Journal of Hydrologic Engineering

An investigation of a groundwater system often requires the prediction of the spatiotemporal distribution of groundwater heads, which determine the groundwater flow direction and magnitude, among other applications. How to quantify the uncertainty of groundwater flow has been an important issue of concern to hydrogeologists for more than $40\mathrm{years}$, and many methods based on stochastic and statistical approaches have been proposed to resolve this issue, with considerable progress being made. However, the statistical characteristics of parameters (e.g., hydrogeological parameters, boundaries, and sources/sinks) are often difficult to obtain precisely in practical applications. From a perspective of the interval uncertainty of groundwater, this study proposes a nonlinear expression of the groundwater head interval (GHI) using an interval uncertainty approach to quantify the uncertainty of numerical simulations in groundwater flow. This study uses a steady-state groundwater flow example to analyze the computational effectiveness of the acquired expression of GHI. A comparison of the results with the equal interval continuous sampling method (EICSM) shows that when the changing rate (which is the ratio of the absolute change in the numerical value to the numerical value itself) of a hydrogeological parameter is no more than 0.2, the relative discrepancy of the obtained GHI in this study and EICSM is no more than 5%. When the changing rate is no more than 0.3, such a relative discrepancy can be controlled within approximately 10%. In general, the proposed method is computationally efficient and robust for quantifying GHI. Finally, the method of this study is applied successfully to a real case.

Evaporation of unsaturated loess refers to the phenomenon that, under the vertical driving force, the water contained in soil moves to the evaporation surface and diffuses into the atmosphere as water vapor. This lowers the water content of unsaturated loess. The existing calculation methods of soil evaporation mainly consider environmental factors and soil water content. However, the soil water gradient (SWG), which has been ignored by previous studies, can also significantly affect loess evaporation. The objective of the present study was to establish an SWG-dependent evaporation prediction model for evaluating the SWG-affected evaporation of loess. First, the parameters of SWG-affected water evaporation of loess samples with changed densities and water contents were assessed by indoor and outdoor evaporation experiments. The results indicated that the ratio of soil evaporation to water evaporation (i.e., the evaporation coefficient) was mainly related to the water supply capacity of soil. The SWG exerted a large effect on loess surface evaporation, especially when the loess had high dry density and low surface water content. For example, the evaporation intensity of a soil sample with 6% (20%–26%) SWG and $1.3\mathrm{g}/{\mathrm{cm}}^3$ dry density was found to be 62.92% higher than that of a soil sample with 20% uniform water content; the evaporation intensity of a soil sample with 6% (15%–21%) SWG and $1.5\mathrm{g}/{\mathrm{cm}}^3$ dry density was found to be 114.24% higher than that of a soil sample with 15% uniform water content. Using both soil water content and SWG as parameters to characterize the soil water supply capacity can establish a model for the prediction of the unsaturated loess surface evaporation coefficient. Moreover, comparison of predicted and measured data ensured that the model can precisely predict the unsaturated loess surface evaporation coefficient.

Stormwater managers use hydrological models to understand the runoff management performance of stormwater systems in urban catchments. However, few published studies attempted to evaluate the performance of hydrological models to represent water sensitive urban design (WSUD) elements at a catchment scale. This study reported on an evaluation of the Storm Water Management Model (SWMM) to represent a study catchment before installing leaky well systems in the catchment (preinstallation) and after introducing leaky wells (postinstallation). The process of representing individual leaky well systems using the user-defined hydraulic storage node or the built-in infiltration trench tools was also compared. The modeling approaches were evaluated against observed flow and infiltration data based on goodness-of-fit statistics, including the Nash-Sutcliffe Efficiency (NSE) and Kolmogorov-Smirnov (K-S) test for the catchment model. The results of goodness-of-fit tests and a K-S test indicated that SWMM can simulate a statistically similar runoff series to the observed series before and after the installation of leaky wells in the catchment. The ability of SWMM to model a leaky well performance using the storage or a built-in infiltration node was also evaluated using goodness-of-fit statistics against adopted criteria for good models. The results of model calibration and validations demonstrated that SWMM was able to simulate the performance of leaky well systems at the catchment scale, and the best performance was achieved by adopting a storage node to represent leaky wells. The model evaluation as per the adopted criteria indicated that the storage node model is suitable for assessing the impacts of leaky wells on stormwater runoff volume and peak flow rates.

