Journal of Hydrologic Engineering

Efficient climate change detection is vital to mitigate the ill effects of climate change. Several methods of change point detection have been proposed earlier that do have some limitations. Nonparametric Mann–Whitney–Pettit (MWP) and signal-to-noise ratio (SNR) methods for climate change point detection have been widely accepted and followed by researchers. SNR method is considered better than other parametric and nonparametric change point detection methods as it considers the natural internal variability of the variable to detect significant change. Here, a simple yet powerful climate change detection method called noise-based change point (NBCP) is proposed which is an advancement of the SNR method. The performance of the NBCP method was compared with the SNR and MWP methods for a synthetically generated series as well as real-world data. Different scenarios for synthetic series were considered by introducing varying trend magnitude, variance, and location of change point. The scenarios signifying limitations of SNR as well as MWP methods were also identified. The performance of MWP was not found satisfactory for climate change point detection as it highly depends on the location of the actual change point and performs well only when the change point occurs in the middle of the series. The NBCP method detected the change point earlier than SNR for all generated scenarios as well as real-world data. This study indicates that the NBCP method could be a better option to early detect climate change points and is more consistent than both SNR and MWP methods.

This paper aims to investigate the safety problems related to coal mining under a loose aquifer, and uses the Factor Analysis (FA) and Fisher discriminant methods to analyze and assess the risk of a water-sand inrush under such conditions. Based on 26 sets of the collected geological parameters of the working faces of coal mines under loose aquifers, a comprehensive analysis shows that the risk factors of a water-sand inrush under a loose aquifer are mainly: the thickness of the bottom aquifer; the water pressure; the height of the caving zone; the thickness of the bedrock; the thickness of the bottom clay layer; the water abundance of the aquifer; and the water inflow at the working face. These factors were used as discrimination factors to test and predict the results by using the back-substitution estimate method. The results show that: the water pressure and the thickness of the bottom clay layer have a major influence on the risk of a water-sand inrush under a loose aquifer; and the predicted results of 23 sets of samples are consistent with the results from actual engineering practice, while the results of the remaining three sets are inaccurate. Accordingly, the error rate of the model is 0.115. Therefore, the FA-Fisher discriminant analysis model is highly reliable when conducting a risk assessment of a water-sand inrush under a loose aquifer, and can provide a scientific reference for the safety evaluation of coal mining under such conditions.

The hyporheic zone connects groundwater and surface water allowing active exchange of water and solute, and thus plays a vital role in many hydrological and ecological processes. Many sophisticated high-dimensional models have been developed to study the hyporheic zone, but they require extensive site information and may involve significant time and effort to set up. This emphasizes the need for a preliminary characterization. This paper proposes an analytical model based on the Rankine body method to estimate the maximum hyporheic depth using minimal site information. The hydraulic conductivity of the alluvium is not needed, and for thick alluvium it is not necessary to know the location of any underlying impervious layer. Applied to an idealized pool-riffle sequence, the proposed method is able to calculate the maximum hyporheic depth to within 20% of those from a two-dimensional (2D) reference numerical model for a wide range of conditions. Comparable results are also achieved with other independent studies for hyporheic zone under various bedforms, surface flows, and alluvium conditions. The proposed method provides a quick estimate of the hyporheic depth, which would be useful in the design of a field data collection program and/or a numerical modeling study.

The study of solute transport processes in open channel flows is important for water quality management and environment protection. Solute transported in open channel flows is usually uncertain because of the underlying stochastic flow and uncertain solute source/sink conditions. Here a stochastic one-dimensional nonreactive transport model based on the Fokker-Planck equation (FPE) approach is developed. The governing equation for the developed stochastic model is a two-dimensional FPE that can explain the effect of both uncertain flow fields and solute source/sink conditions on the stochastic solute transport behavior. The proposed model can provide the spatiotemporal probability density function (PDF) of the solute concentration. Ensemble mean and standard deviation behavior in space and time can then be easily obtained through the corresponding PDF. A numerical experiment is conducted to demonstrate the capabilities of the two-dimensional stochastic transport model. The two-dimensional ULTIMATE QUICKEST finite difference scheme is applied to solve the FPE. Monte Carlo simulation is performed to validate the results obtained from the proposed approach. The validation results indicate that the proposed FPE approach can capture the complete probabilistic dynamics of a one-dimensional solute transport system under uncertain flow fields and solute source/sink conditions by providing the evolutionary PDF of solute concentration in both time and space in a computationally efficient way.

