Rainfall-runoff Model Research Articles

Hyperparameter optimization of regional hydrological LSTMs by random search: A case study from Basque Country, Spain

This paper introduces a novel approach for hyperparameter optimization of long short-term memory networks (LSTMs) to achieve highly accurate hourly streamflow and water level predictions in the realm of regional rainfall-runoff modeling. Leveraging simultaneous systematic hyperparameter optimization of 10 distinct hyperparameters by Random Search, the study achieves high accuracy in terms of predictions across 40 humid flashy catchments in Basque Country, north of Spain. By carefully designing the search space and incorporating domain expertise, the approach quickly converges to optimal and highly accurate network configurations with both efficiency and efficacy. LSTMs ingested precipitation, temperature, and potential evapotranspiration as inputs to predict 2 targets of streamflow and water level, in an hourly timestep. On the test set, the optimized LSTM networks accurately predicted streamflow and water level with Nash-Sutcliffe (NSE) and Kling-Gupta (KGE) efficiencies as high as 0.97, in one of the catchments. Across all 40 studied catchments, the overall average NSE and KGE values for streamflow were 0.89 and 0.87, respectively; water level exhibited average NSE and KGE scores of 0.91 and 0.92.Moreover, statistical analysis reveals significant differences in the performance of the 2 distinct optimized network architectures in different hydrological catchments, underscoring the importance of deliberate network configuration selection post-random search. This selection process is vital for achieving higher performance in as many catchments as possible. The findings highlight opportunities for enhancing the “learning maturity” of regional hydrological deep learning LSTM networks. This research provides valuable insights for researchers and practitioners involved in optimizing regional hydrological deep learning models for a variety of applications and on new datasets.

Sensitivity of montane grassland water fluxes to warming and elevated CO2 from local to catchment scale: A case study from the Austrian Alps

Study region: Montane grassland within the Gulling catchment, Austrian Alps. Study focus: A climate-change experiment in a grassland ecosystem used lysimeters and HYDRUS-1D models to quantify changes in evapotranspiration (ET) and groundwater recharge (GWR) due to warming (＋3 °C) and elevated CO2 concentrations (ΔCO2; ＋300 ppm). Findings at the plot-scale were generalized and transferred to the surrounding catchment, half comprised of grassland, using three lumped rainfall–runoff models and two spatially-distributed Community Water Models, differing in soil hydraulic properties.New hydrological insights for the region: Warming increased ET and decreased GWR and river discharge compared to ambient conditions. ΔCO2 increased stomatal resistance, which partially offset warming effects. In scenarios combining warming and ΔCO2, the impact of warming was higher than ΔCO2 effect. Elevation influenced the sensitivity of ET to warming, which was greater at the catchment scale than at the plot scale, while GWR was more sensitive to warming at the plot scale. Under dry conditions, GWR and discharge exhibited increased sensitivity to warming at both scales. HYDRUS-1D successfully reproduced lysimeter experiment results and their sensitivity to warming and ΔCO2. Despite model agreement on water flux sensitivity to climate changes, the varying response magnitudes highlight the need for a multi-model approach in climate impact assessments. This study provides insights into how climate change might impact hydrological dynamics of montane grassland systems across the Central European Alps.

Rainfall-runoff modeling using artificial neural networks in the Mono River basin (Benin, West Africa)

Hydrological models are developed to simulate river flows over a watershed for many practical applications in the field of water resource management. However, the rainfall-runoff models mostly used in the Mono river basin struggle to better simulate high river flows. This paper presents a modeling approach based on Artificial Neural Networks (ANN) under different input meteorological parameters in the Mono River basin to better take into account the non-linearity of the relationship between rainfall and runoff. To this end, precipitation, potential evapotranspiration, and previously observed flow have been used for the daily flow simulation. The Levenberg-Marquardt algorithm is used to train the ANN rainfall-runoff models over the other optimization training algorithms mostly implemented in the study area. The analysis of the rainfall-runoff variability allowed us to show the strong correlation between rainfall and runoff and the impact of the Nangbéto dam on the flows at Athiémé. The results obtained after the training, validation, and testing of the ANN models are satisfactory (e.g., the coefficient of correlation varies between 0.93 and 0.99). The most efficient model has been identified and implemented in the Mono river basin at Nangbéto. The satisfactory results obtained show that ANN models can be considered good alternatives for traditional rainfall-runoff modeling approaches.

