Journal of Hydrologic Engineering

New optimization and variants of quantile-based methods are developed for bias corrections of monthly and daily general circulation model (GCM)-based statistically downscaled precipitation data. These methods use optimization formulations involving several linear and nonlinear corrections with single and multiple objectives and integrate artificial neural networks (ANNs) with quantile matching (QM) methods. The proposed methods were evaluated at 18 rain gauge sites in Florida using several error and performance measures. Downscaled monthly precipitation data are derived from two statistical downscaling models, including a support vector machine (SVM)-based method developed in this study. Downscaled daily precipitation data from two different climatic zones are also used for the evaluation of bias-correction methods. The methods are assessed based on several performance and error measures, along with their ability to replicate all the moments of the distribution. The selection of the best method among several others for a specific site was found to be dependent on specific performance and error measures adopted for evaluation. The proposed methods not only replicated the observed precipitation data distributions but also minimized the quantitative errors between observed and downscaled precipitation data sets, which could not be accomplished using existing methods. ANN-based methods performed better than QM-based ones in replicating extreme precipitation indices at a daily temporal scale. The multiobjective optimization methods require careful selection of objectives and assignment of weights, with the latter heavily influencing the performance of methods. Variation in performances of methods is observed when methods are calibrated with varying baseline periods with a constant length of test data.

The purpose of this paper is to introduce an effective way to solve the problem of nonstationary data generation. Empirical mode decomposition (EMD) algorithms have been widely used in data diagnosis. A new EMD-based data synthesis method is proposed. The method utilizes the recombination of the intrinsic mode function (IMF) of the segmented data, as well as the characteristics of the residuals, to generate the data. This article takes the 100-year monthly temperature and rainfall data of Tainan, Taiwan, as an example. The Kwiatkowski–Phillips–Schmidt–Shin (KPSS) test is applied in the paper to verify the stationarity of the generated data. The EMD-based data synthesis effectively shows its applicability and provides new ideas for nonstationary data generation.

Earth-fill dam overtopping could pose significant threats to public safety if the necessary reliability assessments are not taken into consideration in the design process. Dam overtopping is caused by several hydrometeorological random variables, including intensive rainfall, flood or flash flood, and wind coupled with waves. Because random variables are the main effective variables, a stochastic processes can be used to capture the leading failure factors. This study proposes a load-resistance-based approach to assess the reliability of Jamishan Dam, Iran, overtopping by considering several random variables simultaneously. Uncertainty sources such as (1) parameters of the hydrological model (e.g., loss, base flow, and unit hydrograph parameters), (2) hydraulic parameters (coefficient and length of spillway parameters), (3) initial reservoir water level, (4) wave height (e.g., wind setup and wave run-up parameters), and (5) rainfall parameters, are quantified using the Monte Carlo simulation (MCS) technique. Besides, the impact of stochastic sources and dam dimension design on overtopping reliability is considered through two indices, called reliability relative difference index ($R_d$) and reliability variation index ($R_v$). The uncertainty of hydrological parameters is quantified using the generalized likelihood uncertainty estimation (GLUE) method, which helps to extract their posterior probability distribution functions (PDFs). Reliability of dam overtopping has negligible sensitivity on the performance measures and behavioral threshold value of the GLUE approach. Results indicate that rainfall depth ($R_d=2.04$) is the most significant random variable affecting the overtopping reliability, with hydraulic random parameters ($R_d=0.1$) as the minimum level. Loss parameters ($R_d=0.47$) have the highest impact on overtopping reliability compared with other hydrological parameters. Overall, the significant of the stochastic sources such as meteorological parameters (e.g., rainfall depth, duration, and pattern), and hydrological parameters (e.g., loss parameters) on dam overtopping should be further studied to obtain reliable perspective for decision makers.

A stormwater drainage system is an important element in the urban infrastructure network. Under the conventional concept, an urban drainage system is designed to collect and pass the extreme storm event when sizing channels, street gutters, and storm drains. An extreme-event drainage system is often designed to target the runoff peak flow reduction, but not the runoff volume. Under the concept of low-impact development (LID), an urban drainage system shall be designed to mimic the nature flow system that conveys the full spectrum of runoff flows. In this study, it is suggested that the full spectrum of runoff flows should be divided into three categories: micro (up to a 6-month event), minor (up to a 10-year event), and major (up to a 100-year event). A 3M Cascading Flow System is proposed using three preselected storm events—micro, minor, and major—to design an urban waterway to be a three-layer flow corridor. In practice, porous devices are placed at the upstream of the entrance into the drainage system to capture the initial runoff volume up to the micro event for on-site runoff disposal, streets are utilized to safely collect and quickly remove minor to major storm events within the allowable gutter depths, and a detention basin is installed at the system exit to reduce the peak flows to the predevelopment condition. In this study, it is proposed that the flow-frequency curve should be used as the basis to quantity the effectiveness of a drainage system. The concept of a 3M Cascading Flow System provides a guide to conduct the regional drainage planning that can be further refined for details at the final design stage.

