Journal of Hydrologic Engineering

The effect of evapotranspiration (ET) on water redistribution in hydrological models has already been explored in many studies, but whether correcting plant transpiration (EP) in a hydrologic model can play a comparable role remains unclear. This study aims to evaluate the effect of EP from the perspective of model calibration. Reference EP data were derived from the Biome Biogeochemical Cycles (Biome-BGC) model, with parameters based on long-term field investigation of the Qinghai-Tibetan Plateau. Soil and Water Assessment Tool (SWAT) model calibration and EP-sensitive parameter analysis were conducted in the plateau’s Huangshui basin. Four modeling experiments were explored: (1) SWAT base model; (2) calibration against EP for each subbasin; (3) calibration against runoff at the gauge station; and (4) calibration against both EP and runoff. The effects of EP and runoff were analyzed. The results indicate that calibration against either EP or the runoff process alone is not adequate to produce spatially and temporally accurate water partitioning. Specifically, calibration against EP improves the runoff simulation during wet and warm seasons and the EP spatial pattern across the basin. Parameters involved in vertical water redistribution are sufficient but not necessary conditions to affect the EP process in the SWAT model. Calibration against runoff gives priority to regulating the EP in the watershed as a whole, retaining a negative spatial correlation with the reference EP. To produce a better simulation of EP and runoff, both factors are recommended in the calibration based on the SWAT model parameterization.

Accurate annual runoff prediction plays an important role in modern water resources planning and management. Here, a hybrid model using evolutionary extreme learning machine and variational mode decomposition (VMD) is developed to forecast annual runoff time series. In the proposed method, the VMD method is first used to decompose the original streamflow into a series of disjoint subcomponents; second, each subcomponent is forecasted by constructing an appropriate extreme learning machine model while the gravitational search algorithm is adopted to tune the model parameters; finally, the aggregated output generated by the forecasting results of all the models is treated as the final simulated output. The annual runoff data series of three huge hydropower reservoirs in China are chosen to testify the performance of the proposed forecasting model. The results show that the developed model can outperform several traditional methods with respect to the employed statistical indexes. Thus, the decomposition-ensemble idea is helpful to yield accurate and stable forecasting results, while the proposed forecasting method can provide strong technical support for operators in water resources and power systems.

Solar irradiation is the dominant source of energy driving hydrologic phenomena on the Earth. This study is a step toward investigating the statistical significance of the relationship between hydrologic data and persistent solar cycles, thereby expanding our knowledge of developing long-term periodic hydrologic models. The task was performed by smoothing the historical monthly mean sunspot data from January 1749 to December 2014, using a discrete Fourier model with a logistic distribution filter capable of removing low amplitude harmonics in the frequency domain, in order to reveal the essential cycles and detect the fundamental solar period. The fundamental period is exploited to promote a Fourier model for generating sunspot forecasts that can be devoted to monitoring, on average, the variation pattern of hydrologic data beyond their records. The results showed that the sunspot pattern can reasonably be synthesized by considering a single fundamental period equal to 2,532 months corresponding to de Vries cycle. A case study of monthly rainfall data was used to develop a sinusoidal model triggering typical solar periodicities of Schwabe, Hale, Gleissberg, and de Vries cycles. The model was able to capture the current long-scale rainfall pattern resulted from hidden periodicities that would only be revealed if a wider range of historical rainfall data was available.

River flow prediction is an important aspect of robust water resources planning and flood warning systems operation. Data-driven approaches have been found efficient to this end. K-nearest neighbors (KNN) is a lazy learning method that can be used for this purpose. In this study, a new method for selecting neighbors named dynamic number of k-nearest neighbor (DKNN) is introduced which uses an optimized distance to select a different number of neighbors for each instance of predictors instead of using a fixed k number as in the classic method. The particle swarm optimization (PSO) algorithm is used for the optimization process to improve the results. Three techniques for prioritizing the contributing neighbors are applied: (1) using the pattern of the predictor data, (2) considering the date of the predictor data, and (3) using both of these features in the prediction procedure. The performance of the proposed method and techniques is tested using 2 years of the daily inflow to the Gheshlagh reservoir in Iran and is compared with the results of classic KNN, artificial neural networks (ANN), random forest regression (RFR), and support vector machines (SVM). The results indicate that the proposed method increased the accuracy of prediction by 4.9% by reducing the root-mean-square error (RMSE) compared to the classic KNN. Using the recorded date of the predictor gives the best performances out of the three proposed techniques and performs better than classic KNN, ANN, RFR, and SVM by showing 49%, 38%, 31%, and 24% improvement in RMSE, respectively. Considering the pattern of the predictor and the combined technique also resulted in 12% and 35% reduction in RMSE, respectively, compared to classic KNN.