Rainfall-runoff modeling using artificial neural networks in the Mono River basin (Benin, West Africa)

Hydrological models are developed to simulate river flows over a watershed for many practical applications in the field of water resource management. However, the rainfall-runoff models mostly used in the Mono river basin struggle to better simulate high river flows. This paper presents a modeling approach based on Artificial Neural Networks (ANN) under different input meteorological parameters in the Mono River basin to better take into account the non-linearity of the relationship between rainfall and runoff. To this end, precipitation, potential evapotranspiration, and previously observed flow have been used for the daily flow simulation. The Levenberg-Marquardt algorithm is used to train the ANN rainfall-runoff models over the other optimization training algorithms mostly implemented in the study area. The analysis of the rainfall-runoff variability allowed us to show the strong correlation between rainfall and runoff and the impact of the Nangbéto dam on the flows at Athiémé. The results obtained after the training, validation, and testing of the ANN models are satisfactory (e.g., the coefficient of correlation varies between 0.93 and 0.99). The most efficient model has been identified and implemented in the Mono river basin at Nangbéto. The satisfactory results obtained show that ANN models can be considered good alternatives for traditional rainfall-runoff modeling approaches.

Natural Flow Simulation in a Baisin considering Artificial Water Cycle

This study seeks to regionalize the parameters of the grid-based rainfall-runoff model (GRM) and simulate the natural flow considering an artificial water cycle. We established a water supply and demand network for the Han River basin, derived data related to the artificial water cycle through various organizations, and performed a water budget analysis by reflecting the simulated natural flow. When comparing the results of regionalization using only the Chungju Dam parameters and those using the Chungju Dam and Soyanggang Dam parameters, both results for the simulated flow showed good agreement with the observed flow, and the fitness evaluation results were also high (pentad: NSE 0.913–0.977, KGE 0.795–0.935, CC 0.974–0.995). The natural flow simulated by the GRM and TANK indicated that both models were able to reproduce the observed data well. Water budget analyses were performed for different time intervals, and daily analysis exhibited relatively lower fitness than pentad analysis (daily NSE 0.887-0.963, KGE 0.792-0.928, CC 0.960-0.990), but no significant difference was observed. The results indicated that the simulation of natural flow through the GRM was appropriate, and the model was sufficiently applicable to a disturbed watershed.

Ensemble of artificial intelligence and physically based models for rainfall–runoff modeling in the upper Blue Nile Basin

ABSTRACT This study investigated the performance of the adaptive neuro-fuzzy inference system (ANFIS), feed forward neural network (FFNN), Soil and Water Analysis Tool (SWAT), Hydrologic Engineering Center's Hydraulic Modeling System (HEC-HMS), Hydrologiska Byråns Vattenbalansavdelning (HBV), and support vector regression (SVR) models for rainfall–runoff modeling using gauged and satellite rainfall, and their fusions in the Gilgel Abay watershed, Ethiopia. Afterward, simple average ensemble (SAE), weighted average ensemble (WAE), and neural network ensemble (NNE) techniques were applied to combine the outputs of individual models under three scenarios. The performance of the models was evaluated using Nash–Sutcliffe efficiency (NSE) and root mean square error (RMSE). The results demonstrated that the ANFIS model outperformed all the other single models with validation stage NSE values of 0.864 and 0.875, and RMSE values of 23.58 and 21.84 m3/s for gauge and fusion rainfall data, respectively. Among the physical-based models, SWAT gave better modeling performance with the validation stage NSE values of 0.81 and 0.821 for gauge and fusion rainfall data, respectively. Moreover, an ensemble of artificial intelligence and physical-based models greatly improved the overall modeling performance. The NNE improved the performance of single models up to 15.7 and 21.2 5% for fusion and satellite-based rainfall modeling, respectively.