In the field of soil hydrology, the Richards equation is commonly used to model water flow in unsaturated soils. The high nonlinearity of the Richards equation makes it very challenging to solve analytically for situations that are meaningful in practical applications. In this paper, an exact and simple analytical solution of the Richards equation under gravity-driven infiltration and constant rainfall intensity is derived. First, the solution is presented under Torricelli’s law, which mimics the soil hydraulic conductivity function and describes the emptying or filling process of a nonlinear water reservoir. Then, following a similar approach, the solution is extended to the Brooks and Corey soil hydraulic conductivity function, which is generally considered to well describe soil hydrological characteristics. The approach followed in this study is a simple hydraulic approach; therefore, the derived solutions are not affected by uncertainty as long as the hypothesis of the gravity-driven infiltration is satisfied for the selected soils. A comparison with the solution numerically derived by the Richards equation for which the gravity-driven assumption is relaxed is performed and discussed. Interestingly, the comparison indicated that the suggested solution delimits the solutions domain of the Richards equation.

Rainfall frequency analysis has been routinely adopted for the estimation of design rainfall for a specific return period. Annual maximum rainfall data are generally used for frequency analysis in practice, but the parameters of the probability distribution are estimated from the limited data that are often available back to the 1970s in many regions, including South Korea. As an alternative, this study aims to utilize century-long the ECMWF twentieth century reanalysis (ERA-20C) daily precipitation data, provided by the European Centre for Medium-Range Weather Forecasts (ECMWF). To reduce the systematic errors in the reanalysis data, a quantile delta mapping method using a composite gamma-Pareto distribution (QDM-GP) is introduced that can better represent temporal trends and extreme events compared with stationary quantile mapping (SQM). In addition, the degree of uncertainty reduction in the estimation of design rainfall is evaluated using bias-corrected ERA-20C within a Bayesian modeling framework. Finally, the bias-corrected data are applied to explore the spatiotemporal change in design rainfall in South Korea in the 20th century. To investigate changes in design rainfall under the nonstationary assumption, this study estimates the design rainfall using data from three different periods (1900–1936, 1937–1973, and 1974–2010). It is found that QDM can substantially reduce the bias in annual maximum rainfall (AMR). The uncertainty ranges of the design rainfall using the bias-corrected ERA-20C reanalysis data are generally within the design rainfalls in the observed, suggesting that the use of bias-corrected reanalysis data can reduce uncertainties in design rainfall by increasing the sample size. Furthermore, this study explores the role of bias-corrected rainfall for uncertainty reduction in design rainfall via three different experiments in the context of prior information within a Bayesian framework. In the experimental study, it is concluded that the uncertainty reduction in design rainfall can be mainly attributed to the use of a prior distribution for the shape parameter, informed by long-term reanalysis data. Moreover, a significant spatiotemporal change in design rainfall is observed over all of South Korea. The significant change in design rainfall is mainly attributed to the recent increase in rainfall intensity, leading to a potential increase in flood risk in most areas.

To understand rain microphysics at the scale of 100 km over the Yangtze River Delta in East China, the characteristics and small-scale variations of raindrop size distribution (DSD) are investigated using data from two-year observations by five OTT Parsivel disdrometers. The characteristics of DSD are quantified and compared for three rainfall classifications: convective, stratiform, and mixed rainfall. Localized $\mu -\lambda$ and $Z\text{-}R$ relationships are further derived based on the quantified DSD. For overall DSD, convective rain shows significant dispersion between different stations, while there is no obvious difference in stratiform and mixed rainfall. The DSD parameter shows that the convective rainfall tends to be enhanced but the stratiform and mixed rainfalls tend to be weakened by urbanization. The convective rainfall has the largest variability in number density ($N_w$) and mass-weighted mean diameter ($D_m$) followed by mixed rainfall, and stratiform rainfall has the smallest dispersion. Besides, both the $\mu -\lambda$ and $Z\text{-}R$ relationships vary with rainfall types. These findings not only help to better understand precipitation physics but also contribute to the quantitative precipitation estimation (QPE) of weather radar, hydrological modeling, and water resource assessment.

Storm-event analysis (SEA) is a method of extracting information from rainfall data by defining rainfall events as storm events. A profile-based model in SEA defines storm events using the interevent time definition (IETD), which is the minimum value between consecutive storms. Based on statistical hypothesis tests done on the difference of the average annual total frequency of storm events obtained when the value of IETD increases, it is found that 6 h is suitable as an IETD value for storms in peninsular Malaysia. In this study, two storm indices, storm intensity (SI) and storm concentration index (SCI), are used to summarize the relationship between storm duration and storm amount as well as to cluster storms. The Ward method is used for cluster analysis, and four distinct regions are found. By categorizing storm events in the four regions, it is shown that the differences are greatly influenced by the monsoon and intermonsoon seasons in peninsular Malaysia. From the result in this study, further analysis can be reduced by performing regionalized frequency analysis instead of at-site frequency analysis.

