Nash-Sutcliffe Efficiency Criteria Research Articles

ANALISIS POLA DISTRIBUSI CURAH HUJAN LEBAT DOMINAN DI DAERAH ALIRAN SUNGAI (DAS) CITANDUY HULU

Rainfall plays an important role in hydrological and environmental planning. Meteorological conditions affect rainfall through its intensity, amount, and duration. However, heavy rainfall can cause significant damage and loss. Research was conducted in the Upper Citanduy watershed, West Java, using rainfall data from 8 stations over a range of 5–14 years. The study took a heavy rainfall event threshold of ≥ 50 mm in 24 hours. The research aims to analyze the dominant heavy rainfall patterns in the Upper Citandyu watershed. Empirical analysis used the Mononobe Method and Alternating Block Method (ABM), while rainfall distribution was analyzed with Inverse Distance Weighted (IDW) using Geographic Information Systems. The dominant heavy rainfall duration in the Upper Citanduy watershed occurred over 4–6 hours. Rainfall distribution analysis showed a pattern that was not influenced by the elevation of the rainfall station. The analysis results show that the Mononobe Method is closer to the dominant observed heavy rainfall distribution for 4 hours, while ABM is closer to the durations of 5 and 6 hours. The suitability of the distribution pattern was measured using Correlation (r), Root Mean Squared Error (RMSE), and Nash-Sutcliffe Efficiency (NSE) criteria, with Mononobe providing better results than ABM method. This research contributes to the understanding of rainfall patterns that can form the basis for analyzing designed rainfall intensity as a step in flood disaster mitigation in the Upper Citanduy watershed.Keywords: heavy rainfall, rainfall distribution pattern, rainfall intensity, upper Citanduy watershed, geographic information system

Developing some models to predict the uniaxial compressive strength of various sedimentary rocks (Case studies: Large dam site and mine in Southeast China)

Direct determination of uniaxial compressive strength (UCS) is time-consuming, expensive, and challenging. Therefore, this study aims to indirectly predict UCS using statistical and machine learning methods. First, laboratory measurements were conducted to determine carbonate percentage, water absorption percentage, moisture content, porosity percentage, P-wave velocity, and UCS of sandstone, dolomite, dolomitic limestone, marl limestone, and limestone samples. Subsequently, various models were established to predict UCS using multivariate linear regression (MVLR), random forest regression (RFR), Gaussian process regression (GPR) based on squared exponential kernel function, and adaptive neuro-fuzzy inference system (ANFIS). The influence of Jurassic limestone texture on physical and mechanical properties in stable and unstable slopes in the Three Gorges Dam site and reservoir (TGDSR) showed that the percentage of mudstone texture in unstable slopes is higher than in stable slopes. The UCS, density, compressional wave velocity, and percentage of carbonate in unstable slopes is less than stable slopes. The results indicated that the average measured UCS is 60.60 MPa. Moreover, the average predicted UCS based on the three intelligent methods was 60.55 MPa. The percentage difference between the measured and predicted UCS was −0.09 %, demonstrating that the average error of the three employed methods is less than one percent, highlighting the high accuracy of these methods in estimating rock UCS. Notably, based on error level, Taylor diagram, Wilmot agreement index (WAI), A10 and A20 indices, Nash-Sutcliffe efficiency (NSE), and PM (performance metric) criteria the RFR exhibited superior precision (R2=99 %, and RMSE= 0.06 MPa) compared to the other used methods. The Kruskal-Wallis test (KWT) results showed that there is no significant difference between the measured and predicted values.