Rainfall event separation and determination of the water quality volume

A procedure is presented and demonstrated for identifying the minimum interevent time for automatic separation of rainfall events and the subsequent use of the event statistics to determine the water quality volume. The proposed procedure for identifying the minimum interevent time is more robust that the conventional approach that requires interevent times to have an exponential distribution. The procedure is validated in South Florida and at locations in Athens (Georgia), Roanoke (Virginia), and St.Louis (Missouri). Minimum interevent times of 6h, 6h, 12h, and 9h, respectively, were found at the study locations, and it was demonstrated that the interevent times do not generally have an exponential distribution. Conventional determination of the water quality volume does not formally consider the influence of the drain time in sizing the water quality storage basin, and a procedure is presented herein for taking the drain time into account. The procedure is demonstrated at the study locations using multiple rainfall-runoff models. It is shown that a maximum drain time of 24h would not generally be adequate in South Florida, but would be adequate in Athens, Roanoke and St.Louis, however, a maximum drain time of 48h would not generally be adequate at all locations.

Assessment of monthly hydroclimatic patterns and rainfall-runoff modeling for hydrometric forecasting in the Upper Inaouene Basin, Northern Morocco

Assessment of monthly hydroclimatic patterns and rainfall-runoff modeling for hydrometric forecasting in the Upper Inaouene Basin, Northern Morocco

Revisiting the tension water storage capacity distribution in conceptual rainfall-runoff modeling: A large-sample approach

Accurately characterizing the spatial variability of tension water storage capacity (TWC) within a catchment is challenging due to limited in-situ hydrologic data availability. Conventional conceptual rainfall-runoff models typically rely on an empirically specified TWC distribution. However, this empirical distribution lacks a physical foundation and fails to effectively redistribute critical hydrologic components, such as local capacity and contributing area, to real-world contexts. To overcome this limitation, the topographic wetness index (TWI) and its generalized form (GTWI) are introduced to bridge local topographic information with hydrologic components. Four TWC distribution curves are contrived based on the empirical parabolic distribution, empirical linear distribution, TWI, and GTWI, respectively. The effects of these alternate TWC distributions on streamflow are investigated within the framework of the HYdrologic MODel (HYMOD) across 460 Australian catchments. The results illustrate that the GTWI-based HYMOD (GTHYMOD) outperforms other models in terms of daily streamflow, with high Kling-Gupta Efficiency (KGE) attained in 74.8% of the study catchments during the validation period. The eastern coast of Australian catchments presents a superior streamflow performance compared to that in the western coast. GTHYMOD demonstrates its superiority in characterizing spatial variability, an aspect HYMOD lacked. This study has the potential to refine the empirical TWC distribution from a physical perspective and advance our comprehension of underlying hydrologic behaviors.

Comparative analysis of data driven rainfall-runoff models in the Kolar river basin

To effectively tackle the challenges posed by climate change, it is crucial to enhance the accuracy of rainfall-runoff models to ensure reliability amidst changing climatic conditions. Neural networks, renowned for their ability to capture complex patterns and relationships within uncertain input and output data, offer valuable tools in this pursuit. This study aims to evaluate the efficacy of two neural network (NN) models: the Radial Basis Function Neural Network (RBFNN) and the Model Tree M5 Neural Network (MTM5NN). These models are assessed both individually and in combination with the Wavelet (WT) data processing technique for rainfall-runoff modeling in the Kolar River watershed located in Madhya Pradesh, India. Fifteen models were developed employing four algorithms: RBFNN models, WRBFNN (RBF model integrating wavelet components of rainfall as inputs), MTM5NN, and WMTM5NN (MT model incorporating wavelet components of rainfall as inputs). Initially, rainfall and runoff data underwent normalization and were applied to RBFNN and MTM5NN networks. Subsequently, time series data for both rainfall and runoff were decomposed using wavelet transforms, resulting in various sub-time series signals such as approximations and decompositions. These derived signals were then utilized as input data for RBFNN and MTM5NN, specifically designated as WRBFNN and WMTM5NN. The most effective model identified was Model 8 of WMTM5NN, which demonstrated R2 values close to 0.97, outperforming the other models. These results underscore the superior performance of the WMTM5NN model, highlighting its effectiveness in achieving heightened accuracy in predicting runoff for this specific watershed.