Coastal aquifers are threatened by saltwater intrusion (SWI), which leads to degradation of groundwater quality by raising salinity. SWI occurs due to human activities such as overabstraction or natural processes such as sea-level rise. SWI control can help in protecting coastal aquifers from deterioration. This paper presents a numerical study using the simulation of three-dimensional variable-density ground-water flow (SEAWAT) code to investigate and control SWI in Gaza aquifer, Palestine, considering overpumping due to population growth and sea-level rise. Application of different scenarios to investigate SWI in Gaza aquifer showed that the intrusion of the equi-concentration line at 35,000 ppm, which is the interface between freshwater and saline water that reached 3,177 m in 2010, were compared to the field investigation in 2010, and by 2055 it will increase to 3,186, 3,518, 5,619, and 6,523 m due to sea-level rise, reducing recharge, and overpumping (percentage changes of 0.29%, 10.74%, 76.87%, and 105.32%, respectively). The results reveal that the intrusion length will double in 2055, and overpumping is considered the main cause for such an increase. Three management scenarios were used to control SWI in Gaza aquifer, including the abstraction of brackish water, recharge of the aquifer using treated wastewater, and a combination of both. Brackish water abstraction led to retardation of SWI at 3.5 km; recharge using treated wastewater led to retardation of SWI at 3.6 km, but a combination of both scenarios led to a retardation of SWI at 3.7 km. The results revealed that recharging the aquifer using treated wastewater could retard intrusion of saline water more than abstraction of brackish water, while the combination of both scenarios gave a higher retardation rate for saline water toward the sea. Aquifer recharge using treated wastewater is considered an effective tool to control SWI and protect coastal aquifers from deterioration.

This paper presents an assessment of the calibration and transfer of artificial neural networks (ANNs) to simulate streamflow at Brazilian Atlantic Rainforest basins. Primary data consisted of rainfall and a streamflow daily series (32 years in extent) of 12 subbasins of the Itapemirim River basin (IRB). First, data from three subbasins were used to adjust three ANNs to estimate daily specific streamflow from input parameters related to rainfall. After, the ANNs were applied to simulate the flows in all other IRB subbasins. The ANNs were able to reproduce the subbasin discharges for which they were adjusted. They also reached satisfactory performance when applied in most of the other subbasins. The obtained results demonstrate that the ANN technique is a viable alternative for simulating flows in regions lacking primary data for hydrological modeling. Besides, calibrating ANNs with subbasin data of an intermediate size or position tends to present a better overall performance than calibrating for the smaller (upstream) or the larger subbasins (downstream).

The Mekong River Basin (MRB) is undergoing rapid dam development, which is altering the river suspended sediment concentration (SSC). In this study, we used satellite remote sensing records spanning 31 years to detect SSC changes (SSC prediction $r^2=0.78$, $\mathrm{RMSE}=21.2\mathrm{mg}/\mathrm{L}$) due to dam development. We focused on the 3S basin of the MRB. We also used satellite data on nighttime lights, which reflect human settlement patterns, and land cover to explain SSC patterns. Our technique allowed for quantification of SSC changes due to dam construction (e.g., $+120\mathrm{mg}/\mathrm{L}$ near basin outlet), reservoir sediment trapping (e.g., $-108\mathrm{mg}/\mathrm{L}$), deforestation, and human settlement (e.g., $+117\mathrm{mg}/\mathrm{L}$ near impacts). Our technique also demonstrated how the SSC of the 3S rivers compared to that of the Mekong mainstem over time (e.g., from $\sim 13\%$ to 100% greater). Our comprehensive analyses of SSC records with dam development indicate that SSC changes will continue with ongoing dam and landscape development in the MRB. From a hydrologic perspective, SSC monitoring will be imperative for effective sediment and water management. Our satellite-based approach answers critical sediment needs of improved monitoring and adaptive management throughout the MRB and other global locations for practitioners who are engaged in real-world management of their river basins.

Regional flood frequency analysis (RFFA) techniques are used in hydrological applications for estimation of design quantiles at ungauged sites or catchments with sparse observational records. The index-flood method, a popular approach for RFFA, is based on the assumption that the flood records within a homogeneous region are identically distributed, except for a site-specific index flood. Because of rapidly changing land-use patterns, human interventions, and climate change, recent studies proposed extension of the index-flood method to account for nonstationarity in flood records. This work compared index-flood–based nonstationary RFFA techniques, in both synthetically generated and real-world homogeneous regions, with sites marked by significant trends in flood records. From the data used in the analysis, it is evident that two recently proposed transformation-based approaches are more suitable compared to other methods, and can capture time-varying behavior of floods effectively.