Journal of Hydrologic Engineering

Developing an efficient and accurate hydrologic forecasting model is crucial to managing water resources and flooding issues. In this study, response surface (RS) models including multiple linear regression (MLR), quadratic response surface (QRS), and nonlinear response surface (NRS) were applied to daily runoff (e.g., discharge and water level) prediction. Two catchments, one in southeast China and the other in western Canada, were used to demonstrate the applicability of the proposed models. Their performances were compared with artificial neural network (ANN) models, trained with the learning algorithms of the gradient descent with adaptive learning rate (ANN-GDA) and Levenberg-Marquardt (ANN-LM). The performances of both RS and ANN in relation to the lags used in the input data, the length of the training samples, long-term (monthly and yearly) predictions, and peak value predictions were also analyzed. The results indicate that the QRS and NRS were able to obtain equally good performance in runoff prediction, as compared with ANN-GDA and ANN-LM, but require lower computational efforts. The RS models bring practical benefits in their application to hydrologic forecasting, particularly in the cases of short-term flood forecasting (e.g., hourly) due to fast training capability, and could be considered as an alternative to ANN.

A hybrid algorithm that combines an improved genetic algorithm (GA) and the Gauss-Newton method (GNM) is proposed for the parameter estimation of the groundwater flow model. GA is capable of searching the whole finite space for nonlinear optimization problems, but it may require a large computational time and has low precision in obtaining the optimal solution. On the other hand, GNM has the advantages of a local search if its initial value is properly assigned. Using the advantages of the two methods, an improved GA is introduced to find a feasible initial solution for GNM, and afterward the global optimal solution is given by GNM. Two examples of two-dimensional (2D) and three-dimensional (3D) unsteady flow models are used to verify the stability and efficiency of the hybrid algorithm. The results demonstrate that the global solution can be found with high precision and fast convergence and that the selection criterion of the GNM for a suitable initial value is feasible. This hybrid algorithm can be applied to solve the parameter estimation of the groundwater flow model as well as other engineering optimization problems.

Time of concentration ($T_c$) is one of the most used time parameters in hydrologic analyses. As topographic slope ($S_o$) approaches zero, traditional $T_c$ estimation formulas predict large $T_c$. Based on numerical modeling and a review of relevant literature, a lower bound for slope ($S_{lb}$) of 0.1% was identified as a threshold below which traditional $T_c$ estimation formulas become unreliable and alternate methods should be considered. In this study, slopes less than $S_{lb}$ are defined as low slopes. Slopes equal to or exceeding $S_{lb}$ are defined as standard slopes where traditional $T_c$ estimation formulas are appropriate. A field study was conducted on a concrete plot with a topographic slope of 0.25% to collect rainfall and runoff data between April 2009 and March 2010 to support numerical modeling of overland flows on low-sloped planes. A quasi-two-dimensional dynamic wave model (Q2DWM) was developed for overland flow simulation and validated using published and observed data. The validated Q2DWM was used in a parametric study to generate $T_c$ data for a range of slopes that were used to develop $T_c$ regression formulas for standard slopes ($S_o\ge 0.1\%$) and low slopes ($S_o<0.1\%$).

Streamflow disaggregation from monthly to daily was performed using a relatively simple, flexible, and adaptive method. Only streamflow acts as a decision variable in this disaggregation process. To disaggregate monthly to daily flow at the target station (TS), monthly counterparts at the source station (SS) were selected on the basis of minimum error criteria, which are calculated with respect to streamflow volume within a three-month time window. Daily streamflow indexes at SS were then calculated to disaggregate monthly to daily streamflow at TS during the disaggregation process. The effectiveness of the proposed method has been demonstrated through its application at both regulated and unregulated waterways located in the northwest states, including Idaho and Wyoming. For both regulated and unregulated monthly streamflow, the proposed method well represents daily streamflow realizations similar to historical flows and preserves both mass balance and a series of statistical characteristics. However, the results also indicate that the quality of disaggregated daily streamflow varies for individual applications depending on the selection of stations, their geographic information, and data availability. The disaggregation model used in this research is transparent, user friendly, less intensive, and less time-consuming so that it can be used at any watershed without difficulty or much effort. Consequently, because development and availability of daily streamflow is important for water resources planning and management, including reservoir operation, water quality study, and environmental/ecological modeling, this research will help bridge the gap among interdisciplinary water research activities, especially for studies of the impacts of hydrologic events possibly driven by extreme weather variability and climate change.