The common strategy for real-time streamflow forecasting involves a precalibrated rainfall-runoff model for streamflow simulation together with a statistical postprocessing model of simulation errors for updating simulated streamflow. Recognizing both autocorrelation and heteroscedasticity inherent in the simulation errors of rainfall-runoff models, the autoregressive-generalized autoregressive conditional heteroscedasticity (AR-GARCH) model is introduced as a statistical postprocessing model of simulation errors in this study, in which the AR model is used to forecast the mean process of simulation errors, and the GARCH model to forecast the variance process of simulation errors. For investigating how well incorporating exogenous variables that contribute to heteroscedasticity improves update accuracy by comparing the GARCH model with and without exogenous variable, in this study, two rainfall-runoff models (one lumped and one distributed) have been chosen to simulate streamflow in three basins. Case studies show that (1) the AR-GARCH model with an exogenous variable showed advantages over AR-GARCH without exogenous variables through both increased forecast accuracy and reduced uncertainty during the validation period, and (2) more than 90% of error heteroscedasticity is due to the internal variable and less than 10% is due to the exogenous variable in this study. Only one exogenous variable was considered in this study, so further research is necessary to identify more exogenous variables which may have a greater contribution to simulate error heteroscedasticity. In conclusion, the developed AR-GARCH model with exogenous variables has the flexibility to deal with the characteristics of error series including autocorrelation and heteroscedasticity. Although the application of this study focused on streamflow forecasting, the developed methodology may be generalized and implemented in other applications to time-series data with complex errors.

Pumping tests are commonly used to estimate aquifer parameters during field work. Classical pumping tests are mainly classified into two general types: constant-rate pumping tests and constant-head pumping tests. However, pumping rates can vary under different circumstances. In this paper, a semi-analytical solution is developed for an exponentially decaying-rate pumping test in an aquitard-aquifer system considering the effects of a finite-thickness skin. The solution was obtained in the Laplace domain and inverted numerically using the Stehfest method. The solution could be simplified to some special cases and was also compared to several previous work. The results show that drawdowns have asymptotic values similar to those of constant-rate pumping tests at early and late times but decrease for a short period at intermediate times. The presence of aquitards makes the drawdowns stable initially because larger transmissivity in aquitards generally leads to smaller drawdowns. It was also observed that the well skin had a significant impact on drawdown values within the skin zone but no impact in the formation zone. The new solution was validated by fitting it with aquifer parameters obtained from a field pumping test in in Jianghan Plain using a particle swarm optimization algorithm. This study provides new insights into aquifer parameter estimation and the impact of variable-rate pumping tests on drawdown phenomena, particularly in aquitard-aquifer systems.

Evapotranspiration (ET) is an important factor in water resource management. This research investigated the performance of four optimization algorithms to hybridize adaptive network-based fuzzy inference systems (ANFIS) models as follow: ANFIS with imperialist competitive algorithm (ANFIS-ICA), ANFIS with biogeography-based optimization (ANFIS-BBO), ANFIS with teaching-learning–based optimization (ANFIS-TLBO), and ANFIS with invasive weed optimization algorithm (ANFIS-IWO). The hybridized algorithms were used to predict reference evapotranspiration (${\mathrm{ET}}_o$) values in Kerman synoptic station. Six observed variables, including mean air temperature ($T_{mean}$), bright sunshine hours ($SSH$), solar radiation ($R_s$), mean speed of the wind at 2-m height ($U_2$), pan evaporation ($E_{pan}$), and three estimated variables, including extraterrestrial radiation ($R_a$), saturation vapor pressure ($e_s$), and actual vapor pressure ($e_a$) were utilized to develop hybrid models. The results showed that the accuracy of hybrid models by using $T_{mean}$, $U_2$, $e_s$, and $e_a$ was better than those using all required variables for developing the FAO-Penman-Monteith (FAO-PM) equation. Among the hybrid models, the ANFIS-ICA with respect to $\mathrm{R}=0.99$, $\mathrm{RMSE}=0.5$, and $\mathrm{NSE}=0.98$ was considered the superior model. A sensitivity analysis has been done to assess the impact of inputs on the output of the superior model. $E_a$ and $T_{mean}$ had the highest and lowest effect on ${\mathrm{ET}}_o$ prediction, respectively. Finally, ${\mathrm{ET}}_o$ values were estimated by relatively new empirical equations and compared with FAO-PM equation. It was observed that the capability of hybrid models was more than the empirical equations in estimation of the ${\mathrm{ET}}_o$ values.

