Journal of Hydrologic Engineering

In flood frequency analysis (FFA), the adequate choice of distribution to fit data is a major problem. The three-parameter lognormal (LN3) distribution has an intermediate tail behavior between the distributions of the Class C (regularly varying distributions) and those of the Class D (subexponential distributions). HYFRAN software performs a complete frequency analysis for approximately twenty distributions often used in hydrology including the LN3 and distributions of Classes C and D. A decision support system (DSS) was added to the HYFRAN software to become the HYFRAN-PLUS software. It allows distinguishing between the distributions of Classes C and D. The objective of the present study is to discriminate between the LN3 distribution and that of Class C of regularly varying distributions (heavier tail) and D of subexponential distributions (lighter tail) and then to improve the current version of the DSS. The power of several normality tests is evaluated for log-transformed variates using a Monte Carlo approach for different alternative hypothesis. The Jarque-Bera test has been found the most powerful on transformed data. Results show a strong dependence between the values of the parameters and the power of the test as well as the quantile estimation errors. Results lead to the development of a LN3 goodness-of-fit procedure, based on the coefficient of variation, the coefficient of skewness and the Jarque-Bera normality test. This procedure will be added to the Decision Support System of the HYFRAN-PLUS software.

Numerical experiments were performed to investigate saltwater intrusion and recirculation of seawater at a coastal boundary. A field-scale two-dimensional cross section was simulated in which freshwater inflow occurred at an upgradient boundary, and saltwater inflow, freshwater outflow, and recirculated seawater outflow occurred at a downgradient boundary. The upgradient boundary is a specified-flux boundary with a zero (i.e., freshwater) concentration, and the downgradient boundary is a specified-head boundary with a specified (i.e., saltwater) concentration. This problem was solved numerically using SEAWAT for two conditions; i.e., first for the uncoupled condition in which the fluid density is constant, and thus, the flow and transport equations are uncoupled in a constant-density flowfield, and then for the coupled condition in which the fluid density is a function of the concentration of total dissolved solids, and thus, the flow and transport equations are coupled in a variable-density flowfield. Equivalent freshwater heads are specified at the downstream boundary for both conditions to account for density differences between freshwater and saltwater at the downstream boundary. For both conditions, it was determined that saltwater intrusion and seawater recirculation are decreased significantly as the dimensionless ratio of the freshwater inflow relative to the density-driven buoyancy flux ($a_z$) is increased. However, the extent of saltwater intrusion is less and the degree of seawater recirculation is greater for the uncoupled condition than for the coupled condition at smaller values of $a_z$, indicating that significant differences can occur between uncoupled and coupled simulations. For the experiments conducted in this investigation, at smaller values of $a_z\sim 0.1$, where the density-driven buoyancy flux dominates the freshwater advective flux, the extent of saltwater intrusion is less and the degree of saltwater recirculation and percent of recirculated seawater are greater for the uncoupled condition than the coupled condition. At larger values of $a_z\sim 10.0$, where the freshwater advective flux dominates the buoyancy flux, the extent of saltwater intrusion for the uncoupled condition becomes equal to the extent of saltwater intrusion for the coupled condition, and the degree of saltwater recirculation and percent of recirculated seawater for the uncoupled condition asymptotically approach the corresponding values for the coupled condition.

An easy-to-use and physically based analytical probabilistic model is developed to evaluate the long-term average hydrologic performance of green roofs. The probabilistic models of local rainfall characteristics are introduced first, the hydrologic and hydraulic processes occurring on and inside a green roof system are then described mathematically, and the closed-form mathematical expressions depicting the storm-water management performance of a green roof system are finally obtained using the derived probability distribution theory. Simplifying assumptions are made to mathematically describe the hydrologic and hydraulic processes. The validity of these assumptions and the overall probabilistic approach is demonstrated by comparing its outcomes with results from a series of continuous simulations using long-term rainfall data from Detroit, Michigan and observations from a real case study in Portland, Oregon.

