerable noise that poses challenges for accurate prediction. A prediction of water level by using existing water level data makes a major contribution to forecasting flood. Enhanced least-squares support vector machine (ELSSVM) is utilized by integrating an additional extra bias error control term. In this study, least-squares support vector machine (LSSVM) and ELSSVM optimized by the genetic algorithm (GA) were chosen to be compared with the help of data decomposition methods to improve daily water level prediction accuracy. Double empirical mode decomposition (DEMD) will be integrated with LSSVM and ELSSVM. Thus, the models are named LSSVM-GA, ELSSVM-GA, empirical mode decomposition (EMD)-LSSVM-GA, EMD-ELSSVM-GA, DEMD-LSSVM-GA, and DEMD-ELSSVM-GA. The proposed models are used in forecasting the water level of Klang River in Sri Muda, Malaysia. The behavior proposed models are investigated and compared based on several performance metrics. The results demonstrated that the DEMD-ELSSVM-GA model outperformed the other models based on the performance analysis in forecasting the water level with RMSE = 0.2536 m and R2 = 0.8596 for testing data that indicate the forecasting accuracy.

Water-Level Prediction Analysis for the Three Gorges Reservoir Area Based on a Hybrid Model of LSTM and Its Variants

The Three Gorges Hydropower Station, the largest in the world, plays a pivotal role in hydroelectric power generation, flood control, navigation, and ecological conservation. The water level of the Three Gorges Reservoir has a direct impact on these aspects. Accurate prediction of the reservoir’s water level, especially in the dam area, is of utmost importance for downstream regions’ safety and economic development. This study investigates the application and performance of four distinct deep-learning models in predicting water levels. The models evaluated include the Long Short-Term Memory (LSTM), Bidirectional Long Short-Term Memory (BiLSTM), Convolutional Neural Network–Long Short-Term Memory (CNN–LSTM), and Convolutional Neural Network–Attention–Long Short-Term Memory (CNN–Attention–LSTM). The performance of these models was assessed using several metrics, namely the Coefficient of Determination (R2), Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), and Mean Absolute Percentage Error (MAPE). The findings indicate that the CNN–Attention–LSTM model outperforms the others in all metrics, achieving an R2 value of 0.9940, MAE of 0.5296, RMSE of 0.6748, and MAPE of 0.0032. Moreover, the CNN–LSTM model exhibited exceptional predictive accuracy for lower water levels. These results underscore the potential of deep-learning models in water-level forecasting, particularly highlighting the efficacy of attention mechanisms in enhancing predictive accuracy. Precise water-level predictions are instrumental in optimizing hydropower generation and providing a scientific basis for effective flood control and water resource management.

Urban storm water prediction by applying machine learning techniques and geomorphological characteristics

ABSTRACT The Storm Water Management Model (SWMM) is a reliable software program for simulating stormwater runoff in combined drainage facilities. However, it faces challenges in efficiency, especially when executing real-time forecasting. This research explores an alternative machine learning (ML) approach to predict water levels in sewer and street drainage systems utilizing SWMM data for model training. The goal is to construct a hybrid ML model for urban flooding prediction, considering hydrological conditions, geomorphologic properties, and drainage facility features. A robust training method is introduced to deliberate complex flow conditions. The suggested approach achieves favourable performance in predictive accuracy and computational efficiency. Findings include: (1) similarity between ML and SWMM flooding depth on street nodes increases from 62.3% to 96.2% with the proposed training; (2) the proposed model is about 50 times faster than SWMM in urban flood simulations; (3) reliable predictions from the ML model are demonstrated through four accuracy metrics.

Surface water quality modelling with data scarcity in semi-enclosed coastal regions encompassed distributed islands

Water quality variation in semi-enclosed urban coastal areas with different pollutant sources is a substantial issue. Pollutant entrapment has a significant impact on the lives of the local people. Surface water quality modelling often requires large datasets covering bathymetry, fluid and pollutants boundary conditions, sources and sinks. This is particularly challenging in regions with complex features and poor data availability, such as coastal water bodies featuring a large number of widely distributed islands and estuaries. In this paper, we developed a model of surface water quality and hydrodynamics for Ha Long Bay, in the North of Vietnam and includes 1969 islands, for a one-year period using a water quality dataset obtained for this study. The model utilized extracted bathymetric and geometric data from Admiralty Charts and Admiralty Tide Tables. Water quality coupled with the TELEMAC Model (WAQTEL) Biomass Module was employed to predict pollutant transport in the domain using point and diffused sources while field studies were conducted to collect data for the setting up and calibration of the water quality model. Thirty scattered sampling points were selected, and the water quality parameters were measured during two campaigns. The predicted water level and velocity values matched the local observation data well with a small error (RMSE = 0.19 and 0.16). Both NO3−-N and PO43--P were high near the shoreline and decrease gradually offshore. The maximum concentrations of NO3−-N and PO43--P reached 0.476 mg/L and 0.048 mg/L at the end of 2021 with the RMSE = 0.13 and 0.011, respectively. The levels of NO3−-N and PO43--P and their distributions showed that Ha Long Bay was eutrophic even during the COVID-19 lockdown period.

