Journal of Hydrologic Engineering

Global climate model (GCM) simulations are generally used in the estimation procedures regarding future rainfall quantiles. However, the use of GCM simulations brings several problems, such as a coarse spatial resolution and an intermodel difference in the simulation results. To overcome these problems, this study proposed a Bayesian method for using the GCM simulations as new information regarding the future. A total of seven GCMs that contained past and future simulations were considered in all four different representative concentration pathway (RCP) scenarios. The hourly rainfall data of Seoul, Korea, were used as the observed data. In the results, first, the rainfall quantiles estimated using the GCM simulations were found to be much smaller than those of the observed data, and they were intermixed, depending on the GCM and RCP scenarios. For example, the observed rainfall quantile of the observed data was estimated to be 318.3 mm for the return period of 50 years, but those for the GCM simulations were estimated to lie in a wide range from the minimum 79.2 mm (NorESM1-M for the RCP 6.0) to the maximum 293.5 mm (BCC-CSM1.1 for the RCP 8.5). Second, the difference became much smaller when the Bayesian method was applied. Although a consistent trend could not be found, all the estimated increase ratios of the future rainfall quantiles were found to lie in a reasonable range, $-8.3$ to $+16.95\%$. Third, the estimated increase ratio of the future rainfall quantiles showed that the increase or decrease patterns were all intermixed depending on the GCM and RCP scenarios. A stronger RCP scenario did not lead to an increase of the rainfall quantiles. In most GCMs, the RCP 4.5 scenario was found to lead to a greater increase of the rainfall quantiles than the other scenarios. Thus, a consideration of the greatest possible number of GCM simulations, along with all four of the RCP scenarios, could be a safe way to minimize the uncertainty potential of climate change considerations.

Roadside swales or drainage ditches are low-impact development (LID) practices for stormwater treatment and control, and there is a need to quantify their infiltration performance for design and planning purposes. A roadside swale calculator has been developed where only the main design parameters that have a significant impact on the runoff volume output are required, in addition to the rainfall distribution of the study site. The inputs of the calculator are the saturated hydraulic conductivity of the soil, width of the swale, width of the road, and location’s rainfall frequency or volume percentile. The calculator and the ancillary information provided can be used to determine: (1) the total percentage of annual volume infiltrated, (2) the percentage of events entirely captured by a roadside swale in a year, and (3) the percentage of road runoff infiltrated by a roadside swale with a specific swale width-to-road width ratio and saturated hydraulic conductivity for a given rainfall depth event.

Diverse hydrologic and hydraulic models of varying complexities have been proposed in the past few decades to accurately predict the water levels and discharges along rivers. Among them, the hydrologic routing models are widely used because of their simplicity, minimal data, and computational requirements. Due to their simplified assumptions, however, they are subject to various sources of uncertainty. To reduce their predictive uncertainty and improve their operational forecast abilities, data assimilation techniques have been proposed to update the states and/or parameters of the mathematic models by integrating real-time river observations with them. However, the characterization of the model errors and the location of the sensors used for data assimilation have an important effect on the model performance. The main objective of this study was to assess the effect of sensor placement and the errors of both the model and the boundary conditions on the assimilation of flow observations in the distributed hydrologic routing models. A Muskingum-Cunge routing model was applied first to a synthetic river reach with a rectangular cross section and then to a more complex natural river, the Bacchiglione River in Italy, with varying geometry of the river cross sections. The Kalman filter was used to assimilate the flow observations. Synthetic and real-world experiments were carried out. The results showed an improved model performance after the assimilation of the flow observations (e.g., a Nash index higher than 0.9 in the synthetic river and 0.85 in the Bacchiglione River); however, the procedure was sensitive to the model error and the locations of the sensors. In particular, when the model error was larger than the boundary condition error, it was suggested to place the sensors in the lower part of the river reach to maximize the model improvement at the river outlet. On average, the model performance was improved by 14% in terms of the Nash index when the sensor was located in the upstream part of the reaches of the Bacchiglione River instead of in the downstream part. Sensors placed in the upper part of the reaches enabled the improved skills to persist for additional lead time of up to 6 h for the forecasting of the water level at the reach outlet. This study presented a method that allowed identifying the optimal locations of the sensors and thus helped to improve the flood forecasts.

