Journal of Hydrologic Engineering

The design of optimal hydrometric networks is an important starting point in water resources planning and management. Redundant or inappropriate networks may require unnecessary monitoring costs, while a sparse network may cause a lack of understanding of the process being monitored. Many studies employ information theory, which uses the Shannon entropy, as a measure of the information to design optimal hydrometric networks measuring various hydrologic parameters, such as streamflow and precipitation. The majority of entropy application methods in hydrometric network design have had two common objectives, i.e., maximizing joint entropy and minimizing total correlation. However, it is still unclear what data lengths should be adequate to properly use the entropy approach to network design and how the data lengths affect the entropy values. In this study, four different data lengths (e.g., 5, 10, 15, and 20 years) of daily time series are used to determine the optimal streamflow and precipitation networks using entropy theory coupled with multiobjective optimization. The spatial distributions of the optimal monitoring locations appeared similarly for each data length. Specifically, the hot-spots where the selection likelihood from optimization results is high were not significantly changed; this is more obvious when the data length of daily time series was 10 years or greater. Additionally, the joint entropy and total correlation of the optimal networks were calculated from 10 days to 20 years with a 10-day increment. The joint entropy increased significantly during the first 5 years and then gradually increased without significant change after 10 years. Similarly, the total correlation stabilized after 5 years of daily time series lengths with no major change after 10 years. Therefore, it is recommended to use at least 10 years of data for information theory–based hydrometric network design when using daily time series.

The study presents a comparative assessment of regionalization approaches for streamflow estimation at ungauged catchments that are independent of any hydrological models. Four regionalization techniques including area-index, inverse distance weighted (IDW), kriging, and stepwise regression were applied on 32 catchments in India to estimate daily streamflow. Each catchment was considered as ungauged in turn during regionalization. The area-index method was dependent upon the drainage area of the catchment, whereas stepwise regression method was based on the hydroclimatological and morphological attributes of the catchments. IDW and kriging were based on spatial distance between the donor and receiver catchments. The performance of each regional model was evaluated in terms of Nash–Sutcliffe efficiency (NSE), root-mean square error (RMSE)–observations standard deviation ratio (RSR) and percent bias (PBIAS). The results obtained from IDW and kriging were found to be superior to other techniques. An average NSE value of 0.6 was found to be better for these two methods than other approaches. The area-index method performed the worst, maybe due to the fact that it considers only drainage area.

Dams are ubiquitous in the United States, with more than 87,000 influencing streamflow across the nation. The significant majority of these dams are small and are often ignored in real-time flood forecasting operations and at-site and regional flood frequency estimations. Even though the impacts of individual small dams on floods is often limited, the combined flood attenuation effects of a system of such dams can be significant. In this study, the authors investigate how a system of spatially distributed small dams affect flood frequency across a range of drainage basin scales using the $660\text{-}{\mathrm{km}}^2$ Soap Creek watershed in southeastern Iowa, which contains more than 144 small dams. Results from continuous simulation of the system of small dams indicate that peak discharges reduced between 20 and 70% with the effect decreasing as the drainage area increases. This means that neglecting the effects of the system of small dams may lead to an overestimation of flood risk, which has implications in both flood frequency estimation and real-time flood forecasting. Considering that more small dams are being built across watersheds in Iowa and elsewhere in the country, the results also highlight how the peak discharge attenuation effects of these dams is an additional factor that invalidates the stationarity assumption that is used in at-site and regional flood frequency analysis.

This paper presents a decision support system (DSS) that can be used by reservoir managers for modeling and visualizing the complex relationships between key variables such as rainfall, streamflow, reservoir level, and water releases. The proposed approach uses a self-organizing map (SOM), a competitive-learning artificial neural network frequently used for vector quantization and clustering. The map uses an unsupervised learning approach and is trained with observed (i.e., historical) input data. This novel nonparametric approach can identify and characterize the different operation conditions and operation policies implicit in the training data. A SOM provides a set of operation profiles, each with specific values of the uncontrollable, state, and controllable variables. The topology-preserving character of SOMs allow for an intuitive and helpful interpretation of the computed prototypes and their relationships. The knowledge provided by these operation profiles can be used in many different ways, some of which are illustrated in the paper. The proposed approach has been tested on the Guadalmellato River reservoir in southern Spain. In this application, intense as well as prolonged rainfall episodes can be identified and characterized, as well as the dry season. The map clearly delineates specific SOM regions for hydroelectricity generation and flow-balancing discharges. The map provides insights into the correlation between these controllable variables and the uncontrollable rainfall and streamflow variables mediated by the state variables (e.g., reservoir level). The trained SOM can be used to study past periods (e.g., of heavy rains) and understand and revise the course of action followed. It can also be used to suggest future operation decisions based on the forecasted values of rainfall, streamflow, and reservoir-level variables.