A new approach for hydrograph data interpolation and outlier removal for vector autoregressive modelling: a case study from the Odra/Oder River

This study presents a new approach for predicting water levels of the Odra/Oder river using vector autoregressive models (VAR). We use water level time series from 27 gauging stations, on which we interpolate no-data gaps using the LinAR method and detect outliers with two separate methods: the extreme values (EV) approach and the isolation forest (IFO) algorithm. Before removing potential outliers, we propose a hydrological evaluation based on multivariate data analysis. Finally, we consider three separate data scenarios, i.e. LinAR (no outlier rejection), EV, and IFO. VAR models for six prediction gauges were built in a moving window manner on the most recent 720 hourly water levels prior to each prediction. The analysis covered the time range from January 2016 to May 2022 and resulted in ≈\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\varvec{\\approx }$$\\end{document} 1,000,000 water level forecasts (3 scenarios x 6 gauges x 55,000 hourly time steps) with lead time of 72 h. The analysis of root mean squared error (RMSE) indicates that the VAR model performs well, especially for 24-hour predictions, with RMSE values ranging from 8 to 28 cm. The model was also found to have skills in predicting a rising limb of a hydrograph. Our numerical experiments showed the susceptibility of the VAR predictions to artefacts. The IFO method was found to detect outliers skilfully, which allowed to produce the most accurate VAR-based predictions.

Application of a Hybrid Model Based on Secondary Decomposition and ELM Neural Network in Water Level Prediction

Application of a Hybrid Model Based on Secondary Decomposition and ELM Neural Network in Water Level Prediction

Using the Methods of Neural Network Learning for Peak Water Level Prediction: A Case Study for the Rivers in the Dvina-Pechora Basin

Using the Methods of Neural Network Learning for Peak Water Level Prediction: A Case Study for the Rivers in the Dvina-Pechora Basin

Evaluation of the performance of SVR and ANN Models in Sea Water Level Prediction for the Izmir Coast of the Aegean Sea

Prediction of sea water level is a very important phenomenon for making future projections for flood control, human living conditions and coastal planning. However, it is not easy to estimate sea water level due to the influence of atmospheric conditions. Thus, Artificial Neural Networks (ANN) and Support Vector Regression (SVR) methods are used for the prediction of seawater level of Izmir Coast based on the time series data. Coefficient of determination (R2 ) and root mean square error (RMSE) are applied as model evaluation criteria in this study. 19 months of seawater level data in the time series were used in this study. The results show that the ANN model can predict the water level with R2 of 0.84 and 0.68 for 1st and 2nd days, respectively, while the SVR model can predict the water level with R2 of 0.83 and 0.69 for 1st and 2nd days, respectively.

Noncontact Automatic Water-Level Assessment and Prediction in an Urban Water Stream Channel of a Volcanic Island Using Deep Learning

Traditional methods for water-level measurement usually employ permanent structures, such as a scale built into the water system, which is costly and laborious and can wash away with water. This research proposes a low-cost, automatic water-level estimator that can appraise the level without disturbing water flow or affecting the environment. The estimator was developed for urban areas of a volcanic island water channel, using machine learning to evaluate images captured by a low-cost remote monitoring system. For this purpose, images from over one year were collected. For better performance, captured images were processed by converting them to a proposed color space, named HLE, composed of hue, lightness, and edge. Multiple residual neural network architectures were examined. The best-performing model was ResNeXt, which achieved a mean absolute error of 1.14 cm using squeeze and excitation and data augmentation. An explainability analysis was carried out for transparency and a visual explanation. In addition, models were developed to predict water levels. Three models successfully forecasted the subsequent water levels for 10, 60, and 120 min, with mean absolute errors of 1.76 cm, 2.09 cm, and 2.34 cm, respectively. The models could follow slow and fast transitions, leading to a potential flooding risk-assessment mechanism.