Applying Machine Learning Methods to Improve Rainfall–Runoff Modeling in Subtropical River Basins

Applying Machine Learning Methods to Improve Rainfall–Runoff Modeling in Subtropical River Basins

Developing Water Quality Formulations for a Semi-Distributed Rainfall–Runoff Model

Hydrological modeling can be challenging due to significant data requirements and computational complexities. Hydrological models must be sufficiently complex to describe physical processes yet simple enough to use. This paper describes the development of a simplified watershed-scale input–output model to simulate runoff quantity and quality during a storm event. This work builds upon an existing semi-distributed rainfall–runoff model by adding calculations for pollutant concentrations based on simplified mass balance equations. The model was tested against various watershed examples of increasing complexity. The results show the change in peak flow and pollutant concentration in different areas of the watershed, demonstrating the model’s ability to account for the dynamics of runoff movement through the watershed. This paper advances watershed management by addressing data scarcity through the development of a simplified hydrological model that effectively incorporates spatial variability within a watershed while requiring minimal data input.

Hydrological modelling using SWAT for the assessment of streamflow dynamics in the Ganga River basin.

Growing concerns over water availability arise from the problems of population growth, rapid industrialization, and human interferences, necessitating accurate streamflow estimation at the river basin scale. It is extremely challenging to access stream flow data of a transboundary river at a spatio-temporal scale due to data unavailability caused by water conflicts for assessing the water availability.Primarily, this estimation is done using rainfall-runoff models. The present study addresses this challenge by applying the soil and water assessment tool (SWAT) for hydrological modelling, utilizing high-resolution geospatial inputs. Hydrological modelling using remote sensing and GIS (Geographic Information System) through this model is initiated to assess the water availability in the Ganga River basin at different locations. The outputs are calibrated and validated using the observed station data from Global Runoff Data Centre (GRDC). To check the performance of the model, Nash-Sutcliffe efficiency (NSE), percent bias (PBIAS), coefficient of determination (R2), and RSR efficacy measures are initiated in ten stations using the observed and simulated stream flow data. The R2 values of eight stations range from 0.82 to 0.93, reflecting the efficacy of the model in rainfall-runoff modelling. Moreover, the results obtained from this hydrological modelling can serve as valuable resources for water resource planners and geographers for future reference.

Event-based flood estimation in un-gauged sub-basins: a comparative assessment of SCS-UH, CWC-UH and Nash-GIUH based rainfall-runoff models in Shilabati River, Eastern India

Event-based flood estimation in un-gauged sub-basins: a comparative assessment of SCS-UH, CWC-UH and Nash-GIUH based rainfall-runoff models in Shilabati River, Eastern India

Rainfall Runoff Modelling of Panchaganga River Using HEC-HMS Software and it's Analysis.

The present study of this project is intended on the formation of a working flood model of the river Panchaganga flood in the Kolhapur stretch which is in Maharashtra state in India. Using HEC-HMS the precipitation is used to estimate the run off along the watershed. The central aim of this project resides in the development of a calibrated working model of the 2019 floods. This model is later used to simulate the various possibilities of flood event. Rainfall Runoff Model which is a mathematical model which relates the rainfall and runoff with the use of catchment area or watershed area. Hydrological modeling can estimate for the runoff in the study area of the model by evaluating the precipitation and discharges inputs given to the HEC-HMS software. Hydrological models on the other hand estimates the surface runoff within the study area. In recent decades, India has experienced an increase in the frequency of flood disasters. Flood is a flow that overtops either hydraulic structure or river bank along a stream resulting in damage to the adjoining property or life. The primary reason behind this is the Climate change and Immense urbanization, which has resulting in alteration of basin characteristics. The development projects near riverbanks have the worst effects on a reach's ability to drain, which increases the likelihood of flood-like conditions. In this project an area of the Panchaganga river is to be taken and a flood mapping using HEC-HMS (Hydrological Modelling System) and GIS software is to be drawn. According to the drawn flood map there will be a suitable mitigation measure suggested and the effect of the suggested mitigation measure is to be modeled.