Construction of recharge dams across highly permeable beds of alluvial streams is one of the most important techniques used for artificial groundwater recharge from muddy surface runoffs. To assess recharge dams, the Meymand and Tangeriz recharge dams located in southern Iran were studied. More than 90% of the Tangeriz recharge dam capacity was artificially created by excavating two vertical sidewall pits behind the dam. During the most extreme recorded rainfall in the study area, the ratio of inflow runoff volume to reservoir capacity was equal to 1.43 and 16.59, respectively, for the Meymand and Tangeriz dams. The Meymand dam was overflowed, while the Tangeriz dam was not overflowed and stored runoff in the pits was quickly recharged. To interpret the large recharge capacity observed in the Tangeriz recharge dam, a theory of vertical ponded flow in two-layered soils with a low-permeability upper layer was developed. Theoretical development supported by numerical modeling and laboratory/field observations showed that vertical infiltration in reservoirs of recharge dams is mainly controlled and limited by sedimentation of a low-permeability silt layer on the reservoir bed. In contrast, due to negligable sedimentation, lateral infiltration through vertical sidewalls is not influenced by siltation. Therefore, excavation of vertical sidewall pits inside the reservoir of recharge dams is proposed as a novel method for improvement of artificial groundwater recharge. A high lateral recharge rate through the vertical sidewall pits of the Tangeriz recharge dam confirms the effectiveness of the proposed method.

The estimation of hydrometric rating curves uncertainty has constituted an active topic of research on hydrology. In this regard, the BaRatin inference framework, which estimates rating curves on the basis of prior hydraulic knowledge, has been considered a promising alternative. In building inference setups, a variety of structural error models have been combined with BaRatin. However, most of them neglect the potentially high scatter levels in the lower portion of rating curves, caused by changes in the channel bottom. For addressing this issue, in this paper we propose a Gaussian heteroscedastic structural error model, which attributes larger uncertainty for both upper and lower portions of the rating curve. The inference framework was applied to two catchments in Brazil with distinct hydraulic controls and channel bed stability conditions. Results demonstrated that, under the proposed error model, the total uncertainty intervals encompassed most measured large flows and even relatively high scatters of low discharges, which suggest the overall suitability of the proposed modeling strategy and its capacity to achieve more realistic intervals of uncertainty.

Estimating reliable projections of precipitation considering climate change scenarios is important for hydrological studies. General circulation models provide future climate simulations at large scale in terms of large-scale atmospheric variables (LSAVs). Those LSAVs can be downscaled to finer special resolution using several downscaling approaches. This paper presents a support vector regression (SVR)-based downscaling approach to downscale rainfall at several locations in a study area. Because the rainfall generation mechanisms cannot be the same for all the sites in a study area, conventional multisite downscaling approaches that assume the same rainfall generation mechanism should not be used. Therefore, a new downscaling approach is proposed that (1) divides the study area in several climatological regions, and (2) develops different downscaling models for each of the climatological regions to obtain future projections of rainfall. The new approach was implemented on rainfall data obtained for Republic of Ireland to demonstrate the effectiveness of the approach compared with existing approaches. Future projections of rainfall were obtained for the period 2012–2050 corresponding to four Representative Concentration Pathway climate change scenarios. The performance of the SVR approach was compared with that of relevance vector machine– and deep learning–based downscaling approaches.

We used a fixed-bed flume test and ultrasonic sensor measurement technology to study the evolution of the characteristics of thin-layer flow roll waves on a sloped surface. The results indicated that there was a positive correlation between the velocity of the roll wave and the Reynolds number when the energy slope and roughness were small. In addition, the amplitude of the increase gradually decreased with increases in roughness, and the height of the roll waves showed a single-hump trend of change. When the roughness was relatively small, the frequency of the roll waves increased as the Reynolds number increased, and the ratio of the wavelength to the water depth gradually decreased. As the energy slope increased, both the wave velocity and frequency of the roll waves gradually increased; they peaked when the energy slope was 0.3420. We found that the dominant factors that affected different roll-wave characteristic parameters varied. The roughness of the bed had a relatively prominent impact on the velocity of the roll waves, but the energy slope had a greater effect on the height of the roll waves.