A deep-learning model for rapid spatiotemporal prediction of coastal water levels

With the increasing impact of climate change and relative sea level rise, low-lying coastal communities face growing risks from recurrent nuisance flooding and storm tides. Thus, timely and reliable predictions of coastal water levels are critical to resilience in vulnerable coastal areas. Over the past decade, there has been increasing interest in utilizing machine learning (ML) based models for emulation and prediction of coastal water levels. However, flood advisory systems still rely on running computationally demanding hydrodynamic models. To alleviate the computational burden, these physics-based models are either run at small scales with high resolution or at large scales with low resolution. While ML-based models are very fast, they face challenges in terms of ensuring reliability and ability to capture any surge levels. In this paper, we develop a deep neural network for spatiotemporal prediction of water levels in coastal areas of the Chesapeake Bay in the U.S. Our model relies on data from numerical weather prediction models as the atmospheric input and astronomical tide levels, while its outputs are time series of predicted water levels at several tide gauge locations across the Chesapeake Bay. We utilized a CNN-LSTM setting as the architecture of the model. The CNN part extracts the features from a sequence of gridded wind fields and fuses its output to several independent LSTM units. The LSTM units concatenate the atmospheric features with respective astronomical tide levels and produce water level time series. The novel contribution of the present work is in spatiotemporality and in prioritization of the physical relationships in the model to maintain a high analogy to hydrodynamic modeling, either in the network architecture or in the selection of predictors and predictands. The results show that this setting yields a strong performance in predicting coastal water levels that cause flooding from minor to major levels. We also show that the model stands up successfully to the rigorous comparison with a high-fidelity ADCIRC model, yielding mean RMSE and correlation coefficient of 14.3 cm and 0.94, respectively, in two extreme cases, versus 12.30 cm and 0.96 for the ADCIRC model. The results highlight the practical feasibility of employing fast yet inexpensive data-driven models for resilient coastal management.

Rolling predictive control of tandem multi-canal pools based on water level elasticity intervals: A case study of the South-North water diversion middle route project

Study regionThe Middle Route of the South-North Water Transfer Project (SNWDMRP) is a major water transfer project to optimize the spatial allocation of water resources in China. Study focusIt is difficult for traditional flow adjustment to ensure the safe operation of open-channel water transfer projects. In this study, the Long Short Term Memory (LSTM) water level prediction model is combined with the elastic interval water level control method to achieve rolling prediction control of the water level in front of the gates of a single pool, and then the model is coupled with storage compensation algorithm for prediction control of the water level of the gate group of tandem multi-canal pools. The method is applied to the SNWDMRP, and the results show that the new method is stable within the restricted water level for both single and multiple canal pools. Moreover, the control algorithm proposed in this study can make full use of the storage capacity to satisfy the need of multiple scheduling scenarios

Abnormally large radiational S2 tides in the Gulf of Tonkin and implications for other diurnal resonant seas

Radiational tides derived from solar radiation are important components of ocean tides. Understanding dynamical features of radiational tides can be beneficial for further promoting the accuracy of water level prediction. Generally, radiational S2 amplitudes are only 10%–30% of gravitational S2 amplitudes, which has been verified by numerous studies. However, a previous study indicated that radidational S2 amplitudes can reach 60% of gravitational S2 amplitudes in the Gulf of Tonkin and related physical processes are unknown. In this paper, we revisit this abnormal phenomenon and find that such large radiational S2 amplitudes are not real, but artifacts induced by the inadaptation of the method based on the credo of smoothness. Specifically, observed semi-diurnal tidal admittances in the Gulf of Tonkin are not smooth due to strong nonlinear interactions between large main diurnal tides. We quantitatively estimate and remove the contributions of quadratic and quartic nonlinear interactions to main semi-diurnal tides and find that the real ratio of radiational S2 amplitude to astronomical S2 amplitude is only 0.17 at Beihai, which is basically consistent with previous model results. Our study highlights the importance of eliminating the interference of nonlinear interactions to semi-diurnal tidal admittances in all diurnal resonant sea areas like the Gulf of Tonkin and the Gulf of Thailand when extracting radiational S2 tides from sea level observations.