Nonlinear and linear reservoirs are popularly used in simulation and characteristic analysis of streamflow recessions after various rainfall events. In this study, the power-law function $-(dQ/dt)=a{Q}^b$ for streamflow ($Q$) recessions over time ($t$) was mathematically expressed by using a two parallel linear reservoirs model as an approximation of fast and slow base flows in the subsurface layers. The two parallel linear reservoirs model displays multiple curves of $-(dQ/dt)=a{Q}^b$ in which the recession slope $b$ and intercept $a$ depend not only on catchment features, e.g., recession timescale parameters of fast and slow base flows, but also on time-varying flow composition and initial streamflow. The two parallel linear reservoirs model is applied in the Lantang Watershed of Southeast China where the subsurface flows come from the soil and underlying weathered bedrock with distinctly different hydraulic characteristics. The model replicates well the individual recessions and the multiple curves of $-(dQ/dt)=a{Q}^b$ in the log-log space. The results indicate that the single-valued relations of $-(dQ/dt)=a{Q}^b$ and storage-discharge with constant values of $a$ and $b$ are a specified approximation of the multiple curves of $-(dQ/dt)=a{Q}^b$ with variables $a$ and $b$, and the two parallel linear reservoir model is flexible to capture the complex recession characteristics in the heterogeneous watershed.

In frequently monitored fluvial systems, such as the middle Mississippi River, large databases of hundreds of cross-section surveys are often needed for assessment of long-term morphologic change and contributing hydraulic parameters. Developing customized interpolation schemes with optimized coefficients for each survey in such large databases is impractical. Using a database of 158 cross-section bathymetric surveys of the middle Mississippi River, automated surface generation of commonly available interpolation algorithms, [i.e., triangulated irregular networks (TIN), inverse distance weighting (IDW), spline with tension, and ordinary kriging] were evaluated and compared with an interpolation algorithm developed for this database, termed rectilinear IDW (RIDW). TIN and RIDW interpolation algorithms had the lowest root-mean square error (RMSE) among all methods evaluated, with the largest difference between them occurring along the perimeter of the interpolated surfaces. TIN interpolation resulted in sharp, geometric edges along the boundaries of the interpolated surfaces requiring postprocessing, whereas RIDW interpolation produced smoother edges that sometimes elongated scour holes.

The Muskingum model is one of the most popular models for hydrologic channel flood routing. However, as the parameters of this model are estimated by analyzing both inflow and outflow data, the application of this model to an ungauged channel reach has been limited. In this study, as a solution to this problem, a methodology is proposed to estimate the parameters of the Muskingum model using the basin characteristics representing the inlet and outlet of the channel reach. That is, the Muskingum model could be expressed as a combination of the linear reservoir model and the Dirac delta function, whose weighting factors are determined as a function of the Muskingum model parameter $x$. It was also found that the storage coefficient of the linear reservoir model is expressed by both the Muskingum model parameters $x$ and $K$. Finally, it was shown that the Muskingum model parameters can be estimated using the information of concentration time and storage coefficient representing the inlet and outlet of the given channel reach. In the application example with the observed data from the dam discharge experiment in the Geum River Basin, Korea, it was shown that the Muskingum model parameters could be reasonably estimated to be very close to those estimated by the conventional graphical method.