A national-scale hybrid model for enhanced streamflow estimation – consolidating a physically based hydrological model with long short-term memory (LSTM) networks

Abstract. Accurate streamflow estimation is essential for effective water resource management and adapting to extreme events in the face of changing climate conditions. Hydrological models have been the conventional approach for streamflow interpolation and extrapolation in time and space for the past few decades. However, their large-scale applications have encountered challenges, including issues related to efficiency, complex parameterization, and constrained performance. Deep learning methods, such as long short-term memory (LSTM) networks, have emerged as a promising and efficient approach for large-scale streamflow estimation. In this study, we have conducted a series of experiments to identify optimal hybrid modeling schemes to consolidate physically based models with LSTM aimed at enhancing streamflow estimation in Denmark. The results show that the hybrid modeling schemes outperformed the Danish National Water Resources Model (DKM) in both gauged and ungauged basins. While the standalone LSTM rainfall–runoff model outperformed DKM in many basins, it faced challenges when predicting the streamflow in groundwater-dependent catchments. A serial hybrid modeling scheme (LSTM-q), which used DKM outputs and climate forcings as dynamic inputs for LSTM training, demonstrated higher performance. LSTM-q improved the mean Nash–Sutcliffe efficiency (NSE) by 0.22 in gauged basins and 0.12 in ungauged basins compared to DKM. Similar accuracy improvements were achieved with alternative hybrid schemes, i.e., by predicting the residuals between DKM-simulated streamflow and observations using LSTM. Moreover, the developed hybrid models enhanced the accuracy of extreme events, which encourages the integration of hybrid models within an operational forecasting framework. This study highlights the advantages of synergizing existing physically based hydrological models (PBMs) with LSTM models, and the proposed hybrid schemes hold the potential to achieve high-quality large-scale streamflow estimations.

UIDS: A Matlab-based urban flood model considering rainfall-induced and surcharge-induced inundations

The Urban Inundation-Drainage Simulator (UIDS) is a new coupled model for simulating urban flooding dynamics, developed as an open-source, MATLAB-based platform. It integrates a rainfall-runoff model with a two-dimensional overland flow model (OFM) and a one-dimensional sewer flow model (SFM). Unlike conventional models limited to either rainfall-induced or sewer surcharge-induced flooding, UIDS captures bidirectional surface-underground interactions to simulate both processes simultaneously. The OFM employs an explicit time-stepping scheme and robust wet-dry front treatment, while a weir equation describes roof-to-ground flow exchange for numerical stability. Timing synchronization facilitates continuous OFM-SFM coupling. Benchmarking and case studies of Gangnam flood events demonstrate UIDS's ability to accurately simulate urban flooding, particularly subcritical flows. The open-source nature of UIDS allows user flexibility in accessing and modifying the MATLAB code. Ultimately, UIDS is expected to serve as an accessible and adaptable tool for urban flood modeling and risk assessment.

Improved understanding of calibration efficiency, difficulty and parameter uniqueness of conceptual rainfall runoff models using fitness landscape metrics