Use of flood statistics in the regionalization of watersheds may improve the homogeneity of regions. However, it is not possible to use the feature vectors, including the flood statistics, to estimate the flood quantiles for ungauged watersheds by most of the regionalization methods, especially those based on cluster analysis. In the current study, the effect of merging a flood statistic into a feature vector consisting of watershed features has been assessed and then, some algorithms based on the cluster analysis have been proposed for regionalization of watersheds. The proposed algorithms can be applied to the feature vectors corresponding to the gauged and ungauged watersheds simultaneously. The algorithms were applied to the data related to the Karun-e-Bozorg basin in Iran. The results showed that joining a flood statistic with an appropriate weight to the feature vectors used in regionalization may improve the results of regionalization according to the number of the watersheds assigned to the homogeneous regions. Also, the proposed regionalization algorithms can increase the number of watersheds assigned to the homogeneous regions noticeably.

Huff curves are probabilistic time distributions of rainfall expressed as dimensionless cumulative percentages of storm depth and duration. Previous studies have documented development factors, spatial robustness, and the utility of Huff curves in practical applications. However, the effects of trending rainfall on Huff curve intensity patterns have not yet been studied. The goal of this paper is to fill this gap by studying Huff curve patterns in a watershed with demonstrated increasing trends of temperature and precipitation, with the intention that it can be generalized to other areas in the US and the world. To achieve this goal, the high temporal resolution precipitation data collected from a high spatial density, 72-year precipitation-gauge network on the $4.25\text{-}{\mathrm{km}}^2$ North Appalachian Experimental Watershed in east-central Ohio were used. Seasonal storm pattern trends from 1939 to 2010 were investigated using dimensionless depth (with the frequency of 50%, $d_{50}$) and the curve variability ($V=d_{80}-d_{20}$) at three dimensionless within-storms time periods (three verticals). The Spearman rank correlation procedure (correlation coefficient, $\rho$ and significance probability, $p$) was used to statistically determine trends over time using 8 periods of 4-season sets of Huff curves over the 72 years. Two averages of $\rho$ and $p$ were computed: (1) by averaging the individual $\rho$ and $p$ obtained from 10 gauges (AvgI); and (2) by grouped averaging of all individual gauge values of $d_{50}$ and $V$ and then computing $\rho$ and $p$ (AvgG). The test results of individual gauges showed that 4 cases for $d_{50}$ and 23 cases for $V$ were significant for all seasons and verticals (total of 120 cases for each variable). The test results of AvgI for $d_{50}$ and $V$ and AvgG for $d_{50}$ showed no significant trends in all seasons and verticals. Only the AvgG for $V$ led to a significant trend for $V$ in spring and fall at different times within storm patterns. The data do not provide sufficient evidence at the $p=0.05$ significance level to reject the null hypothesis of unchanging position of the dimensionless depth of the 50% Huff curves for individual or averages for all seasons and verticals. Also, there is insufficient evidence to reject the null hypothesis of unchanging variability, $V$, using the average $\rho$ of individual gauges (AvgI for $V$) for all seasons and verticals. AvgG results showed a significant trend in $V$; however, this analysis may be affected by the nonindependence of storm data. The results suggest that it is likely that there is little if any effect of trending climate over approximately 70 years on Huff curve patterns. The results of this study add to the robustness characteristics of Huff curves and to their potential use in hydrological practice as design storms, as the foundation of stochastic storm generation, and for storm disaggregation, and they deserve further investigation. These different forms of inputs to watershed models have the potential to improve runoff estimation. The results suggest that, if verified in other studies, they have applicability to provide useful stationary precipitation patterns across the US and other areas of the world in areas of nonstationary climate. Also, individual rain gauge data may not be representative of trends even over small areas, and seasonal differences were noticeable as found in other studies. Recommendations are provided.

Porous basins are popular low-impact-development (LID) facilities for on-site runoff disposal and water quality enhancement. Bioretention basins and rain gardens are recommended to intercept the initial runoff up to the water quality capture volume, which depends on the catchment area and imperviousness. Underneath the basin’s porous bottom exists a top layer of sand filter and a bottom layer of back-filled gravel. The seepage flow through the filtering layers is collected into a perforated subdrain pipe. The challenge facing the design of a porous basin is the trade-off between its residence time and drain time. On the one hand, the longer the drain time, the better the water quality enhancement. On the other hand, the faster the flow is released, the greater is the amount of runoff volume treated. Therefore, the drain time is the key parameter for the design and operation of a porous basin. This paper derived a theoretical solution to estimate the drain time to deplete the seepage flows through filtering layers underneath a porous basin. The discharge coefficient was calibrated to take into consideration the friction losses through the sand and gravel layers. This method provides a direct solution to predict the drain time and significantly improves the simplified formula used in numerical modeling.