Explicit prediction of the suspended sediment loads in rivers or streams is very crucial for sustainable water resources and environmental systems. Suspended sediments are a governing factor for the design and operation of hydraulic structures, like canals, diversions and dams. In recent decades, to model hydrological phenomena which are complex in nature the machine learning models are used commonly. In the present study, support vector machine (SVM) with wavelet transform (WASVM) has been employed for prediction of daily suspended sediment load (SL) for two south Indian watersheds (Marol and Muneru) using hydrometeorological data. A 40-year daily observed data (1972–2011) have been used for the analysis, where past SL, streamflow ($Q$), and rainfall ($R$) data were used as the model inputs, and SL was the model output. Using conventional correlation coefficient analysis between input and output variables, the best input of WASVM model was identified. The reliability of SVM and WASVM models were evaluated on the basis of different performance criteria, i.e., coefficient of determination ($R^2$), root mean square error (RMSE), normalized mean square error (NMSE), and Nash-Sutcliffe coefficient (NS). Initially, 1-day ahead SL prediction was performed using the best WASVM model. The results showed that, 1-day predictions were very precise, showing a close agreement with the observed SL data ($R^2=0.94$, $\mathrm{NS}=0.94$ for the Marol watershed, and $R^2=0.77$, $\mathrm{NS}=0.77$ for the Muneru watershed) in the testing period. The same WASVM model was then used for the prediction of SL for the higher lead periods. The NMSE value for the Marol watershed was found as low as 0.06 for 1-day ahead prediction, and increases subsequently as 0.29, 0.46, and 0.70 for 3-, 6-, and 9-day higher leads, respectively. Likewise, for the Muneru watershed, the NMSE value was found as low as 0.21 for 1-day ahead prediction, and increases subsequently as 0.42, 0.53, and 0.68 for 3-, 6-, and 9-day higher leads, respectively. Further, the model was evaluated on the basis of its capability of predicting peak SL and cumulative SL for 1- to 6-day leads. The statistical analysis shows that the developed WASVM model can predict the target value successfully up to a 6-day lead and is not suitable for higher lead specifically in the selected watersheds having similar hydroclimatic conditions like the ones selected in this study. Predictions by the WASVM model were found significantly superior to the ones obtained by the conventional SVM model. The results revealed that the WASVM model provides a very good accuracy in predicting SL and can be used as an effective forecasting tool for hydrological applications.

Groundwater vulnerability assessment delineating areas that are susceptible to contamination from future scenarios has aroused worldwide attention. In this study, the authors (1) estimate future groundwater vulnerability in Hunan province, China, under urban-related land-use change and climate change scenarios; and (2) analyze the importance of related parameters to future groundwater vulnerability. The DRASTIC model [including seven parameters: depth to water table (D), net recharge (R), aquifer type (A), soil type (S), topography (T), impact of vadose zone (I) and conductivity (C)] together with an extra parameter, land-use patterns, was used to generate the map of groundwater vulnerability in future scenarios. The results indicated that vulnerability classes had an increasing trend from low to high vulnerability in the future scenarios. Hunan province may face high groundwater pollution risk in the future. The sensitivity analysis indicated that the depth-to-water table may be the dominant factor, and the land-use pattern was the most sensitive parameter on the predicted future groundwater vulnerability in Hunan province. Decision makers should identify the potential future groundwater vulnerability and take early steps to protect groundwater resources.

Forecasting of monthly and annual groundwater levels is important for water resources management, irrigation, and assessment of climate change. This study employs entropy spectral analysis for forecasting monthly groundwater levels. For spectral analysis, the domain of consideration for defining entropy is the frequency domain, in which three types of entropies are known: Burg entropy, configurational entropy, and relative entropy. These entropies lead to three types of spectral analysis: (1) Burg entropy spectral analysis (BESA), (2) configurational entropy spectral analysis (CESA), and (3) relative entropy spectral analysis (RESA). BESA, CESA, and RESA are employed to analyze spectra and forecast monthly groundwater levels, and then they are compared to determine which spectral analysis method better forecasts the monthly groundwater level. Monthly and annual groundwater data were obtained from South Carolina to verify the three methods. Both monthly and annual groundwater level data showed significant decreasing trends at almost all stations. It was found that relative entropy yielded the highest resolution in determining the spectral density, while for simulating groundwater levels, all three methods fitted the observed values well. This was indicated by the average value of Nash-Sutcliffe efficiency (NSE) for BESA, CESA, and RESA being 0.69, 0.70, and 0.70, respectively.