Journal of Hydrologic Engineering

Drought is a natural hazard driven by extreme macroclimatic variability, and generally resulting in serious damage to the environment over a sizable area. Accurate, reliable, and timely forecasting of drought behavior plays a key role in early warning of drought management. In this study, a hybrid hidden Markov model coupled with multivariate copula (HMC) is proposed for probabilistic drought forecast. It is an extension of the regular hidden Markov model (HMM) in which the mixture distribution for each forecast is a weighted combination of posterior copula conditional distributions, which are allowed to vary with different predictors. Bayesian inference is used to optimize model structure and parameters. The cascaded sampling procedure is used to obtain conditional probability of a pair copula. The HMC model is performed for multistep meteorological drought forecast at the stations of Hanchuan and Tianmen, China, with the widely used Standardized Precipitation Index (SPI) time series. HMM, artificial neural network (ANN), and autoregressive moving average (ARMA) drought forecasting are also implemented for comparison. Results demonstrate that HMC drought forecast is much more accurate than HMM, ARMA, and ANN for point forecasts as well as interval forecasts. This study is of great significance for understanding drought uncertainty and extending drought early warning.

The flood protection topic has been the centerpiece of many studies due to the amplification of coastal vulnerabilities, especially on the east coast of the United States. In this study, increasing the reliability of the urban infrastructure has been addressed by a more accurate understanding of flood hazards and coastal protection strategies. The first step in this study is the hydrologic data analysis. Climate change and human activities have often caused an invalidation of stationarity assumption for simulation and forecasting flood hazards. Also, hydrological events often coincide with two or more natural events. Therefore, for a more accurate analysis, three approaches are introduced to assess the nonstationarity in data. Then, the nonstationary frequency analysis on a univariate distribution function and the joint probability distribution have been applied to define the frequency of flood hazards. First, a trade-off curve is plotted to contain all results of a nonstationary joint probability analysis that can be used to select the design value curve based on decision-makers’ preferences. The second step is the selection of coastal protection strategies. Strategies are selected based on their suitability and the characteristics of the protected area. The final step is to utilize a hydrologic distribution model, with the ability to simulate both rainfall and storm surge for obtaining a flood inundation map in order to test the strategies. Although, this model has certain limitations of not including wave dynamics and setup and wind speed. It succeeds in the integration of hydrologic and systems analyses, which is the main advantage of the methodology. The results show the significant value of a nonstationary analysis for obtaining design values and selecting comprehensive coastal protection strategies. The proposed methodology can be implemented in other coastal geographic settings.

The soil water retention curve (SWRC) is significant for predicting soil physical properties and solute processes. However, the existing methods for obtaining SWRC are mainly focused on laboratory measurements and have some limitations in efficiency. Thus, this study attempts to build a new SWRC estimation method using soil resistivity data in porous media based on field conditions. The key of obtaining the SWRC is determining the soil moisture and the corresponding soil suction. In this study, soil resistivity and moisture were determined in the laboratory. The relationship between resistivity and soil moisture was then determined by the Archie equation. The quantitative relationship between resistivity and soil suction was constructed using measured data and empirical models. The results show that the Archie equation could be applied to precisely determine the relationship between soil resistivity and soil moisture when the soil resistivity ranged from 100 to 800 Ω · m. Both the van Genuchten (VG) model and the Gardner model were able to fit the SWRC. For the sandy soils involved in this study, the fitting results of the VG model were better than those of the Garden model. The soil resistivity–matrix suction relationship was established to obtain the SWRC by combination with the soil resistivity–water content model. The proposed method was then tested, and it showed great accuracy and reliability in estimating the SWRC and provided a method and theoretical basis for accurately measuring the SWRC on a regional scale.

Dam removal tends to produce a series of changes in river morphology. Whereas decisions pertaining to dam removal are typically evaluated from a deterministic viewpoint, this study develops a stochastic framework for assessing the effects of hydrological variability on morphological changes in river channels resulting from dam removal. Our approach yields probability distributions for channel responses that can be used in decision-making. Application of the framework to the 25-m-tall Shihgang Dam on the Dajia River in central Taiwan reveals that the area most geomorphically sensitive to removal of the dam lies 1,200–3,600 m downstream of the dam site. We also investigate the influence of flow regime on geomorphic response using indicators of hydrologic alteration (IHA) and find that higher flows result in more pronounced channel evolution and greater thalweg migration.