Superior performance of hybrid model in ungauged basins for real-time hourly water level forecasting – A case study on the Lancang-Mekong mainstream

In the context of climate change and human influence, timely and reliable information about water level variations is crucial for downstream flood control, navigation, and water resource management, particularly for the Lancang-Mekong River, the largest transboundary river in Southeast Asia. However, accurate real-time water level forecasts remain challenging, especially in data-limited regions. This study focused on the Chiang Saen Station, the uppermost Mekong River hydrological station, comparing three methods to predict 48-hour water levels: the Variable Infiltration Capacity (VIC) and Hydrology-Hydraulic (HH) physics-based models, the Gated Recurrent Unit (GRU) model, and hybrid models (VIC-GRU and HH-GRU). Evaluations were conducted at the 1-h, 3-h, 6-h, 12-h, 24-h, and 48-h forecast lead times. Assessments at various lead times showed that hybrid models incorporating physics-based mechanisms significantly enhanced predictions. VIC-GRU demonstrated superior performance, with KGE values surpassing 0.94, NSE values exceeding 0.95, MAE below 0.17 m, and PBIAS within ± 1 % across all lead times. Compared to GRU, PBIAS decreased by 84 %, while MAE dropped by 76 % to 81 % relative to VIC and HH, respectively. Notably, seasonal variations affected hybrid model performance, especially in the dry season. Optimal parameter selections enhanced model accuracy by 48 % to 49 %. These results underscore the potential of combining physics-based and deep learning models for more accurate, high-temporal-resolution real-time water level forecasts in data-scarce regions. A comprehensive study of streamflow characteristics enhances the advantages of this integration. This research contributes valuable insights to water level prediction in data-scarce regions and informs flood risk reduction and management in the Lancang-Mekong mainstream.

Analysing the performance of the NARX model for forecasting the water level in the Chikugo River estuary, Japan

Estuaries are dynamic environments which are driven by various natural processes like river discharge, tides, waves, influx of saline water and sediments, etc. These ecosystems are the most sensitive to sea level rise and fluctuations in river discharge associated with climate change. A direct response of sea level rise and river discharge can be observed in the water level of estuaries. However, existing models have not considered these parameters for forecasting water level. This paper focuses on developing a water level forecast model for the Chikugo River estuary in Japan using Nonlinear Autoregressive with Exogenous inputs (NARX Model). NARX neural network was used to do the one-step-ahead prediction of water level considering the various parameters that can very well be influenced by climate change: previous water level, river discharge, and salinity. Accordingly, three models were developed: (i) Model I considering previous water level; (ii) Model II additionally considering river discharge; and (iii) Model III additionally considering salinity. All the models showed appreciable performance in forecasting the water level. Model III had the best correlation with the water level with a cross-correlation value of 0.6030, while the river discharge had only a cross-correlation of 0.1113 indicating that the Chikugo River estuary is tide-dominated. The model was trained using different combinations of available data - previous water level, river discharge, and salinity. Cross-correlation results showed a better correlation between water level and salinity than various other combinations trained. Therefore, tidal intrusion influences the water level in the estuary, thereby depicting that sea level rise can affect the water level, and its influence can be well predicted by the developed model. The water level significantly affects the flora and fauna and the predictability of future estuarine floods can help in taking necessary mitigation strategies.

A Transfer Learning Approach Based on Radar Rainfall for River Water-Level Prediction

River water-level prediction is crucial for mitigating flood damage caused by torrential rainfall. In this paper, we attempt to predict river water levels using a deep learning model based on radar rainfall data instead of data from upstream hydrological stations. A prediction model incorporating a two-dimensional convolutional neural network (2D-CNN) and long short-term memory (LSTM) is constructed to exploit geographical and temporal features of radar rainfall data, and a transfer learning method using a newly defined flow–distance matrix is presented. The results of our evaluation of the Oyodo River basin in Japan show that the presented transfer learning model using radar rainfall instead of upstream measurements has a good prediction accuracy in the case of torrential rain, with a Nash–Sutcliffe efficiency (NSE) value of 0.86 and a Kling–Gupta efficiency (KGE) of 0.83 for 6-h-ahead forecast for the top-four peak water-level height cases, which is comparable to the conventional model using upstream measurements (NSE = 0.84 and KGE = 0.83). It is also confirmed that the transfer learning model maintains its performance even when the amount of training data for the prediction site is reduced; values of NSE = 0.82 and KGE = 0.82 were achieved when reducing the training torrential-rain-period data from 12 to 3 periods (with 105 periods of data from other rivers for transfer learning). The results demonstrate that radar rainfall data and a few torrential rain measurements at the prediction location potentially enable us to predict river water levels even if hydrological stations have not been installed at the prediction location.