This research presents a Bayesian artificial intelligence model averaging (BAIMA) method that incorporates multiple artificial intelligence (AI) models to estimate hydraulic conductivity and evaluate estimation uncertainties. Uncertainty in AI model outputs stems from errors in model input and nonuniqueness in selecting different AI methods. Using one single AI model tends to bias the estimation and underestimate uncertainty. The BAIMA employs a Bayesian model averaging (BMA) technique to address the issue of using one single AI model for estimation. The BAIMA estimates hydraulic conductivity by averaging the outputs of AI models according to their model weights. In this study, the model weights are determined using the Bayesian information criterion (BIC) that follows the parsimony principle. The BAIMA calculates the within-model variances to account for uncertainty propagation from input data to AI model output. Between-model variances are evaluated to account for uncertainty because of model nonuniqueness. The authors employ Takagi-Sugeno fuzzy logic (TS-FL), an artificial neural network (ANN), and neuro-fuzzy (NF) to estimate hydraulic conductivity for the Tasuj plain aquifer, Iran. The BAIMA combines three AI models and produces better fitting than individual models. Although NF was expected to be the best AI model owing to its utilization of both the TS-FL and ANN models, the NF model is nearly discarded by the parsimony principle. The TS-FL model and the ANN model show equal importance, although their hydraulic conductivity estimates are quite different. This results in significant between-model variances that are normally ignored by using one AI model.

Riverine nutrient concentrations and loads in the Midwestern United States have been identified as a major source of nutrients for the Mississippi River, causing hypoxia in the Gulf of Mexico. The Illinois River is one of the major contributors of nutrients to the Mississippi River, draining more than 40% of Illinois and parts of Indiana and Wisconsin. The main goals of this study were to calculate annual average nitrate concentration and annual total loads in the Lower Illinois River and its tributaries for 1975–2010, and to determine how the selection of different approaches affects the calculated trends. The Illinois Environmental Protection Agency (IEPA) has been collecting surface water quality data in Illinois since the early 1970s. This study used nitrite-N+nitrate-N (${\mathrm{NO}}_2+{\mathrm{NO}}_3$) samples provided by IEPA and collected from the Illinois River at Havana and Valley City, Sangamon River at Oakford, and La Moine River at Ripley for the period 1975 to 2010. Discharge data concurrent with routine water quality samples were applied to a suite of approaches to estimate daily concentration and load for the period of analysis, including the seven-parameter regression equation along with several modeling residual adjustment techniques and the weighted regression on time, discharge, and season (WRTDS) method. The calculated annual mean concentration and annual total loads for ${\mathrm{NO}}_2+{\mathrm{NO}}_3$ for the period of record were discussed in relation to different approaches.

Retrieving precipitation data from a rain gauge network is a classical and common practice in hydrology and climatology. These data represent the key input in hydrological modeling to reproduce, for example, the characteristics of a flood phenomenon. The accuracy of the model results is strongly dependent on the consistency of the monitoring network in terms of spatial scale, i.e., network density and location of the rain gauges and time resolution. In this context, several studies have been carried out to analyze how the rainfall sampling influences the estimation of total runoff volume. The aim of this paper is to use a physically based and distributed-parameter hydrologic model to investigate how the number and the spatial distribution of a rain gauge network influence the estimation of the hydrograph and its characteristics in conjunction with different spatial and temporal characteristics of rainfall forcing and different soil-type characteristics. The TIN-based real-time integrated basin simulator (tRIBS) hydrologic model was used to simulate hydrologic response at Baron Fork Basin, Oklahoma. Downscaled next-generation radar (NEXRAD) measurements were assumed to represent the true precipitation distribution over the basin. Additional precipitation fields have been derived from the interpolation of eight fictitious rain gauges randomly placed in the area. The hydrological response from tRIBS that is driven by these precipitation fields has been compared with the response of the model forced with NEXRAD precipitation. The analysis has been carried out assuming first simplified spatial distributions of soil characteristics and then the real soil-type distribution. Results have shown the dependence of the best rain gauges configuration for the estimation of runoff on the spatiotemporal characteristics of storm events and the soil-type distribution.

