Journal of Hydrologic Engineering

Surface water resources management relies on the river flow in the region which, in turn, depends on numerous factors, resulting in the complexity of predicting the runoff. In this study, data-driven methods have been used to identify the relation between the river flow and regional climatic parameters and the teleconnection indexes. To achieve this, three nonlinear models of artificial neural networks, namely, generalized feedforward neural networks (GFNNs), Jordan–Elman network (JEN), and k-nearest neighbor (KNN), have been used to model monthly flow in a period of 30 years. The sensitivity analysis of input data was done using gamma test, and upon determination of the effective input parameters, modeling was done in four scenarios. The results reveal that among data-driven models, Jordan–Elman neural networks, compared with the other two models, show higher capabilities. On average, the JEN model, in comparison with the KNN and GFNN models, shows 23.4 and 23.04% less errors, respectively. Applying climatic parameters with remote sources, for instance, North Atlantic Oscillation and East Pacific/North Pacific, can enhance the efficiency of GFNN and JEN models.

Existing entropy-based methodologies for deriving two-dimensional (2D) velocity distributions in open-channel flow are based on the extension of one-dimensional (1D) formulations. Such extensions contain too many parameters and require either experimental calibration or that velocity be known at several points. This information is not always available or does not apply under some particular conditions. Following the approach developed for a rectangular cross section, this paper develops an entropy-based method for deriving a 2D velocity distribution for a generic-shaped cross section. The derived distribution is parsimonious and can be applied to natural watercourses, which usually do not have a regular cross section. The model is consistent with previous results because the velocity distribution for a simple domain can be derived as a particular case of a more complex domain. The method was validated against velocity data collected in almost 200 data series along three rivers. The velocities calculated by the new method were found to be in good agreement with measured data. The results were also compared with two methods available in the literature. One of these methods showed better performance than the proposed model, although it requires much more information.

The use of hydrological models to undertake climate change research raises the question of their robustness for use in a long-term projection when calibrated over a period in the recent past. The aim of this study was to assess the temporal transposability of the soil and water assessment tool (SWAT) model in the face of changes in climate and land use information. The model is set up on the Garonne River watershed in southwest France and two different tests were conducted: (1) two five-year periods were selected from the past 50 years: a dry and warm (DW) period and a wet and cold (WC) period. In this case, the model was calibrated for each period, and then each was cross-validated with the other; (2) the same process was undertaken using two sets of land use data from 1990 and 2006. The model was calibrated for the same time period (2000–2010), and each land use data set was cross-validated with the other. All four of the calibration/validation tests showed that the model was able to simulate the monthly stream effectively. Most of the 20 sites spread over the watershed have Nash-Sutcliffe criterion (NSe) and Nash-Sutcliff criterion calculated on the squared roots values of the discharge (NSeSqrt) above 0.75. The robustness of the SWAT model when confronted with changes to the climate or land use data was also very satisfactory: a variation of less than $\pm 0.10$ is induced for 14 sites and both metrics. Sites where simulations appeared more sensitive to those changes were either impacted by snowfall/melt processes during the winter season or by the low-flow support provided by hydraulic structure during the summer season in order to sustain irrigation. A change in land-use information caused even less instability: all 20 sites showed a variation in performances of less than $\pm 0.10$ for both metrics. It was concluded from this study that the SWAT model could be used over the Garonne River watershed in a climate change condition.

This study applies the copula-based Markov process to model water quality time series. The bivariate copula is applied to investigate the first-order Markov processes. The D-Vine copula is applied to investigate the more complicated higher-order ($k\ge 2$) Markov processes. The Value-at-Risk (VaR), computed using the best-fitted copula-based Markov process, is applied for the risk analysis. Using water quality time series at the Snohomish River watershed (Washington) and the Chattahoochee River watershed (Georgia), the results show that the copula-based Markov processes (1) are able to properly model the temporal dependence for dissolved oxygen (DO) series [i.e., forecast root-mean-square error (RMSE) $<1\mathrm{mg}/\mathrm{L}$ at both watersheds] and temperature (T) series (i.e., forecast RMSE = 1.5°C at the Chattahoochee watershed); and (2) can only predict the overall trend for nitrate and conductivity series, due to the fact that these two series also depend heavily on other factors (e.g., runoff). Overall, the study indicates that the copula-based Markov process may be an efficient tool in the assessment of water quality and the associated risks with the following advantages: (1) constructing the transitional probability explicitly and properly; (2) studying the temporal dependence independently from the marginal distributions; (3) avoiding the strict assumptions of the classic time series modeling approach (e.g., the time series belonging to the Gaussian process); and (4) providing a reasonable risk measure through the VaR.

A two-dimensional hydrodynamic model, named CHRE2D, is developed for simulating surface runoff in watershed. The model is based on the solution to shallow water equations (SWEs) using the Godunov-type finite volume method. The novelty of the model is a stable and accurate algorithm to calculate flow velocities in cells of very shallow flow depth. Stability analysis shows that the friction terms in the momentum equations are stiff when flow depth is very small (e.g., ${10}^{-3}\mathrm{m}$). In this study, the velocities in those cells are calculated by the diffusion wave approximation (DWA). This leads to a perfect balance between the surface gradient and the friction terms and prevents the occurrence of stiffness. Furthermore, the model possesses the well-balance property for both overland and channel flow. This allows the use of the original digital elevation model (DEM) data without removing topographic depressions. The model has been tested and verified by three experimental and one field cases including the rapidly varied dam-break flow and the sheet flood in arid regions. Simulated flow depth, velocity, and discharge hydrographs were compared with available measurements in two experimental testing cases. Since the resolution of the DEM significantly affects the modeling results of flow depth for field applications, only the simulated hydrograph was compared with the observation. The simulated results demonstrate that the model is highly capable of simulating both hydrodynamic and hydrological flow processes.