Hydrological Modeling of the Cambamba Watershed in Angola Using the HEC-HMS Model

When evaluating the impact of climate change on water resources in river basins it is crucial to accurately estimate the availability of water and such can be attained through hydrological modeling of the basin. The studies of the Hydrological Modeling System (HEC-HMS) for the Cambamba River Basin in Angola has been calibrated and validated for the prediction of its hydrologic response. Due to the complexity of Hydrologic models, a good calibration of the model should be established, therefore, improving its skills and effectiveness. A combination of both the energy budget and a physically-based rainfall-runoff model, which are used for the estimation of lake evaporation, were applied in the lake water balance model. Soil storage, as well as groundwater storage coefficient, are essential parameters for estimating the simulated streamflow. For an adequate performance evaluation of the model which entails the simulation of the streamflow on Cambamba watershed and quantification of water availability, The Nash-Sutcliffe model efficiency criterion; root mean square error; percentage error in volume PEV; and percentage error in peak PEP, were implemented. Model performance is improved by using semiannual parameter sets that consider the change in hydrological conditions, as evaluated in this study. Key words: HEC-HMS, watershed, rainfall-runoff, flood hazards

Design and Simulation of Stormwater Control Measures Using Automated Modeling

Stormwater control measures (SCMs) are decentralized technical elements, which can prevent the negative effects of uncontrolled stormwater flow while providing co-benefits. Optimal SCMs have to be selected and designed to achieve the desired hydrological response of an urban catchment. In this study, automated modeling and domain-specific knowledge in the fields of modeling rainfall-runoff (RR) and SCMs are applied to automate the process of optimal SCM design. A new knowledge library for modeling RR and SCMs, compliant with the equation discovery tool ProBMoT (Process-Based Modeling Tool), was developed. The proposed approach was used to (a) find the optimal RR model that best fits the available pipe flow measurements, and (b) to find the optimal SCMs design that best fits the target catchment outflow. The approach was applied to an urban catchment in the city of Ljubljana, Slovenia. First, nine RR models were created that generally had »very good« performance according to the Nash–Sutcliffe efficiency criteria. Second, six SCM scenarios (i.e., detention pond, storage tank, bio-retention cell, infiltration trench, rain garden, and green roof) were automatically designed and simulated, enabling the assessment of their ability to achieve the target outflow. The proposed approach enables the effective automation of two complex calibration tasks in the field of urban drainage.

Calibration and Validation of HEC-HMS Model for Chalakudy River Basin

Prediction of flood prone areas in a basin and evolution of the impact of climate change on water resources needs a correct estimation of the availability of water which will solely be achieved by hydrological modelling of the basin. However, modelling the hydrology of a basin is a complex task and models should be well calibrated to increase user confidence in its predictive ability which in turn makes the application of the model effective. In this study rainfall-runoff simulation model viz., Hydrologic Modelling System, developed by the Hydrologic Engineering Centre USA (HEC-HMS) has been calibrated and validated for Chalakduy river basin in Kerala, in Sothern India for prediction of its hydrologic response. The result shows Curve Number (CN), Lag time and initial abstraction (Ia) to be the sensitive parameters for the simulated stream flow. The statistical analysis of Nash-Sutcliffe model efficiency criteria, the percentage error in peak, percentage error in volume, and net difference of observed and simulated time to peak, which were used for performance evaluation, have been found to range from (0.70 to 0.87), (4.39 to 19.47%), (1.9 to 19%) and (0 to 1day) respectively, indicating a very good performance of the model for simulation of stream flow.

MSGP-LASSO: An improved multi-stage genetic programming model for streamflow prediction

This paper presents the development and verification of a new multi-stage genetic programming (MSGP) technique, called MSGP-LASSO, which was applied for univariate streamflow forecasting in the Sedre River, an intermittent river in Turkey. The MSGP-LASSO is a practical and cost-neutral improvement over classic genetic programming (GP) that increases modelling accuracy, while decreasing its complexity by coupling the MSGP and multiple regression LASSO methods. The new model uses average mutual information to identify the optimum lags, and root mean-square technique to minimize forecasting error. Based on Nash-Sutcliffe efficiency and bias-corrected Akaike information criterion, MSGP-LASSO is superior to GP, multigene GP, MSGP, and hybrid MSGP-least-square models. It is explicit and promising for real-life applications.