Because of its relatively easy quantification in comparison with the variables of other hydrological components, streamflow is an important and primary component for modeling in a watershed study. This study attempts to enhance one of the channel routing tools, the Muskingum routing method (MRM) used in the soil and water assessment tool (SWAT) model. This study advocates its replacement by a well-tested alternative physically based model known as the variable parameter McCarthy-Muskingum (VPMM), capable of varying the routing parameters at every routing interval, and thereby accounting the nonlinear characteristics of flood wave method movement in steep, intermediate, and small slope channels and rivers. However, the routing capability of the VPMM model is subject to the limitation of the inflow hydrograph being characterized by the criterion $(1/So)\partial y/\partial x<0.5$, where $\partial y/\partial x$ denotes the slope of the longitudinal water surface gradient. A small watershed of approximately $986.9{\mathrm{km}}^2$ in the Vansadhara basin of the Odisha state in India with upstream and downstream sites at Gunupur and Kashinagar, respectively, is studied to demonstrate the routing capability of the VPMM scheme in comparison with the MRM, and variable storage routing method (VSRM) routing schemes used in the SWAT model. The routing simulations carried out using these three schemes are evaluated using measures like the Nash-Sutcliffe efficiency (NSE), coefficient of determination ($R^2$), and mean absolute percent error (MAPE). The NSE estimates for VPMM, MRM, and VSRM routing schemes for the calibration and validation period were estimated to be 0.89, 0.92 and 0.91, and 0.72, 0.71, and 0.63, respectively; similarly, the $R^2$ estimates for VPMM, MRM, and VSRM routing schemes for the calibration and validation period were estimated to be 0.89, 0.93 and 0.92, and 0.71, 0.70, and 0.63, respectively. As the performance of the VPMM model is at par with the the MRM routing module of the SWAT model and has a sound physical basis than the MRM routing scheme, it can be recommended that the VPMM routing scheme can be incorporated in the SWAT model for enhancing its routing capability. The added advantage of the VPMM scheme is that it also estimates the stage hydrograph corresponding to the routed discharge hydrograph and thereby increasing the utility of the SWAT model for sediment transport, in-stream nutrient, and operational purposes.

Drought indices are statistical tools used for monitoring the departure from normal conditions of water availability. Recently, the multivariate nature of droughts was addressed through composite indices capable of including different factors contributing to the occurrence of a drought. However, some issues (like the autocorrelation or the proper definition of the multivariate index) are still open and need to be addressed to make these indices applicable in current practice. Here, a composite agrometeorological drought index (AMDI-SA) has been introduced, accounting for meteorological and agricultural droughts, considering specifically seasonality and autocorrelation. The AMDI-SA combines, through the copula concept and the Kendall function, two drought indices [namely multivariate standardized precipitation index (MSPI) and the multivariate standardized soil moisture index (MSSI)] in a statistically consistent (normal distributed) drought indicator. Nonparametric distributions have been used for the variables of interest and the calculation of MSPI and MSSI, whereas parametric and nonparametric (empirical) copulas are used to build the AMDI-SA. A prewhitening procedure has been applied to the MSPI and MSSI to remove the autocorrelation. An application to the Urmia lake basin in Iran has been presented, drought indices compared, and their spatial variability investigated. Results showed that MSPI and MSSI are able to justify 72 and 89% of the variability throughout the year. The AMDI-SA reflects the combined effect of soil moisture and precipitation, and has a behavior in between whitened MSPI and MSSI. In addition, having no memory and being a composite index, the AMDI-SA is able to clearly detect the temporal variability of recorded droughts to a greater extent than the MSPI and MSSI.

Downscaling techniques are required to describe the linkages between global climate model (GCM) outputs at coarse grid resolutions and surface variables at suitable finer scales for climate change impact and adaptation studies. The present paper proposes an improved statistical approach to downscaling of daily maximum (Tmax) and minimum (Tmin) temperature series located at many different sites concurrently. This new approach is based on a combination of a multiple-regression model and the modeling of its stochastic component by the singular-value decomposition (SVD) technique to represent more effectively and accurately the space-time variabilities of these extreme daily temperature series. Results of an illustrative application using data from a network of 10 weather stations located in the southwest region of Quebec and southeast region of Ontario in Canada and from the available National Centers for Environmental Prediction/National Centre for Atmospheric Research (NCEP/NCAR) reanalysis data set indicated the effectiveness and the accuracy of the proposed approach. In particular, this new approach was found to be able to reproduce accurately the basic statistical properties of the Tmax and Tmin time series, including their mean, standard deviation, Tmax 90th percentile, and Tmin 10th percentile. In addition, the at-site autocorrelations, interstation correlations, and intervariable correlations of the daily Tmax and Tmin series have been accurately reproduced. Furthermore, the proposed approach was able to adequately reproduce the interannual variability of the Tmax and Tmin.