The ease and efficiency with which conceptual rainfall runoff (CRR) models can be calibrated, as well as issues related to the uniqueness of their parameters, has received significant attention in literature. While several studies have tried to gain a better understanding of the underlying factors affecting these issues by examining the features of model error surfaces, this has generally been done in an ad-hoc fashion using lower-dimensional representations of higher-dimensional surfaces. In this paper, it is suggested that exploratory landscape analysis (ELA) metrics can be used to quantify key features of the error surfaces of CRR models, including their roughness and flatness, as well as their degree of optima dispersion throughout the surface. This enables key error surface features of CRR models to be compared in a consistent, efficient and easily communicable fashion for models with different combinations of attributes (e.g. model structure, catchment climate conditions, error metrics, and calibration data set lengths). Results from the application of ELA metrics to the error surfaces of 420 CRR models with different combinations of the above attributes indicate that increasing model complexity results in an increase in relative error surface roughness and relative optima dispersion and that, while increasing catchment wetness increases the relative roughness of error surfaces, it also decreases optima dispersion. This suggests that for the models considered in this study, optimisation efficiency is likely to decrease with increasing model complexity and catchment wetness, while optimisation difficulty is likely to increase and parameter uniqueness likely to decrease with model complexity and catchment dryness. While implications for choice of model complexity will need further work, this study highlights the potential value of the proposed approach to understanding the calibration efficiency, difficulty and parameter uniqueness of conceptual rainfall runoff models.

Using supervised machine learning for regional hydrological hazard estimation in metropolitan France

Study regionThis study is carried out for 1929 gauged catchments in France, ranging from 1 to 10,000 km², where quality hydrometric observations are available for flood frequency analysis. Study focusThe regional estimation of hydrological hazards is studied for flood risk management and prevention in hydrology. For gauged catchments, flow quantiles can be estimated from observations using statistical approaches based on suitable probability distributions or simulation approaches based on rainfall-runoff transformation models. For ungauged catchments, the lack of hydrological observations means that we have to extrapolate our knowledge of hazards from gauged catchments to ungauged catchments, using regionalization methods. It is therefore necessary to combine regionalization methods with the implemented hazard estimation approach. In this paper, two popular machine learning methods, Random Forest and Neural Networks, are tested and compared as regionalization methods. A classical regionalization method using multiple linear regression is also applied as a benchmark to evaluate the performance of all configurations. All these regionalization methods are applied to a simulation-based approach (the SHYREG method) and to a statistical-based approach using generalized extreme value distribution (GEV). New hydrological insights•Regionalization approaches based on multiple linear regression have limitations to explain parameters with environmental descriptors in Regional Flood Frequency Analysis (RFFA) domain.•Regionalizing RFFA parameters using Random Forest allows more explanatory variables to be considered through non-linear relationships, resulting in better parameter estimation.•Machine learning techniques can better handle environmental descriptors for regionalization, this providing a notable performance improvement, especially for the statistical approach.•The tested simulation-based approach is less sensitive to the choice of spatial interpolation method than the studied statistical approach.

A hybrid CNN–RNN model for rainfall–runoff modeling in the Potteruvagu watershed of India

AbstractAccurate rainfall‐runoff analysis is essential for water resource management, with artificial intelligence (AI) increasingly used in this and other hydrological areas. The need for precise modelling has driven substantial advancements in recent decades. This study employed six AI models. These were the support vector regression model (SVR), the multilinear regression model (MLR), the extreme gradient boosting model (XGBoost), the long‐short‐term memory (LSTM) model, the convolutional neural network (CNN) model, and the convolutional recurrent neural network (CNN‐RNN) hybrid model. It covered 1998–2006, with 1998–2004 for calibration/training and 2005–2006 for validation/testing. Five metrics were used to measure model performance: coefficient of determination (R2), Nash‐Sutcliffe efficiency (NSE), mean absolute error (MAE), root‐mean square error (RMSE), and RMSE‐observations standard deviation ratio (RSR). The hybrid CNN‐RNN model performed best in both training and testing periods (training: R2 is 0.92, NSE is 0.91, MAE is 10.37 m3s−1, RMSE is 13.13 m3s−1, and RSR is 0.30; testing: R2 is 0.95, NSE is 0.94, MAE is 12.18 m3s−1, RMSE is 15.86 m3s−1, and RSR is 0.25). These results suggest the hybrid CNN‐RNN model is highly effective for rainfall‐runoff analysis in the Potteruvagu watershed.