A novel method for lake level prediction: deep echo state network

Accurately prediction of lake level fluctuations is essential for water resources planning and management. In the present study, the potential of a novel method, deep echo state network (Deep ESN), is investigated for monthly lake level prediction and its results are compared with three data-driven methods, artificial neural networks (ANNs), extreme learning machine (ELM), and regression tree (Reg. Tree). The methods are validated using root mean square errors (RMSE), determination coefficient (R2), and Nash-Sutcliffe efficiency (NSE) criteria. The investigated method (Deep ESN) outperforms the ELM, ANNs, and Reg. Tree by improving accuracies by 61–62–96%, 10–14–84%, and 8–23–80% in prediction 1 month, 2 months, and 3 months ahead lake level fluctuations in terms of RMSE criteria, respectively. Also, accuracy of ELM, ANNs, and Reg. Tree was significantly increased using Deep ESN model by 1.1–1.1–443%, 1.1–1.6–250%, and 1.6–6.5–184% in terms of NSE indicator for different lead-time horizons. Among the ELM, ANNs, and Reg. Tree, the third method provides the worst predictions while the first method performs superior to the second one in all tree time horizons.

Tree-based nonlinear ensemble technique to predict energy dissipation in stepped spillways

In this study, under the skimming flow regime, energy dissipation was investigated with data-driven methods. The data set obtained from the laboratory experiments were modelled by different machine learning (ML) methods, including support vector machine (SVM), K-star (K*) algorithm and artificial neural networks (ANN). Afterwards, for the first time in the literature, linear and nonlinear ensemble models were established in order to improve the accuracy of single models in predicting energy dissipation. Simple average (SA-E) and weighted average (WA-E) were performed as linear ensemble models while M5 Model Tree (M5-MTE) and Random Forest (RF-E) were used to establish non-linear ensemble models. The model results were evaluated according to performance metrics, such as Coefficient of Correlation (CC), Percent Bias (PBIAS), Performance Index (PI), Willmott’s index of agreement (WI) and Nash-Sutcliffe efficiency criteria (NSE). The NSE values are calculated as 0.986, 0.909 and 0.985 for SVM, K* and ANN models, respectively. Moreover, for the ensemble models, higher NSE values were obtained for both linear (NSESA-E = 0.9887, NSEWA-E = 0.9916) and tree-based non-linear (NSEM5 MT-E = 0.9963, NSERF-E = 0.9974) models. Overall, it can be stated that tree-based ensemble models make better predictions for energy dissipation calculation in step spilways compared to single ML methods.

Assessment of applicability of mike 11-nam hydrological module for rainfall runoff modelling in a poorly studied river basin

Introduction. The need to simulate hydrological processes is caused by, among other factors, the complexity of hydrological systems and data insufficiency due to the unavailability or a small number of instrumental observations. Recently, the reanalysis of the climate data supplied by the world’s leading meteorological centres has been used quite successfully in the regions that suffer from the deficit of instrumental information. This paper assesses the applicability of climate reanalysis data to rainfall runoff (“rainfall runoff”) modelling in the poorly studied river basin in Eritrea. Materials and methods. Climate Forecast System Reanalysis (CFSR) data generated by the National Centre for Environmental Prediction (USA) were used. Besides, high-resolution topographic information, generated by the SRTM international research project, was also applied to set the drainage area boundaries and to simulate the river network using such tools as MIKE and GIS. In addition, calibration and validation (evaluation) of the hydrological model (simulation quality) were performed using the Nash-Sutcliffe efficiency criterion, the determination coefficient, and the root mean square error of volumetric and peak flow rates. Results. The results suggest that a considerable overestimation of precipitation in the reanalysis data set, which in turn has a significant effect on other variables such as potential evapotranspiration, leads to a significant discrepancy between water balance values which are simulated and registered by the hydrographs. Conclusions. The applicability of Climate Forecast System Reanalysis (CFSR) data to river flow modelling in arid and semi-arid regions such as Eritrea is questionable. The incompatibility of spatial and temporal variations of initial variables (e.g. precipitation), derived from reanalysis data sets and instrumental observations, is undoubtedly the main reason for errors. Thus, the application of reanalysis data sets and development of hydrological models for the region under study requires further intensive research aimed at identifying most effective mechanisms designated for the harmonization of differences between reanalysis data and field observations. In the course of further research, CFSR information is to be converted into more realistic data; climate reanalysis indicators, provided by other sources and designated for different time scales in the context of the “rainfall runoff” model are to be assessed, and the efficiency of other software systems is to be compared with MIKE 11-NAM.