River width is an essential physical variable for the simulation of rainfall-runoff processes using distributed hydrologic models. In this paper, a model of river width integrating the effects of an upstream drainage area and terrain topography is proposed for the derivation of a spatial variation of a river width of medium-sized mountainous watersheds based on a digital elevation model (DEM). Tests of the model for three medium-sized watersheds, including Daheba, Changhua, and Yuqian, suggest that the spatial variation of the river width is dominated by topography in terms of slope near the riverbed. Wider river reaches are often located at flatter terrains with a larger upstream drainage area. In addition, the river width estimated method of the TOPographic Kinematic Approximation and Integration (TOPKAPI) model, which only uses the upstream drainage area to estimate river widths, is compared with the proposed model. Relative to a percent bias of 15%–19% of the river width model (GeoRW), the error of the method in TOPKAPI is obviously larger, at about 32.9%–65.4%. A comparison of the proposed model with the river width estimated method of the TOPKAPI model based on the upstream drainage area justifies its usefulness for medium-sized mountainous watersheds.

The present study quantifies the variation in water balance components of the semiarid Upper Girna basin (UGB), Maharashtra, India, and the flow quantiles of a dam downstream of the UGB under changing climate using the semidistributed hydrologic model Soil and Water Assessment Tool (SWAT). The SWAT model was developed using digitized actual ground-surveyed data (1971–1979) spanning ${\mathrm{4,730}\mathrm{km}}^2$; it was calibrated (1991–1998) and validated (2006–2010) using hydrometeorological data. The water balance components of UGB as simulated by the calibrated SWAT model for the historical observed period (1981–2010) have water yield and evapotranspiration contributions of 17% and 56%, respectively, out of the total 100% precipitation input to the system. All the water balance components as simulated by SWAT for all future climate scenarios [for two representative concentration pathways (RCPs) 4.5 and 8.5 $\mathrm{W}/{\mathrm{m}}^2$ under three time slices 2011–2040, 2041–2070, 2071–2100] represented by CMIP5 general circulation models (GCMs) are lower than those historically observed. Additionally, based on the yearlong and monsoon month flow duration curves, a significant change is observed in simulated streamflow between the historical climate and future climate scenarios. The variation in the magnitudes of all the flow indexes between the CMIP5 GCMs for the early and midcentury time slices is less than the variation for the late century time slice, as demonstrated from dispersed flow duration curves. The present case study is the first of its kind to carry out climate change impact analysis for the UGB; according to the spatiotemporal maps of precipitation and other water balance components, Subbasins 2 and 5 in the UGB are plausible for severe depletion of water resources. The present work provides significant insights for resource managers and policy makers for efficient water resources planning and management in the UGB.

Preparing for the impacts of climate change, especially extreme rainfall events, is not a “one size fits all” process. Exhaustive case studies must be reported to understand the impact of climate change in a local area. However, there have been some difficulties in presenting all procedures used to derive the impact of climate change. Therefore, the current study presents a local case study of how a local small basin is prepared to mitigate the effects of climate change on extreme rainfall events. From the case study, the full procedure to produce an intensity-duration-frequency (IDF) curve regarding a number of future global circulation model (GCM) daily precipitation scenarios is described in detail. The major portion of this work is focused on simply estimating extreme rainfall intensity with an IDF curve considering climate change scenarios from a GCM ensemble. From all available GCMs (19), the IDF ensemble is estimated with the following procedure: (1) daily GCM outputs obtained from the grid point that is closest to the target area were bias-corrected with gamma distribution after checking the suitability of the distribution model; (2) the bias-corrected daily precipitation data were downscaled; and (3) the IDF curves for the future scenarios were estimated and an ensemble was used to produce the final IDF curve. The result indicates that the IDF curve of future scenarios effectively inherits the behaviors of the original GCM daily precipitation outputs. The future IDF estimate will be employed to prepare for the effects of future climate change on extreme rainfall events on a local scale.