Current practice using probabilistic methods applied for designing hydraulic structures generally assume that extreme events are stationary. However, many studies in the past decades have shown that hydrological records exhibit some type of nonstationarity such as trends and shifts. Human intervention in river basins (e.g., urbanization), the effect of low-frequency climatic variability (e.g., Pacific Decadal Oscillation), and climate change due to increased greenhouse gasses in the atmosphere have been suggested to be the leading causes of changes in the hydrologic cycle of river basins in addition to changes in the magnitude and frequency of extreme floods and extreme sea levels. To tackle nonstationarity in hydrologic extremes, several approaches have been proposed in the literature such as frequency analysis, in which the parameters of a given model vary in accordance with time. The aim of this paper is to show that some basic concepts and methods used in designing flood-related hydraulic structures assuming a stationary world can be extended into a nonstationary framework. In particular, the concepts of return period and risk are formulated by extending the geometric distribution to allow for changing exceeding probabilities over time. Building on previous developments suggested in the statistical and climate change literature, the writers present a simple and unified framework to estimate the return period and risk for nonstationary hydrologic events along with examples and applications so that it can be accessible to a broad audience in the field. The applications demonstrate that the return period and risk estimates for nonstationary situations can be quite different than those corresponding to stationary conditions. They also suggest that the nonstationary analysis can be helpful in making an appropriate assessment of the risk of a hydraulic structure during the planned project-life.

Global climate change is expected to lead to changes in precipitation patterns and increased frequency and intensity of extreme weather events, which may produce conditions outside current design criteria for dams. This study investigated climate change effects on future safety of the Aurland hydropower dams during extreme floods. The design inflow floods for present and future climate scenarios were calculated using two approaches. Flood frequency analysis was applied to the annual maximum series from the simulated daily flows for present and future periods. Analysis of extreme precipitation and floods was performed using a hydrological model to compute the corresponding extreme flood values for the present situation and future scenarios. The outflow flood with the associated water level was calculated using a reservoir routing model linking all the Aurland reservoirs. In this paper, the authors present current and future design floods for the Aurland case and the implications of the changes in flood levels on dam safety for single reservoirs and for the reservoir system as a whole. Results from this study show that there will be a change in seasonal shift in the peak inflow flood from summer to autumn for the future scenario; from the range of results of the climate models and emission scenarios, the design inflow flood and the probable maximum flood (PMF) in the autumn is projected to increase for the future scenario.

Understanding the variability of rainfall with El Niño Southern Oscillation (ENSO) events at the watershed level is important for water resource managers to discern the pattern of water available for agricultural planning and water and soil conservation. This study was conducted to determine the responses of rainfall and vegetation to El Niño and La Niña events occurring between 2001 and 2011 in the Upper Wardha watershed in Maharashtra, India. Daily rainfall data pertaining to 14 stations in and around the study area were analyzed in conjunction with 16-day global 250-m Normalized Difference Vegetation Index (NDVI) data from the Moderate Resolution Imaging Spectroradiometer (MODIS) sensor abroad the Terra (EOS AM) and Aqua (EOS PM) satellites. Results indicate correlations between ENSO events, rainfall, and vegetation. Both rainfall and vegetation show a positive anomaly during La Niña events and a negative anomaly during El Niño events, but the effects of all events are not the same. A low volume of rainfall with the early onset of monsoon, multimodal rainfall distribution over time, and intermittent rains are the effects of El Niño on rainfall. The effect of La Niña was to produce wetter conditions in the study area. The impacts of moderate to severe ENSO events (in 2009 and 2010) can be predominantly observed on rainfall and vegetation.

This study was undertaken to discern and quantify the hydrologic functions of wetlands within the context of the Rideau River watershed. Using numerical modeling techniques (Mike11), the potential cumulative effect of the loss of wetlands on peak flood discharges and low flows at selected locations within the watershed were quantified. It was estimated that the combined effect of all wetlands (approximately 15% of the basin area) is a 10% reduction in flood peak. It was demonstrated that the flood risk would increase if wetlands were removed. The impact of wetland loss on flood risk diminishes downstream of long channels and lakes. The low flow will likely increase if wetlands are removed, although this impact diminishes rapidly downstream of long channels and lakes. This kind of site-specific technical analysis can be used for decision making and policy formulation regarding interference with and land development adjacent to wetlands.