Pan evaporation estimation and derivation of explicit optimized equations by novel hybrid meta-heuristic ANN based methods in different climates of Iran

Pan evaporation (Ep) estimation is important in scheduling and computing irrigation water requirement. This study evaluated the ability of novel meta-heuristic optimization algorithms including Genetic Algorithm (GA), Grey Wolf Optimization (GWO), and Whale Optimization Algorithm (WOA) that hybridized with artificial neural networks (ANNs) in estimating daily Ep and optimizing explicit predictive equations. Five meteorological stations in five climate areas (Humid, sub-humid, semi-arid, arid, hyper-arid) located in Iran, and each station having ten scenarios of available input data, were used. The Subset Selection of Maximum Dissimilarity (SSMD) was used for pre-processing of data set and automatic selection of the training and testing subsets. The modeling results were compared based on the global performance indicator (GPI), root mean square error (RMSE), coefficient of determination (R2), and Nash-Sutcliffe Efficiency (NSE) criteria in addition to Taylor diagram and Box-plots. The results indicated that the performance of models of ANN-GA-4 (I4: Tmean, RH; R2=0.83, RMSE=0.95, NSE=0.82, GPI=0.68) for Astara station, ANN-GA-1 (I1: Tmax; R2=0.79, RMSE=1.39, NSE=0.78, GPI=0.68) for Gorgan station, ANN-GA-5 (I5: Tmean, n; R2=0.86, RMSE=1.98, NSE=0.86, GPI=0.96) for Tabas station, ANN-6 (I6: Tmean, U2) for Esfahan (R2=0.77, RMSE=1.57, NSE=0.77, GPI=0.6) and Hamedan (R2=0.69, RMSE=1.43, NSE=0.68, GPI=1.06) stations were superior than the other models with limited climatic data. After selecting best model in each station, the explicit optimized equations to calculate the daily Ep in each climate are provided as another major contribution of the present study.

From marginal to conditional probability functions of parameters in a conceptual rainfall–runoff model: an event-based approach

ABSTRACTA parameter estimation strategy for a conceptual rainfall–runoff (CRR) model applied to a storm sewer system in an urban catchment (Chassieu, Lyon, France) is proposed on the basis of event-by-event Bayesian local calibrations. The marginal distribution formed by locally-estimated parameters is divided into conditional functions, clustering the event-based parameters based on their transferability to similar rainfall events. The conditional functions showed to be consistent with an observed bimodality in the marginal representation, reflecting two different hydrological conditions mainly related to the magnitude of the rainfall intensities (high or low). The improvements achieved by expressing the parameter probability functions into a conditional form are shown in terms of accuracy (Nash-Sutcliffe efficiency criterion), precision (average relative interval length) and reliability (percentage of coverage) for simulating flow rate in 255 and 110 calibration/verification events.