Deep learning for Multi-horizon Water levelForecasting in KRS reservoir, India

In recent times, the densely populated Bengaluru metropolis in India has faced challenges related to water scarcity, particularly relying on the Krishna Raja Sagara (KRS) dam. The forecasting of reservoir water levels has become challenging due to spatio-temporal fluctuations in meteorological conditions and complex physical processes. As a result, developing suitable water management to meet the population's water demand requires an accurate and dependable estimate of the dam's water level. This work attempted to use a daily reservoir and weather data by utilizing long short-term memory (LSTM) networks. Seven high-performance models (viz., M1 to M7) with varying window sizes and horizons have been trained on this data and their performance is compared. The performance metrics revealed that the M7 model outperformed the other models for reservoir water level prediction, with coefficient of determination (R2) score of 0.93, root mean square error (RMSE) of 2.94, and mean absolute percentage error (MAPE) of 0.01. This study also provides a valuable dashboard for tracking forecasted water levels in the largest reservoir in the Cauvery basin. Finally, the innovative integration of LSTM technology in water level predictions for the Krishna Raja Sagara (KRS) dam not only addresses the challenges posed by spatio-temporal fluctuations but also sets a new standard for precision in forecasting, thereby establishing a crucial decision-support tool for real-time monitoring and enhanced water resource management in India's Bengaluru metropolis.

Prediction of the water level at the Kien Giang River based on regression techniques

AbstractModel accuracy and runtime are two key issues for flood warnings in rivers. Traditional hydrodynamic models, which have a rigorous physical mechanism for flood routine, have been widely adopted for water level prediction in river, lake, and urban areas. However, these models require various types of data, in‐depth domain knowledge, experience with modeling, and intensive computational time, which hinders short‐term or real‐time prediction. In this paper, we propose a new framework based on machine learning methods to alleviate the aforementioned limitation. We develop a wide range of machine learning models such as linear regression (LR), support vector regression (SVR), random forest regression (RFR), multilayer perceptron regression (MLPR), and light gradient boosting machine regression (LGBMR) to predict the hourly water level at Le Thuy and Kien Giang stations of the Kien Giang river based on collected data of 2010, 2012, and 2020. Four evaluation metrics, that is, R2, Nash–Sutcliffe efficiency, mean absolute error, and root mean square error, are employed to examine the reliability of the proposed models. The results show that the LR model outperforms the SVR, RFR, MLPR, and LGBMR models.

Forecasting environmental water availability of lakes using temporal fusion transformer: case studies of China's two largest freshwater lakes.

Preserving lacustrine ecosystems is vital for sustainable watershed development, and forecasting the environmental water availability of lakes would support policymakers in developing sound management strategies. This study proposed a methodology that merges the lake water level prediction and environmental water availability evaluation. The temporal fusion transformer (TFT) model forecasted the lake water levels for the next 7 days by inputting the streamflow and lake water level data for the past 30 days. The environmental water availability was assessed by comparing the forecasted lake water levels with the environmental water requirements, resulting in adequate, regular, scarce, and severely scarce environmental water availability. The methodology was tested in two case studies: Poyang Lake and Dongting Lake, the two largest freshwater lakes in the Yangtze River Basin, China. The TFT model performed well in forecasting the lake water levels, as shown by the high coefficient of determination and finite root mean square error. The coefficients of determination exceeded 0.98 during the model training, validation, and test for both Poyang Lake and Dongting Lake, and the root mean square errors ranged from 0.06 to 0.46 m. The accurate prediction of lake water level promoted the precise forecasting of the environmental water availability with the high Kappa coefficient exceeding 0.90. Results indicated the rationality and effectiveness of integrating the lake water level prediction and environmental water availability evaluation. Future research includes the applicability of the TFT model to other lakes worldwide to test the proposed approach and investigate strategies to cope with environmental water scarcity.

Water level prediction model based on blockchain and LSTM

Aiming at the shortcomings of traditional water level prediction methods such as insufficient information mining ability and unclear mechanism of heuristic algorithms, this paper proposes for the first time a water level prediction method based on blockchain technology fused with long short-term memory (LSTM) network. The method utilizes blockchain and LSTM neural network to build a combined model, and directly uploads monitoring data such as import and export water flow and water level to predict the water level, which avoids the secondary error brought by the indirect calculation of flow. In this paper, the flow compensation strategy is proposed for the first time, and the monitoring data with large deviations are compensated accordingly to reduce the prediction error from the source. The results show that the combined Blockchain-LSTM model has the smallest prediction error after adopting the compensation strategy, with the MAE of 0.290 and the RMSE of 0.490, which are smaller than those of other models, and has high prediction accuracy and practicability, which provides technical support for real-time scheduling of the South-to-North Water Diversion Reservoir.