To reproduce historical stream flows, climate and land-use change studies require watershed models with physically based parameters, rather than empirical models that are simply calibrated. With this in mind, soil moisture accounting and the temperature index (degree-day) snowmelt models embodied in the Hydrologic Engineering Center’s hydrologic modeling system (HEC-HMS) are applied to three Great Lakes watersheds—Kalamazoo, Maumee, and St. Louis—with different climatic and land-use characteristics. Watershed and subwatershed models are calibrated and validated on a daily time step using gauge precipitation measurements, observed snow water equivalent data, and physically based parameters estimated using geospatial databases. Results are compared with area-scaled outputs from the National Oceanic and Atmospheric Administration (NOAA) large basin runoff model (LBRM) for historical conditions. The results show modest improvements resulting from the increased spatial resolution of the HEC-HMS models, in addition to the benefits of the more process-based snow algorithm in HEC-HMS, particularly for the snow-dominated St. Louis watershed. However, both LBRM and HEC-HMS models had difficulty reproducing peaks in late winter and early spring runoff, and discrepancies could not be attributed to any systematic errors in the snowmelt models.

Evaporation in reservoirs plays a prominent role in water resources planning, operation, and management because a considerable amount of water is lost through evaporation, especially in large reservoirs. Estimating evaporation from surface water usually requires ample data that are not easily measurable. At present, in India, reservoir evaporation is estimated from the pan evaporation and average water spread area. Because reservoir evaporation exhibits a nonlinear relationship with the reservoir storage and other meteorological parameters, accurate prediction of evaporation by the conventional method is a cumbersome process. Recently evolved data-driven techniques will excel in nonlinear processes modeling. In this study, reservoir evaporation is predicted using three different data-driven techniques—artificial neural network (ANN), model tree (MT), and genetic programming (GP)—by time-series modeling. The daily Koyna reservoir evaporation prediction models are developed using 49 years of daily evaporation data. Approximately 70% of the data set is used for training the model, and the remaining 30% is used for testing. From this study, all of the data-driven techniques predicted the reservoir evaporation very accurately, with better performance and a correlation of approximately 0.99. This shows that if the input data series exhibits a good pattern with less noise, the data-driven techniques result in better performances. Among the data-driven techniques used in this study, GP predicts the reservoir evaporation slightly better than ANN and MT models.

Knowledge of pore-water pressure responses to rainfall is vital in slope failure and slope hydrological studies. The performance of four artificial neural network (ANN) training algorithms was evaluated to identify the training algorithm appropriate for modeling the dynamics of soil pore-water pressure responses to rainfall patterns using multilayer perceptron (MLP) ANN. The ANN model comprised eight neurons in the input layer, four neurons in the hidden layer, and a single neuron in the output layer representing an $8\text{-}4\text{-}1$ ANN architecture. The training algorithms evaluated include the gradient descent, gradient descent with momentum, scaled conjugate gradient, and Levenberg-Marquardt (LM). The performance of the training algorithms was evaluated using standard performance evaluation measures—root mean square error, coefficient of efficiency, and the time and number of epochs required to reach a predefined accuracy. It was found that all the training algorithms could be used in the prediction of pore-water pressures. However, the LM algorithm required the least time and epochs for training the network and gave the minimum error during both training and testing. The LM training algorithm is therefore proposed as an ideal and fast training algorithm for modeling the dynamics of soil pore-water pressure changes in response to rainfall patterns.

Reliable hydrometeorological forecasts are necessary to help reduce damages caused by extreme events. Coupling atmospheric and hydrological models can help improve those forecasts; therefore, the main objective of this work is to propose, apply, and evaluate a one-way coupled system involving an atmospheric and a hydrological model. The models were applied to a watershed in the Serra do Mar region, São Paulo State, Brazil. Because of the complex terrain and native tropical forest (Mata Atlântica), hydrometeorological modeling in this region requires specific studies and adequate models. The coupling was carried out between the atmospheric regional Eta model and the distributed hydrology soil vegetation model. The primary challenge in coupling hydrological and atmospheric models is the large difference in spatial and temporal scales in which these models operate. This study proposes a downscaling methodology to increase the spatial resolution of precipitation and temperature predicted by the atmospheric model from 10 to 2 km. The coupling was tested in three precipitation events that accumulated large amounts of precipitation. The results show that the quality of the predicted streamflows depends primarily on the quality of the predicted precipitation that is input to the hydrological model. The downscaling methodology applied does not significantly influence quality of the simulated and predicted basin outflow.