Wavelet-Exponential Smoothing: a New Hybrid Method for Suspended Sediment Load Modeling

In this study, four conventional and a newly proposed method of wavelet-exponential smoothing (WES) - with two presented scenarios (WES1 and WES2) – are employed to estimate daily and monthly suspended sediment load (SSL) in two rivers (Lighvanchai river in Iran and Upper Rio Grande in the USA), which have different hydro-geomorphological characteristics of the related watersheds. In the proposed WES method, first, wavelet transform (WT) is applied to the original observed time series to decompose them into approximation and detailed subseries to separate different components of time series. For the first scenario (WES1), only two time series, i.e., an approximation and a detail time series are utilized as inputs of model, whereas for the second scenario (WES2), all subseries are separately fed into different exponential smoothing (ES) models. The results revealed that for both rivers, the proposed WES2 and wavelet based artificial neural network (WANN) models could lead to superior performance in comparison to the autoregressive integrated moving average (ARIMA), seasonal ARIMA (SARIMA), ES ad-hoc and artificial neural network (ANN). The WES2 method could enhance the overall performance of SSL forecasting both in daily and monthly modeling of the case studies regarding Nash-Sutcliffe (E) efficiency criteria, respectively up to 13%, 42% and 87%, 116% in daily and monthly scales for SSL modeling of the Lighvanchai and Upper Rio Grande Rivers. As a result, combining WT with ES method and ANN led to more accurate modeling.

Calibração conjunta de um modelo hidrológico com parâmetros de curva-chave para simulações de inundações bruscas em áreas urbanas

ABSTRACT Calibration process is usually time demanding and as much streamflow information as possible in rainfall-runoff models. Nevertheless, from a practical point of view, the available information is measurement of water level, which is essential to design rating curves. This manuscript proposes a set of joint calibration of hydrological model parameters with a range of rating curves, developed for the main channel of the catchment in a crowded urban area. As an alternative of free calibration, the simulations were carried out based on a list of proficient parameters. Four streamflow gauging stations were analysed and used to subdivide the basin. The hourly lumped rainfall-runoff model GR4H was applied to four critical flash flood events to create a rank of Nash-Sutcliffe efficiency criteria (NSE) applying the best set of parameters. The results indicated that the hydrological model errors were compensated by hydraulics errors and they presented an equifinality in the process.

Determining optimal combination of roller compacted concrete pavement mixture containing recycled asphalt pavement and crumb rubber using hybrid artificial neural network–genetic algorithm method considering energy absorbency approach

The present study investigates the effectiveness of evolutionary algorithms such as genetic algorithm (GA) evolved neural network in estimating roller compacted concrete pavement (RCCP) characteristics including flexural and compressive strength of RCC and also energy absorbency of mixes with different compositions. A real coded GA was implemented as training algorithm of feed forward neural network to simulate the models. The genetic operators were carefully selected to optimize the neural network, avoiding premature convergence and permutation problems. To evaluate the performance of the genetic algorithm neural network model, Nash-Sutcliffe efficiency criterion was employed and also utilized as fitness function for genetic algorithm which is a different approach for fitting in this area. The results showed that the GA-based neural network model gives a superior modeling. The well-trained neural network can be used as a useful tool for modeling RCC specifications.

Evaluating the robustness of conceptual rainfall-runoff models under climate variability in northern Tunisia

To evaluate the impact of climate change on water resources at the catchment scale, not only future projections of climate are necessary but also robust rainfall-runoff models that must be fairly reliable under changing climate conditions. The aim of this study was thus to assess the robustness of three conceptual rainfall-runoff models (GR4j, HBV and IHACRES) on five basins in northern Tunisia under long-term climate variability, in the light of available future climate scenarios for this region. The robustness of the models was evaluated using a differential split sample test based on a climate classification of the observation period that simultaneously accounted for precipitation and temperature conditions. The study catchments include the main hydrographical basins in northern Tunisia, which produce most of the surface water resources in the country. A 30-year period (1970–2000) was used to capture a wide range of hydro-climatic conditions. The calibration was based on the Kling-Gupta Efficiency (KGE) criterion, while model transferability was evaluated based on the Nash-Sutcliffe efficiency criterion and volume error. The three hydrological models were shown to behave similarly under climate variability. The models simulated the runoff pattern better when transferred to wetter and colder conditions than to drier and warmer ones. It was shown that their robustness became unacceptable when climate conditions involved a decrease of more than 25% in annual precipitation and an increase of more than +1.75°C in annual mean temperatures. The reduction in model robustness may be partly due to the climate dependence of some parameters. When compared to precipitation and temperature projections in the region, the limits of transferability obtained in this study are generally respected for short and middle term. For long term projections under the most pessimistic emission gas scenarios, the limits of transferability are generally not respected, which may hamper the use of conceptual models for hydrological projections in northern Tunisia.