Estimation of peak outflow resulting from dam failure is of paramount importance for flood risk analysis. This paper presents a new hybrid clustering model based on artificial neural networks and genetic algorithms (ANN-GA) for improving predictions of peak outflow from breached embankment dams. The input layer of the ANN-based model comprises height and volume of water behind the breach at failure time plus a new parameter of cluster number. The cluster number is obtained from partitioning the input data set using the $k$-means clustering technique. The model is demonstrated using the data samples collected from the literature and compared with three benchmark models by using a cross-validation method. The benchmark models consist of a conventional regression model and two ANN models trained by nonlinear techniques. Results indicate that the suggested model is able to estimate the peak outflows more accurately, especially for big flood events. The best prediction for the current database was obtained from a five-clustered ANN-GA model. The uncertainty analysis shows the five-clustered ANN-GA model has the lowest prediction error and the smallest uncertainty.

To assess spring vulnerability level, the vulnerability estimator for spring protection areas (VESPA) index has been proposed. The VESPA index method is based on the joint use of discharge, electrical conductivity, and temperature data, measured at the spring location. In this paper the VESPA index method has been examined through the analysis of observed time series with respect to the following aspects: (1) the time extension of the data set, (2) the use of averaged data, and (3) the number of missing data. For this purpose four springs fed by different geological systems were considered. The outcomes of the statistical analysis show that the VESPA index is a reliable way to identify higher vulnerability levels while some uncertainty occurs in the identification of low and medium vulnerability levels.

Climate change impacts on streamflow and groundwater recharge were evaluated for two nearby watersheds, each with different development pressures and land-use patterns. Dynamically downscaled daily climate projections from a multimodel ensemble were used to drive rainfall-runoff simulations with a distributed hydrologic model. A total of 13 general circulation models (GCMs) with three different emission scenarios were used to analyze the potential impacts of streamflow and recharge in two basins in southern New Jersey. Both basins showed similar responses of streamflow to climate change with increased low flows (95% exceedance). Average annual recharge in the Batsto River basin increased by 5% by 2065, but recharge in the Maurice River basin did not increase based on simulation results. The response of groundwater recharge recharge in the summer was similar in both basins, with simulations indicating a 10–20% increase. It seems likely that different patterns of land use change, population growth, and subsequent changes in water demand will result in different responses to climate change in these basins despite their proximity. In particular, development in the Maurice basin, the need for new stormwater infrastructure, and upgrades to existing infrastructure suggest taking steps towards maximizing retention and groundwater recharge.

Daily evaporative loss ($E_p$) forecasting models are decisive tools with potential applications in hydrology, the design of water systems, urban water assessments, and irrigation management. This paper performs a case study for forecasting daily $E_p$ for Brisbane city using least-square support-vector regression (LSSVR). A limited set of predictor data with solar radiation and exposure, maximum/minimum temperatures, wind speed, and precipitation (March 1, 2014 to March 31, 2015) is adopted to develop the predictive model. The results are evaluated with Gaussian process regression (GPR), minimax probability machine regression (MPMR), and genetic programming (GP) models. In the testing phase, a correlation coefficient of 0.895 is attained between the observed and forecasted $E_p$ by LSSVR that contrasted 0.875 (GPR), 0.864 (MPMR), and 0.628 (GP). A sensitivity test of predictor variables shows that approximately 28.5% of features are extracted from solar radiation data with 18.1% (wind speed), 16.6% (precipitation), and 10–15% (minimum and maximum temperature). The root-mean square error for LSSVR is lower than the GPR, MPMR, and GP models by 16.2, 11.4, and 79.4%, and the cumulative frequency of forecasting error attained for LSSVR is the highest within the smallest error band. The results confirm the better utility of LSSVR in relation to GP, GPR, and MPMR models for forecasting daily evaporative loss.

The groundwater systems of small coral islands are highly dynamic, fluctuating in response to variable rainfall patterns, sea-level rise, and human management of water resources. Using a three-dimensional modeling approach, this study investigates the behavior of the freshwater lens in relation to major geohydrologic factors, including annual recharge rate, hydraulic conductivity ($K$) of the Holocene aquifer, depth to the contact between the Holocene and Pleistocene-age aquifers, and annual rate of coconut palm transpiration. Multiple islands within the Federated States of Micronesia are assessed, and therefore, the relations to island width, island surface area, and island orientation on the atoll (leeward versus windward) also are investigated. Results indicate that Holocene $K$ and the rate of annual recharge are most influential on lens behavior. In general, a hydraulic conductivity of $25\mathrm{m}/\mathrm{d}\mathrm{a}\mathrm{y}$ ($0.03\mathrm{cm}/\mathrm{s}$) and between 100 and $200\mathrm{m}/\mathrm{d}\mathrm{a}\mathrm{y}$ ($0.12–0.23\mathrm{cm}/\mathrm{s}$) is found for leeward and windward islands, respectively. Results can be used to guide freshwater resource management by estimating extractable volume of a given atoll island aquifer if island surface area, atoll orientation, and rainfall are known.

A methodology is presented for combining the conventional Natural Resources Conservation Service (NRCS) 24-h rainfall distributions with Horton and Green–Ampt infiltration processes to estimate the curve number (CN) and initial abstraction ($I_a$) parameters of the NRCS curve number model. It is shown that the rainfall distribution, rainfall depth, and infiltration characteristics all affect CN and $I_a$, and the influence of these factors on CN and $I_a$ under selected conditions are quantified. The results presented in this paper are particularly applicable to cases in which the dominant abstraction mechanism is infiltration.