Information about soil bulk density (SBD) is important for soil-water conservation, carbon content, and nutrient reserves. The authors studied the spatial variability of SBD in gravel-mulched fields of different ages in an arid area of northwestern China. The SBDs of a new gravel-mulched field (NGM), a moderately aged gravel-mulched field (MGM), an old gravel-mulched field (OGM), and bare land (CK) ranged from 1.07 to 1.41, 1.15 to 1.41, 1.18 to 1.38, and 1.04 to $1.29\mathrm{g}/{\mathrm{cm}}^3$, respectively, with a general trend of $\mathrm{OGM}>\mathrm{MGM}>\mathrm{NGM}>\mathrm{CK}$. The coefficients of variation were less than 7.58%, indicating weak spatial variability. SBD was strongly correlated with soil-water content (SWC), indicating that the distribution of high and low SBD can be approximately estimated using SWC distribution. Kriging could effectively predict the spatial distribution of SBD in the study area with a high precision of interpolation, but estimating SBD using cokriging by measuring SWC was more accurate.

Farm ponds are constructed for different purposes in many parts of the world, including irrigation, livestock water, and so forth; however, their impacts are not fully captured in stormwater application. As part of this study, the environmental benefits of farm ponds were assessed, and their cumulative impact at the watershed level was evaluated. The following three criteria were used to assess the impact of farm ponds: (1) the aggregate capability in reducing non-point-source pollution (measured in terms of phosphorous load reduction); (2) the cumulative impact in reducing channel erosion; and (3) the support to the ecological function of the buffer areas known as resource protection areas (RPA). The investigation was carried out on 494 farm ponds located in Fairfax County, Virginia. The results showed a profound effect of these small water storage features in protecting the downstream receiving channels from erosion. It is estimated that the level of peak flow reduction provided by 70% of the farm ponds exceeds 78% for the 1-year storm, and 55% for the 2-year storm event. This means that the presence of these farm ponds significantly reduces the stresses caused by the increase in impervious areas as a result of urbanization and its impact on channel incision. The net environmental benefit of the farm ponds in terms of their pollutant load reduction capability is found to exceed $\mathrm{USD}0.58\mathrm{million}/\mathrm{ha}$ ($\mathrm{USD}0.23\mathrm{million}/\mathrm{acre}$) of pond surface area. In Fairfax County, it is estimated that a total of 2.7 t (6,000 lbs) of phosphorous load is being trapped by farm ponds on average annual basis. The role of these water bodies in supporting the ecological function of the RPA buffer was studied by first analyzing the connectivity of farm ponds and their locations following the core functions of RPA buffer features. The study found that a total loss of $1.5{\mathrm{km}}^2$ (0.61 sq mi) in the RPA buffer would be anticipated in the event that these farm ponds are drained. The outcome of this study is intended to inform decision makers on the right policy that could be integrated as part of the adopted jurisdictional regulatory framework. Such an approach could be scalable to other jurisdictions contingent upon addressing the local contextual issues of other localities.

In-depth evaluation and analysis of the error properties associated with satellite-based precipitation estimation algorithms can play an important role in the future development and improvements of these products. This study evaluates the Precipitation Estimation from Remotely Sensed Information using Artificial Neural Networks (PERSIANN) daily data set from 2000 to 2011 in 69 pixels over a semiarid basin in northwest Iran and compares it with the data set of the existing rain-gauge network. Different analytical approaches and measures are used to examine PERSIANN performance seasonally and categorically. The residuals are also decomposed into true positive (hit), false negative (miss), and false alarm (FA) estimate biases in addition to systematic and random error components. The results show seasonal variability of PERSIANN precision in rainfall detection with substantial errors during winter and summer that are associated with high rates of FA ratio (more than 60%). The value of miss and FA biases (124 and $-\mathrm{77,000}\mathrm{mm}$, respectively, within the total data set) are considerably larger than hit and total bias (27 and 74,000 mm, respectively) because these components contribute conversely and compensate each other by their opposite signs. Moreover, PERSIANN detects heavy rainfalls well with a probability of detection (POD) over 80%, but with serious biases. Generally, although the detection ability of PERSIANN improves as the rate of rainfall increases, its systematic error in simulation of the rainfall process also increases (from 5% systematic error to 90% in heavier rainfalls), leading to a low level of accuracy in the estimation of precipitation rate.

A regional frequency analysis using L-moments was performed with time series clustering approaches to identify homogeneous regions using dynamic data sets in Northern Cyprus. In this context, the conventional approach, based on station characteristics and different time series clustering approaches, classified as shape-based, feature-based, and model-based, were compared. Hierarchical Ward’s method with the correlation-based similarity measure of the feature-based approach was determined as the best method regarding the results of the jackknife validation procedure, which was performed for assessment of clustering approach uncertainty. Therefore, the cluster analysis ended up with five homogeneous subregions, and according to the goodness-of-fit measure, the Pearson Type III, generalized logistic, and generalized normal distributions were chosen as the best fit for different subregions. The accuracy of the estimated quantiles was evaluated through Monte Carlo simulations and, consequently, the quantiles for different return periods were estimated, which demonstrated spatial consistency in terms of increasing trend from the low-lying Mesaoria Plain toward the north coastal strip, including the Kyrenia Mountains and the Karpass Peninsula.