Spreading of treatment reagent into a contaminated aquifer is a key factor for determining the efficacy of an in situ groundwater remediation method. Engineered injection and extraction (EIE) is a novel method of enhancing spreading in aquifers by inducing unsteady flow fields using sequential injection and extraction of freshwater around the treatment area. Chaotic advection in response to EIE stretches and folds the contaminant–reagent interface, thereby increasing the efficiency of remediation. Several numerical simulations investigating the effect of chaotic advection in response to EIE in porous media are available in literature but are limited to two-dimensional flow conditions. Considering the three-dimensional nature of real aquifer systems, an innovative addition to the existing two-dimensional models of EIE is introduced in this study. The three-dimensional perspective of EIE was established in this study by replacing fully screened wells in conventional EIE techniques with partially screened wells, thereby performing injection and extraction at a different vertical level. Three-dimensional numerical simulation of EIE in a confined aquifer was carried out using groundwater modeling software. The effect of heterogeneity and anisotropy on spreading in aquifers in the context of EIE also was examined. Results of simulation with variable active screen depth showed a considerable increase in spreading, which was quantified here in terms of area and volume of spread. Lyapunov exponents calculated from three-dimensional simulation results showed the presence of characteristic chaotic advection in three dimensions. Real field conditions such as heterogeneity and anisotropy were found to complement the spreading due to EIE in porous media. The insights obtained from this study can provide crucial guidance for the successful and effective application of EIE in the field.

Scour is one of the most relevant causes of bridge collapse. Most existing scour models have been focused on estimating scour depth as input for bridge design. This estimation is mainly deterministic. Parameter uncertainty has been considered to estimate expected scour as well. This framework is suitable for bridge design but has limitations for risk analysis of bridge networks, in which exceedance probabilities are needed. In this paper we use the first-order reliability method to estimate the probability that the actual scour will exceed the design scour, considering the uncertainty in hydraulic and hydrological parameters. The procedure was applied to a case study in Chile using the flood-flow history between 30 and 60 years of eight fluviometric stations. The local, contraction, and general scour were estimated for return periods between 2 and 500 years. The exceedance probability obtained was highly dependent on the uncertainty in annual maximum flow, Froude’s number, top width, and riverbed longitudinal slope.

Accurate and rapid simulation and monitoring of urban inundation play important roles in urban flood prediction and warning. This paper presents a case study of campus-scale rainfall-flood inundation, including simulation and monitoring work, conducted at Xi’an University of Technology. A surface hydrodynamic numerical model based on graphics processing unit (GPU) acceleration technology is proposed. It is termed as GPU-accelerated surface water flow and associated transport (GAST), and is employed to simulate and analyze the rainfall runoff process and the drainage pipe network process of the study area. A uniform grid of $624\times 550$ units with a high resolution of 1 m was used. Moreover, a monitoring system was established in the study area to dynamically monitor and acquire data in real time, including the water level in the drainage pipe network, rainfall, and inundation. Eventually, the monitoring system provides observed data for validating the model. The findings from this study indicate that inundation and the drainage process can be effectively computed by the model. The GPU accelerates the simulation time approximately 2.9 times faster than real time. This study proposed a novel approach for disaster prevention and the mitigation of urban flooding. In addition, the system can produce a data set to help validate numerical models for small-scale urban flood processes.

This study explores the impact of urbanization on stream channel erosion. The effect of changes in flow regimes and the associated changes in hydrodynamic forces were studied using monitored data for nine outfalls in Fairfax County, Virginia. The Stormwater Management Model (SWMM) simulated the flow and velocity hydrographs at each outfall site, using continuous rainfall records for the period 2004–2014. The cumulative stream flow work was used to study the impact of change in hydrology and hydraulics on the dynamics of channel evolution. The cumulative stream flow work at each section was determined by integrating the stream power (discharge time velocity) with time. The results show heterogeneity in the responses of the downstream channel reaches. However, the aggregation of results shows that the cumulative stream flow work accounts for about 70% of the variance in the observed changes in stream cross sectional areas. The cumulative stream flow work was suggested as a method for sizing stormwater detention facilities to meet the channel protection criteria for adequacy of outfalls. The advantage of the method is that it accounts for both the generated runoff at point of discharge and the connectivity of the site to the downstream watercourse. The study suggests that the channel protection criteria for adequacy of outfall should be based on limiting the post-developed cumulative stream flow work to its pre-developed conditions.