Groundwater–surface water (GW-SW) exchange processes are important due to their critical role in controlling the transport of pollutants and ecologically related materials in rivers. In this paper, the surface water–groundwater coupling models of five wavy riverbed topographies were developed, and surface water velocity and pressure distribution under different wavy bed surface morphologies were studied. The upflows and downflows of sediment–water interface (SWI) were determined and quantified; in addition, the depth of hyporheic exchange and the locations of slow flow points were obtained. The obtained results showed that rippled bed morphologies disturbed the surface water flow field and separated adherent water flow at the water–sediment interface. The slope angle of water–slope of rippled beds was negatively correlated with the pressure of the water–sediment interface and positively correlated with flow velocity at the peak of the rippled bed. Upwelling and downwelling distributions basically depended on the geometry of the riverbed and did not change by changing the surface water velocity. The crest was a stable demarcation point between upflows and downflows, and the hyporheic exchange flux was negatively correlated with the angle of sloping slope. A stagnation zone existed in the hyporheic zone where both lateral and vertical flow velocity components approached zero, and its depth was similar to the depth of hyporheic zone. A lateral stagnation zone was located near the boundary of the reverse exchange zone in the hyporheic zone, and vertical stagnation zones were located on both sides of the maximum pressure of the water–sediment interface.

The prairie region in Canada has been characterized as a graveyard of hydrological models due to its challenging cold-regions processes and complex landscape with numerous land depressions that influence runoff pathways. Efforts were made at the small-basin scale to propose new algorithms and/or modify existing physically based hydrological models in order to achieve some semblance of a coherent mathematical runoff modeling system. To date, there has been very little research on modifying conceptual bucket-type models to include lateral pothole flow complexities for peak-flow estimation. In this study, the conceptual Hydrologiska Byråns Vattenbalansavdelning (HBV)-light model is modified to work in the prairies by incorporating a conceptual lateral-flow component to represent the pothole storage complexities. The modification of the HBV-light model resulted in a HYdrological model for Prairie Region (HYPR) that can be used for prairie streamflow simulation. The traditional HBV-light and HYPR conceptual models are tested on different pothole-dominated watersheds within the Qu’Appelle River Basin in Saskatchewan, Canada. The incorporation of a pothole storage-modeling component in HYPR results in a better streamflow simulation than that of HBV-light. Also, a new approach is proposed in this study to better identify the proper calibration period to arrive at a successful streamflow simulation. Although HYPR’s processes representation is simplified, the model shows potential for simulating the overall hydrograph and peak flows. HYPR shows strengths as a possible tool for operational and flood prediction purposes in the prairies, especially when data are limited.

Negative binomial distribution has been suggested to describe the first arrival time of the $k$th flood exceeding the design flood under independence and both stationary and nonstationary conditions. However, hydrological processes often exhibit temporal dependence, which can cause persistent fluctuations in observed series and clustering of extreme events that might be confused with nonstationary effects. This study focuses on a distribution of waiting time of the $k$th event exceeding a prescribed design value under stationarity and serial dependence. This probability distribution is known as beta negative binomial, which complements the models proposed for (non)stationary independent processes, and enables the comparison with results corresponding to stationary dependent processes. We discuss the properties of the beta negative binomial distribution and show its validity for theoretical occurrence processes with power-law and exponentially decaying autocorrelation functions. The proposed model is applied to peak flows and maximum temperatures recorded across the conterminous United States. Results show that the beta negative binomial distribution can capture the effect of serial dependence on the distribution of waiting time of extreme events.

This research presents a bivariate methodology for the hydrological characterization of the overtopping failure for an existing river levee. Conventional procedures usually consider only one variable for the hydrologic forcing (peak discharge, $Q$). Such an approach might fail if the volume of the hydrograph ($V$) is also significant. The proposed methodology is based on the generation of a set of plausible inflow hydrographs characterized by their peaks and volumes in the $Q\text{-}V$ space. The shapes of the hydrographs are classified according to their tendency to produce overtopping, introducing a new index: the Overtopping Hydrograph Shape Index (OHSI). The levees in two river reaches located in Tuscany, Italy, were used as a case study. As a preliminary result, it was found that the hydrographs that produce overtopping failure in both cases lay on a zone in $Q\text{-}V$ space delimited by a curve, the Critical Overtopping Flood Hydrograph (COFH) curve. The existence of COFH demonstrates that overtopping failure is not determined by a unique variable ($Q$) but rather by the combination of both $Q$ and $V$. The limiting case corresponds to a family of hydrographs with varying $Q$ and $V$ values. The COFH allows dividing the $Q\text{-}V$ space into two zones, the no-failure zone and the failure zone, that simplify the evaluation of the failure probability for a river levee.