In this paper, a distributed catchment-hydrology model and a physically based lake hydrodynamic model were used to simulate the large-scale and highly dynamic lake catchment system of Poyang Lake, in the middle reach of the Yangtze River basin, China. The simulation of the hydrodynamics of the lake is a significant extension to previous efforts to simulate Poyang Lake’s considerable variability in lake extent and flow rates. Further, the combination of the distributed catchment-hydrology model and the lake-hydrodynamic model, applied to a highly dynamic and large-scale system, is a rare attempt to develop a physically based management model of this complexity and scale. Model calibration and validation were undertaken to evaluate the model’s performance and to enhance its effectiveness in simulating catchment discharges, lake water levels, lake water surface areas, and lake flow patterns. The results showed a satisfactory agreement with field observations, with Nash-Sutcliffe efficiencies of 0.71–0.84 for catchment discharges, 0.88–0.98 for lake water levels, and 0.80 for lake outflows during the calibration period. The Nash-Sutcliffe efficiency values for the validation period, ranging from 0.62 to 0.97, were largely consistent with the calibration values. Further investigation of the results showed that the modeling approach simulated adequately the lake hydrodynamics in terms of the flow fields within the lake and the seasonal changes in the lake water surface area. The outcomes of this paper will benefit future modeling efforts by providing a tool for predicting the hydrology of Poyang Lake and its catchment under climate variability and land-use changes.

In this study, the trends in precipitation in the northwest (NW) of Iran were identified using the four different versions of the Mann-Kendall method, i.e., the conventional Mann-Kendall method (MK1); the Mann-Kendall method following the removal of the effect of significant lag-1 autocorrelation (MK2); the Mann-Kendall method after the removal of the effect of all significant autocorrelation coefficients (MK3); and the Mann-Kendall method by considering the Hurst coefficient (MK4). Identification of trends was carried out on different time scales (monthly, seasonal, and annual) using the precipitation data of 50 years from 1955 to 2004 of the sixteen stations selected from the NW region of Iran. The Theil-Sen method was used to estimate the slopes of trend lines of precipitation series. Results showed that: (1) on a monthly time scale, the statistically significant $Z$-statistics were negative for all but one (July) month; and the strongest negative (positive) precipitation trend-line slope among all the negative (positive) cases was found to be $-0.89(0.38)\mathrm{mm}/\mathrm{year}$ at Bijar (Kermanshah) station in NW Iran; (2) on a seasonal time scale, the median of trend-line slopes was found to be negative in all four seasons; the winter and spring season’s precipitation series witnessed negative trends for almost all the stations using all four different versions of the MK test; and in the summer and autumn seasons, both upward and downward trends were observed for most of the sites of NW Iran; (3) in an annual time scale, all stations had witnessed negative trends using both the MK1 and the MK4 tests. However, application of the MK4 instead of the MK1 reduced the absolute value of the $Z$-statistic for most of the time series. The strongest negative annual trend-line slope was $-4.04\mathrm{mm}/\mathrm{year}$ at Bijar station. Therefore, the observed decreases in precipitation in NW Iran in the recent half of the past century may have serious implications for water resources management under the warming climate with probably a higher rate of the population growth and the higher consumption of freshwater as a result of the rise in standards of living of the population of NW Iran.

This study proposed a method for estimating the concentration time and storage coefficient of the Clark model using the rainfall-runoff measurements. The proposed method is based on the analytically derived concentration time and storage coefficient of the Nash model along with the relationship between these two parameters. More fundamentally, the proposed method recursively searches the convergent concentration time and storage coefficient, or the number of linear reservoirs and storage coefficient of the linear reservoir of the Nash model. The proposed method was applied to the Bangrim stream gauge station of the Pyungchang River Basin, Korea. The results are summarized as follows. First, this study theoretically derived the basin storage coefficient and concentration time by using the Nash model. Especially, the relationship between the storage coefficient and concentration time could be expressed as a function of the number of linear reservoirs only. This result has a special meaning to understand the effect of basin subdivision on the concentration time and storage coefficient. Second, when estimating the concentration time and storage coefficient by using the observed rainfall-runoff event data, the most important thing to consider was that these two parameters are interrelated. Thus, these two parameters could not be determined uniquely by applying any optimization technique. In this study, the structure of the Nash model was introduced to make the parameters be more realistic. Finally, in the application to the Bangrim stream gauge station, the concentration time was estimated to be 4–15 h with an average of 10 h, and the storage coefficient was estimated to be 11–24 h with an average of 15 h. This result indicates that the proposed method could estimate physically reasonable values of the concentration time and storage coefficient of the Clark model.