The construction of dams has significant effects on hydrologic regimes downstream. The combined effect of the Danjiangkou Reservoir and the downstream cascade reservoirs on the changes in the hydrologic regime at Huangzhuang station were investigated using the indicators of hydrologic alteration (IHA) and the range of variability approach (RVA) in the middle-lower Hanjiang River. The single operation of the Danjiangkou Reservoir had effects with an overall alteration degree of 40.6%, in which the number of reversals, and flows in March and September were highly altered. The combined operation of Wangfuzhou Reservoir changed the flows in March and August and the extreme minima from high or moderate alteration to low alteration and increased the alteration of the extreme maxima, pulse behavior, and rise rate, with an overall alteration degree of 46.7%. Furthermore, with the combined operation of Cuijiaying Reservoir, the overall alteration degree increased to 63.5%, and 56.2% of the indicators were of high alteration. These changes are mainly caused by the reservoir regulations, which have profound effects on the river aquatic environment and biological habitats.

Recently, several bivariate and multivariate methods have been used to analyze the relationship of flood peak discharges with corresponding hydrograph volumes and durations in many basins. In some flat tidal areas, such as the Taihu Basin of China, the flood peak level and the corresponding flood duration have received more attention than the flood peak discharge and other flood characteristics in flood control and flood management. In this study, avoiding the complex problem of positive-negative flows, a pragmatic approach for bivariate flood frequency analyses based on flood peak level data is proposed. Commonly used bivariate copulas and techniques for fitting such copulas are applied to analyze the correlation between flood peak levels and durations based on peak-over-threshold flood series in Pingwang and Wuzhen. The goodness-of-fit test shows that the extreme-value copulas are not suitable for these two cases. Along with the tail dependence analysis, the Student copula ($\nu =5$) is chosen as the best-fitting model. A comparison between the joint return period (years) associated with the event ($H>h_T$ or $D>d_T$) obtained through the Student copula ($\nu =5$) and other bivariate copulas demonstrates that ignoring the tail dependence results in a significant overestimate of flood risk. The joint return periods of arbitrary combinations of flood peak levels and flood durations are obtained through the final fitted copula. Furthermore, a comparison between the cumulative probability curves of flood peak levels obtained through different univariate and bivariate methods confirms the correctness of this work.

The quantity and various sources of water for irrigation are important controlling factors in the agricultural development of any area. The aim of this work is to determine an optimal cropping pattern to maximize the net annual returns, to conserve groundwater resources, and to use optimal land under cultivation for a region that has a scarcity of surface water resource and uncertain rainfall. Hence, mathematical models are developed based on data collected from government agencies, interviews of farmers in the region, and a practical survey of the region. An optimization problem is formulated based on the developed mathematical models. For this purpose, an optimization problem is formulated. Sinnar Taluka of Maharashtra state in India is the study area, in which groundwater is the only source for irrigation in this region. Hence, the Minor Irrigation Department (MID) of Nasik has planned an artificial recharge for the study area. The existing cropping pattern is studied extensively, and a new cropping pattern is suggested by solving the optimization problem using advanced optimization algorithms such as Jaya and particle swarm optimization (PSO). The PSO algorithm is widely used in water management. The Jaya algorithm is new in the area of water management and has significantly shown its competitiveness in this study. It has allocated land and water optimally in less computational time and provided higher net annual returns than PSO.

Groundwater level (GWL) is an important subsoil characteristic, closely related to geological, geographical, and hydrological conditions. In urban areas, a key influence component for GWL is river stage (RS) because of the near-river location and low rainfall infiltration into the ground. In this study, the spatial and temporal variation of GWL was analyzed, and a GWL prediction model based on river stage was proposed. For this purpose, a series of the finite-element (FE) analyses were performed considering various geological and hydrological conditions. The spatial and temporal responses of GWL to river stage from the analyses were quantified for various permeability and river stage conditions. Based on results of the FE analyses, a correlation model between GWL and river stage was established. The model parameters were evaluated and given in forms of design equations. The proposed method provides a simple and effective way for predicting GWL based on river stage, without an elaborate modeling process and sophisticated numerical scheme that requires heavy computation efforts. To check the validity of the proposed GWL prediction method, a test site was selected and adopted into the comparison. The predicted GWL using the proposed method indicated close agreement with the measured GWL.