Subdaily precipitation gauging stations are limited and unevenly distributed across Canada. To support the design of sustainable stormwater infrastructure, especially in the data-sparse regions of Canada, this study presents a novel, gridded intensity–duration–frequency (IDF) dataset of precipitation storms of 5, 10, 15, 30, and 60 min and 1, 2, 6, 12, and 24 h durations and 2, 5, 10, 25, 50, and 100 year return periods. The dataset has been prepared using atmospheric variable (AVs) estimates from two reanalysis products: the North American Regional Reanalysis (NARR) and ERA-Interim. A state-of-the-art machine-learning algorithm, named a support vector machine (SVM), is used to establish the link between AVs and extreme precipitation magnitudes. First, the most relevant AVs shaping precipitation extremes in different parts of Canada are identified, and preliminary estimates of gridded IDFs are produced. The preliminary IDF estimates are corrected for systematic distribution of spatial errors to obtain corrected gridded IDF estimates. Modeled gridded IDF estimates are compared with observations and are found to exhibit a root mean squared error varying between 5% and 25% across different regions of Canada. The gridded IDFs are also found to capture the observed spatial pattern of extreme precipitation reasonably well.

Records of discharge for the connecting channels within the Great Lakes Basin are important to national governments of Canada and the United States and the various water management agencies and users in the basin. For more than 100 years, the official discharge records for the St. Clair and Detroit Rivers, two connecting channels within the Great Lakes Basin, have been computed using various stage-fall-discharge (SFQ) methods. However, as a result of technological advancements, newer methods have recently been considered for discharge computations. In this study, three discharge computation methods were compared: two SFQ methods and the index-velocity discharge (IVQ) method. Although the two SFQ methods have significantly different assumptions and use different data from the index-velocity method, the differences between the computed discharges derived from the methods are small, especially as the time step approaches monthly discharge values. Statistical analyses of discharge measurements and discharges computed using each of these methods indicate that there is no substantive difference in the discharges computed using the three methods. However, the IVQ method provides distinct advantages over the SFQ methods, including increased temporal resolution of computed discharge (minutes versus daily) and the ability to account for changes caused by aquatic vegetation and ice. Based on the results of the comparisons described herein, the IVQ discharge computation method is the most appropriate method for discharge computation in the St. Clair and Detroit Rivers. Updated SFQ equations for the St. Clair and Detroit Rivers, also presented herein, can be used to compute discharge during periods of missing or invalid IVQ record.

Climate and land-use impacts on stream water quality in a karst watershed were studied to determine how they might be useful to model stream water quality parameter (SWQP) distribution from diffuse source pollution. This was achieved using principal component analysis (PCA), canonical correlation analysis (CCA), and artificial neural networks (ANNs) for the Upper Green River Watershed, Kentucky. Correlation analysis and eigenvalue analysis gave rise to what are known as effective input parameters for the subsequent ANN modeling. The results of ANN modeling indicated that although land-use contributions were better correlated when they were combined with temperature and precipitation for some of the SWQPs, most of the SWQP distribution trends were better predicted using the field data of temperature and precipitation at the watershed scale. This was in contrast to the findings of the many nonkarst watersheds where SWQPs showed a significant dependence on land use alone according to the literature. This, in essence, implies that the land-use influences are secondary, and precipitation and temperature affect the Upper Green River karst watershed’s SWQPs primarily, whereas for nonkarst watersheds the land use influences SWQPs primarily from diffuse source pollution.

The Australian water balance model (AWBM) is automatically calibrated to 204 data sets from the main stream-gauging network in Australia. This daily rainfall-runoff model calculates surface runoff and baseflow as separate components of runoff, and the curve number (CN) is calibrated to the surface runoff component in each data set. The catchment areas ranging from 52 to $\mathrm{1,970}{\mathrm{km}}^2$ contain mixes of soils and vegetation, but the calibrated CN can be used in the catchment areas and surrounding regions. There was a distinct relationship between the median calibrated CN for average antecedent rainfall (CNII) value of each group of catchments and the mid latitude of the group with an increase of one in the median CNII for each one degree increase in latitude. The method has potential for providing a very large amount of surface runoff data for calibrating the CN from the 3,000 or so stream-gauging stations in Australia.

Nitrate-nitrogen (nitrate) pollution remains one of the most common causes of surface water quality degradation. Previous studies explored the fractal scaling behavior of controlling factors, but few studies investigated the change of scaling characteristics with time and the influence of a combination of factors on streamflow nitrate concentration (SFC). This study showed that a suite of hydrological factors impacting the SFC variations at multiple temporal scales can be determined using wavelet coherence analysis. The scaling behavior for controlling factors was evaluated using multifractal detrended fluctuation analysis. Results showed that streamflow and base flow exhibited persistent behavior, whereas the precipitation exhibited variable scaling behaviors that contributed to significant SFC fluctuations at multiple scales. Streamflow had substantial control on the SFC fluctuations when the time scale was shorter than 90 days. The combination of precipitation and streamflow formed the dominant processes impacting SFC at larger temporal scales ($>90\mathrm{days}$). These results provide an effective way to identify the factors influencing SFC variations across scales and predict the temporal evolution of SFC.