Trend analysis occupy a significant role in the climate change studies for almost three decades. It is significant to try and identify monotonic trends in a given time series so as to make future predictions about the possible consequences on the urban environment, water resources, agriculture, and many other socioeconomic aspects of life. Although there are now classically accepted and frequently used trend tests in the open literature, such as Mann-Kendall trend analysis and Spearman’s rho test, they are based on some restrictive assumptions as normality, serial independence, and rather long sample sizes. Also, they search for a single monotonic trend without any specification such as low, medium, and high values, which may have different trend patterns. Many climatological records have serial dependence, and therefore, it is very helpful to provide a methodology that is not affected from such a restriction. It is the main purpose of this paper to provide simulation results and applications of an earlier innovative trend analysis methodology based on the $1:1$ (45°) line comparison of the scatter points on a Cartesian coordinate system. The plots are a result of available time series first-half values versus second half after sorting in ascending order. This method does not have any restriction, and it is applicable whether the time series is serially correlated, nonnormally distributed, or has short record length. It helps to identify trends in low, medium, and high records. The application of the methodology is provided for a set of temperature records from the Marmara region in Turkey.

MODFLOW is a three-dimensional finite-difference groundwater flow model. IHM (integrated hydrologic model) is a recent integrated model developed to simulate the interaction between surface water and groundwater, which formally couples HSPF and MODFLOW. The purpose of this research was to define a set of comparisons between IHM-MODFLOW96 (IHM-MF96) and standard MODFLOW96 (MF96) to certify that IHM-MF96 is acceptable to use with the IHM. The certification tests were necessary to determine whether model runs with modifications to the source code of MF-96, contained in the source code of IHM-MF96, would alter the head or flux results when compared to model runs with the standard source code. The tests were conducted over a range of problem sets, which are anticipated to cover the range of hydrogeologic conditions and anticipated applications for which IHM will be used in westcentral Florida. The tests results showed the MF96 implementation in IHM appears to produce virtually the same results as the standard MF96. None of the changes to MF96 as implemented in IHM had a material effect on any of the tested conditions. Based on the test results, the IHM MODFLOW package was expected to produce the same results, as would the standard MF96.

Pumping tests are used to determine the transmissivities and storage coefficients of aquifers. For nonleaky aquifers, the solution reduces to the exponential integral, which is also called the Theis well function. It is beneficial to have a simple and handy approximation for the Theis well function. A simple (minimum number of terms) and reliable approximation that is efficient, yet sufficiently accurate, is preferable. This research provides a simple and accurate approximation to the Theis well function valid for all values of its arguments. The approximation is constructed by a combination of two solutions that are valid for small and large arguments. The approximation contains unknown coefficients, which are determined by using an optimization procedure. An accurate approximation to the well function should also be able to accurately compute the derivative of the well function. The proposed approximation minimizes errors in both the well function and its derivative. The maximum relative error in the proposed approximation and its derivative are less than 0.2% and 0.22%, respectively, making it useful for routine groundwater applications.

Most groundwater equations for flow toward wells use a set of assumptions and idealizations about the aquifer–well configuration so that analytical expressions can be derived for steady-state and unsteady-state flows. In this article, the main assumption in these equations is that constant hydraulic conductivity is relaxed and instead allows radial variability. The basic question is how the hydraulic conductivity gradient affects groundwater flow. Changes in hydraulic conductivity influence groundwater flow; any local changes in the hydraulic conductivity cause local changes in hydraulic gradient and in groundwater velocity. This problem is solved using water balance equations with changes in linear radial hydraulic conductivity. Simple but more general equations for groundwater flow toward wells are derived and applied to steady-state groundwater flows in a confined aquifer. This formulation reduces to the classical Theim solution for constant hydraulic conductivity. The use of this methodology is presented for steady-state groundwater measurement from a well in the Arabian Peninsula. It is observed that constant hydraulic conductivity underestimates transmissivity, compared to the numerical example given in this article, by about 41%.