This paper proposes deriving regional intensity-duration-frequency (IDF) curves for Edmonton, Canada, based on the scaling property of precipitation data using ensemble empirical mode decomposition (EEMD). Selected sets of annual maximum precipitation data were decomposed by the EEMD to intrinsic mode functions (IMFs), and the scaling property was investigated. Next, representative scale exponents were extracted. The results show that quantiles derived from general extreme value (GEV) probability distribution (PD) with parameters derived by the probability-weighted moment (PWM) are more accurate than those derived from the extreme value type I (EVI) PD with parameters derived by the method of moment (MOM), whose underestimation of rainfall intensity becomes obvious for high return period (greater than 25 years) and short duration (less than 1 h). The results also show that for Edmonton, generally three of four IMFs of the precipitation data showed a simple scaling property, and regional IDF curves derived from the scaling IDF and EEMD approach predict accurate storm intensities for rain-gauging sites at both the calibration and validation stages, but there could be errors associated with predicted storms of high return periods (100 year).

Protecting and improving water resources play an important role for meeting the future demand for drinking water and preventing flood events. Increasing population tends to cause a shift from rural to urban areas, which results in the alteration of concentration time, water quantity, and flow rate. Because of these changes, specific solutions need to be investigated to prevent flood and overflow. For this purpose, calibrated hydrodynamic models should be developed to understand the hydrodynamic behavior of the catchment. Sazlıdere Watershed, which provides an average of $55\mathrm{million}{\mathrm{m}}^3$ of water per year, is one of the most important watersheds of Istanbul, Turkey. There is a tremendous population increase in the area, and moderate flood events are frequently observed. Therefore, in this study, a calibrated hydrodynamic model for Sazlıdere Watershed was developed by using the Environmental Protection Agency Storm Water Management Model (EPA SWMM). The model is calibrated by using the rainfall and flow data obtained from the stations we set in the area. The most sensitive parameters of the hydrodynamic model are determined to be Manning’s roughness coefficient for both channels and subcatchments followed by the infiltration parameters, i.e., hydraulic conductivity and the initial soil-moisture condition. The flow developed on the watershed was simulated both under typical rainfall events and under extreme cases, namely, high urbanization and storm events with a high return period. It is observed that even under typical storms, floods developed on the watershed; and finally, several flood control options are suggested.

Decreasing annual maximum flood peaks on the rivers and streams in the Wisconsin driftless area have been reported in recent studies. Various explanations have been offered, generally suggesting different episodes or change points separating the early periods of higher peak flows from the more recent lower peak flows. The present research used two statistical tests to detect a change point in annual flood peaks at Freeport on the Pecatonica River for the period 1914–2008. Both tests indicated that the most significant change occurred in 1954. Next, to find an explanation for the decreasing peaks, this research carried out a seasonal analysis of flood timing. The decrease in winter flood peaks was partly explained by the decrease in snow depth and the increase in winter temperature, providing less favorable conditions for winter flooding. In turn, the decrease in winter peak flows made once smaller summer peak flows more dominant in recent years, causing the shift in flood timing. Similar analysis showed a significant degree of resemblance between the Pecatonica River and several streams in its vicinity.

The Wakavali region of the Maharashtra state in India is characterized by high-intensity rainfall over a short duration. Precipitation in Wakavali occurs mainly from May to October. Rain gauge charts of self-recording rain gauge stations were used for rainfall characteristics analysis. Intensity of storms were recorded along with the depth of individual storms. It was found that the variation in magnitude of total kinetic energy was more attributable to unit increase in magnitude of rainfall depth than unit increase in rainfall intensity by keeping other variables constant. Values of 30-min maximum rainfall intensity ($I_{30}$) were found to be very high, reaching up to $140\mathrm{mm}/\mathrm{h}$. Most of higher value of $I_{30}$ occurred in June and early July. Rainfall intensity of $8\mathrm{mm}/\mathrm{h}$ was treated as threshold rainfall intensity for the Wakavali region, above which the erosion process begins. There was significant correlation between erosivity indexes and rainfall parameters, namely, $I_{30}$ and daily precipitation for the threshold rainfall intensity of $8\mathrm{mm}/\mathrm{h}$ out of four sets of rainfall intensity. Annual average erosivity index for threshold rainfall intensity $8\mathrm{mm}/\mathrm{h}$ set was found as $13,460\mathrm{MJ}·\mathrm{mm}/\mathrm{ha}·\mathrm{h}$. Probability analysis of annual and monthly (one day maximum) rainfall at Wakavali reveals that July has the maximum precipitation in single day, followed by June, August, September, and October. From the precipitation concentration index, it was found that relative distribution of rainfall was a highly influencing parameter.