Hydrological model parameterization using NDVI values to account for the effects of land cover change on the rainfall–runoff response

Abstract Classic rainfall–runoff models usually use historical data to estimate model parameters and mean values of parameters are considered for predictions. However, due to climate changes and human effects, model parameters change temporally. To overcome this problem, normalized difference vegetation index (NDVI) derived from remotely sensed data was used in this study to investigate the effect of land cover variations on hydrological response of watersheds using a conceptual rainfall–runoff model. The study area consists of two sub-watersheds (Hervi and Lighvan) with varied land cover conditions. Obtained results show that the one-parameter model generates runoff forecasts with acceptable level of the considered criteria. Remote sensing data were employed to relate land cover properties of the watershed to the model parameter. While a power form of the regression equation could be best fitted to the parameter values using available images of Hervi sub-watershed, for the Lighvan sub-watershed the fitted equation shows somewhat lower correlation due to higher fluctuations of the model parameter. The average values of the Nash–Sutcliffe efficiency criterion of the model were obtained as 0.87 and 0.55, respectively, for Hervi and Lighvan sub-watersheds. Applying this methodology, the model's parameters might be determined using temporal NDVI values.

Improving runoff prediction using agronomical information in a cropped, loess covered catchment

Predicting runoff hot spots and hot-moments within a headwater crop-catchment is of the utmost importance to reduce adverse effects on aquatic ecosystems by adapting land use management to control runoff. Reliable predictions of runoff patterns during a crop growing season remain challenging. This is mainly due to the large spatial and temporal variations of topsoil hydraulic properties controlled by complex interactions between weather, growing vegetation, and cropping operations. This interaction can significantly modify runoff patterns and few process-based models can integrate this evolution of topsoil properties during a crop growing season at the catchment scale. Therefore, the purpose of this study was to better constrain the event-based hydrological model Limburg Soil Erosion Model by incorporating temporal constraints for input topsoil properties during a crop growing season (LISEM). The results of the temporal constraint strategy (TCS) were compared with a classical event per event calibration strategy (EES) using multi-scale runoff information (from plot to catchment). The EES and TCS approaches were applied in a loess catchment of 47 ha located 30 km northeast of Strasbourg (Alsace, France). A slight decrease of the Nash–Sutcliffe efficiency criterion on runoff discharge for TCS compared to EES was counterbalanced by a clear improvement of the spatial runoff patterns within the catchment. This study showed that limited agronomical and climatic information added during the calibration step improved the spatial runoff predictions of an event-based model. Reliable prediction of runoff source, connectivity, and dynamics can then be derived and discussed with stakeholders to identify runoff hot spots and hot-moments for subsequent land use and crop management modifications.