Green roofs and other urban green spaces can provide a variety of valuable benefits linked to evaporative processes, including storm-water management, reduction of urban heat island, and carbon sequestration. Accurate and representative estimation of urban evapotranspiration (ET) is a necessary tool for predicting such benefits. However, many common ET estimation procedures were developed for agricultural applications, and thus carry inherent assumptions that may not be applicable to urban green spaces, including green roofs. The objective of this paper is to evaluate the performance of two combination methods for the prediction of ET from a green roof. Two ET estimation methodologies were compared, using on-site and regionally available data sets for daily time steps, to weighing lysimeter measurements of actual ET at a green roof site in the Bronx, New York. Regionally available estimates of potential ET did not accurately predict lysimeter measured actual ET on 30 nonconsecutive, non-water-limited days in months from September through December. Over the same period, the ASCE Standardized Reference Evapotranspiration Equation performed well in predicting actual ET with an RMSD of only $0.03\mathrm{mm}{\mathrm{d}}^{-1}$. Additionally, the ET equation for short reference types, using on-site climatic data and coupled with a variation of the Thornthwaite-Mather approximation, which accounts for variable media moisture conditions, gave reasonable predictions of actual evapotranspiration for 89 days analyzed (representing months from June through January) with an aggregate underestimation of 10.1%. However, this method was highly sensitive to input parameters, specifically media field capacity. Further on-site data collection is necessary to fully evaluate the performance of the equations over different seasons at this location, and monitoring of supplementary urban green spaces and green infrastructure sites will also lend further insights regarding urban evapotranspiration.

Recently, a simple algebraic formula, equivalent in accuracy to the Penman equation, was derived for computing evapotranspiration from readily available measured data. The derivation of the formula was based on simplifications made to the standardized form of the Penman equation. In this paper, a weighted coefficient is introduced that accounts for different impacts of the aerodynamic component at two different ranges of relative humidity. The new formula for estimating reference evapotranspiration (${ET}_0$) is obtained by calibration using meteorological data from the CLIMWAT global database. The performance of the new derived formula was tested under various climatic conditions using daily data for 12 weather stations obtained from the CIMIS database. For many places where reliable wind speed data are not available, another expression was also suggested. An alternative method not requiring wind data was also tested: the reduced-set food and agriculture organization Penman-Monteith method, according to the standardized procedure for estimating missing wind data. Comparison of the two methods not requiring wind data indicated that the proposed formula is a better option than the reduced-set Penman-Monteith method to estimate ${ET}_0$.

This paper describes the application of two different neural network models, the standard-back propagation (SBP) model and the radial basis neural network (RBNN) model, to predict monthly sediment yield as a function of monthly rainfall and runoff during the rainy season for a watershed area in India. Four scenarios were considered to determine the type and number of inputs for the artificial neural network (ANN) model. It was observed that in the small and forested watershed of Nagwa, the inclusion of monthly precipitation and average discharge values improved the performance of the ANN model in the estimation of monthly sediment yield. The momentum rate, number of nodes at the hidden layer, number of nodes at the prototype layer, linear coefficient, learning rule, and transfer functions were optimized based on lowest root-mean-square error and highest correlation coefficient values. The optimized parameters were used for the SBP and RBNN models. During validation periods, the RBNN model was closer to the observed values than SBP. The mean annual observed sediment yield was $3.7\mathrm{t}/\mathrm{ha}$. The mean annual simulated sediment yields were found to be 3.1 and $3.5\mathrm{t}/\mathrm{ha}$ in SBP during training and validation periods. RBNN simulated mean annual sediment yields of 3.6 and $3.5\mathrm{t}/\mathrm{ha}$ during training and validation periods. The results are indicative that the RBNN model is more appropriate for forecasting/simulating the sediment yield at a single point of interest in agricultural watersheds.