Estimating catchment-scale groundwater dynamics from recession analysis – enhanced constraining of hydrological models

Abstract. In this study, we propose a new formulation of subsurface water storage dynamics for use in rainfall–runoff models. Under the assumption of a strong relationship between storage and runoff, the temporal distribution of catchment-scale storage is considered to have the same shape as the distribution of observed recessions (measured as the difference between the log of runoff values). The mean subsurface storage is estimated as the storage at steady state, where moisture input equals the mean annual runoff. An important contribution of the new formulation is that its parameters are derived directly from observed recession data and the mean annual runoff. The parameters are hence estimated prior to model calibration against runoff. The new storage routine is implemented in the parameter parsimonious distance distribution dynamics (DDD) model and has been tested for 73 catchments in Norway of varying size, mean elevation and landscape type. Runoff simulations for the 73 catchments from two model structures (DDD with calibrated subsurface storage and DDD with the new estimated subsurface storage) were compared. Little loss in precision of runoff simulations was found using the new estimated storage routine. For the 73 catchments, an average of the Nash–Sutcliffe efficiency criterion of 0.73 was obtained using the new estimated storage routine compared with 0.75 using calibrated storage routine. The average Kling–Gupta efficiency criterion was 0.80 and 0.81 for the new and old storage routine, respectively. Runoff recessions are more realistically modelled using the new approach since the root mean square error between the mean of observed and simulated recession characteristics was reduced by almost 50 % using the new storage routine. The parameters of the proposed storage routine are found to be significantly correlated to catchment characteristics, which is potentially useful for predictions in ungauged basins.

Quantifying shallow subsurface water and heat dynamics using coupled hydrological-thermal-geophysical inversion

Abstract. Improving our ability to estimate the parameters that control water and heat fluxes in the shallow subsurface is particularly important due to their strong control on recharge, evaporation and biogeochemical processes. The objectives of this study are to develop and test a new inversion scheme to simultaneously estimate subsurface hydrological, thermal and petrophysical parameters using hydrological, thermal and electrical resistivity tomography (ERT) data. The inversion scheme – which is based on a nonisothermal, multiphase hydrological model – provides the desired subsurface property estimates in high spatiotemporal resolution. A particularly novel aspect of the inversion scheme is the explicit incorporation of the dependence of the subsurface electrical resistivity on both moisture and temperature. The scheme was applied to synthetic case studies, as well as to real datasets that were autonomously collected at a biogeochemical field study site in Rifle, Colorado. At the Rifle site, the coupled hydrological-thermal-geophysical inversion approach well predicted the matric potential, temperature and apparent resistivity with the Nash–Sutcliffe efficiency criterion greater than 0.92. Synthetic studies found that neglecting the subsurface temperature variability, and its effect on the electrical resistivity in the hydrogeophysical inversion, may lead to an incorrect estimation of the hydrological parameters. The approach is expected to be especially useful for the increasing number of studies that are taking advantage of autonomously collected ERT and soil measurements to explore complex terrestrial system dynamics.

Parameter Estimation of a Distributed Hydrological Model Using a Genetic Algorithm

Water is a vital resource, and can also sometimes be a destructive force. As such, it is important to manage this resource. The prediction of stream flows is an important component of this management. Hydrological models are very useful in accomplishing this task. The objective of this study is to develop and apply an optimization method useful for calibrating a deterministic model of the daily flows of the Miramichi River watershed (New Brunswick). The model used is the CEQUEAU model. The model is calibrated by applying a genetic algorithm. The Nash-Sutcliffe efficiency criterion, modified to penalize physically unrealistic results, was used as the objective function. The model was calibrated using flow data (1975-2000) from a gauging station on the Southwest Miramichi River (catchment area of 5050 km2), obtaining a Nash-Sutcliffe criterion of 0.83. Model validation was performed using flow data (2001-2009) from the same station (Nash-Sutcliffe criterion value of 0.80). This suggests that the model calibration is sufficiently robust to be used for future predictions. A second model validation was performed using data from three other measuring stations on the same watershed. The model performed well in all three additional locations (Nash-Sutcliffe criterion values of 0.77, 0.76 and 0.74), but was performing less well when applied to smaller sub-basins. Nonetheless, the relatively strong performance of the model suggests that it could be used to predict flows anywhere in the watershed, but caution is suggested for applications in small sub-basins. The performance of the CEQUEAU model was also compared to a simple benchmark model (average of each calendar day). A sensitivity analysis was also performed.