In recent decades, several parts of the world have been facing severe droughts and frequent floods against a background of anthropogenic influences and changes in climatic variables. For efficient management and adaptation measures, it is important to understand the relative effect of climate variability and anthropogenic influences. The Nagavalli River Basin (NRB), located in India, has experienced significant changes in hydroclimatic variables and land use in the last decade and therefore serves an ideal basin for investigating the attribution of rainfall variability and land-use changes. In this study, the characteristics of precipitation, temperature, and streamflow were analyzed for the period 1970–2012. It was observed that there is a significant increasing trend in precipitation and streamflow. Further, there have been substantial land-use change in terms of scrubland conversion. Investigations on rainfall-runoff coherency using wavelet coherence showed that there were noteworthy changes during the periods of 1991–2001 and 2002–2012. The contributions of land-use and rainfall variability to the changes in the streamflow were quantified using a semidistributed hydrological model (Soil Water Assessment Tool). The results showed that for the whole NRB, the variations of mean annual streamflow in 2002–2012 were primarily affected by rainfall variability with reference to 1990s, whereas human activities played a complementary role. The quantitative assessment revealed that rainfall variability resulted in an increase in runoff by 103 mm in 2002–2012 for the whole catchment, accounting for 41.52% of runoff changes relative to the 1990s. However, land-use changes are responsible for a decrease in runoff by 59 mm during the period of 2002–2012, which accounts for $-23.54\%$ of runoff changes. Overall, it was observed that the agricultural intensification in terms of scrubland conversion counteracted the effect of rainfall variability, resulting in the reduction in the margin of increase in streamflow.

Hydrologic analyses are used in dam safety evaluations to determine the flow a dam must pass without failing. Many current guidelines model flood runoff solely by an infiltration-excess mechanism. Saturation-excess runoff and subsurface stormflow mechanisms are known to be important for common events in forested regions, but few studies have analyzed their role in extreme events. The objectives of this study are to determine the active runoff mechanisms for large historical storms, design storms in the Colorado Front Range, and propose methods to model these mechanisms that dam safety consultants can use. Hydrologic models are developed for five basins to simulate historical flood events in 1976 and 2013 as well as various design storms. The model results (and available in-situ soil moisture observations) show that the entire soil layer approached saturation during the 2013 storm, which had a long duration and low rainfall intensities. Thus, saturation-excess runoff was likely the dominant mechanism. In contrast, the modeled soil layer does not approach saturation for the 1976 storm, which had a short duration and high rainfall intensities, so infiltration-excess runoff was likely the dominant mechanism. Similarly, infiltration-excess runoff dominates for short duration (2-h) design storms, while saturation-excess dominates for longer (6-h and above) design storms in the Front Range basins.

The linking of onsite rates of sedimentation and river water quality represent a major research need within the field of erosion, and it is considered to be an important scale problem in sediment yield prediction. Understanding sedimentation assessment and the sediment transport process is critical because it helps to estimate the amount of sediment moving through a river. Therefore, different methods are examined in this study to predict the suspended sediment load based on the collected data from various hydrometric stations. In order to calibrate and validate the modeling results, the suspended sediment load and flow data of the Abiaca Creek watershed over the period 1991–2003 are used. In addition, normal root mean square error, the Nash–Sutcliffe efficiency coefficient, the correlation coefficient, the Akaike information criterion (AIC), and the fitting between the observed and the predicted values of suspended sediment are calculated using a number of well-known statistical indexes. Furthermore, several stochastic methods with constant and variable coefficients, sediment rating curves, decomposition additives, and multiplicative are applied to predict the suspended sediment load. The results of this study show that stochastic modeling with variable coefficients has higher accuracy than the other methods in which the indicator of the normal root mean square error (NRMSE) is 0.27 and 0.37 for the simulation and prediction of the suspended sediment load, respectively.

A representative of meteorological data-constrained basin, Ayeyarwady, in Myanmar, Southeast Asia, is set for flow simulation and forecasting at 15 locations using a range of hydrological modeling approaches: conceptual lumped (GR4J), hybrid-lumped [Identification of unit Hydrographs And Component flows from Rainfall Evapotranspiration and Streamflow Catchment Wetness Index (IHACRES CWI)], semidistributed [Hydrological Engineering Center-Hydrological Modeling System (HEC-HMS)], and relatively distributed [Soil and Water Assessment Tool (SWAT)]. Using daily rainfall data from 51 surface rainfall stations (over an area of approximately $\mathrm{400,000}{\mathrm{km}}^2$) and coarse monthly evaporation inputs from global sources, the models are calibrated (validated) against observed flows for 2001–2009 (2010–2014) using the performance indicators Nash-Sutcliffe efficiency (NSE), percentage bias (PBIAS), RMSE-observations standard deviation ratio (RSR), and volumetric efficiency . The developed models were then integrated with rainfall forecasts from the Weather Research and Forecasting Model for 2015–2018 and assessed for biases against observed flows. The NSE values were favorable for GR4J (median $\mathrm{NSE}=0.9$), followed by IHACRES ($\mathrm{NSE}=0.86$), SWAT ($\mathrm{NSE}=0.81$) and HEC-HMS ($\mathrm{NSE}=0.77$) during calibration and GR4J ($\mathrm{NSE}=0.87$) and the latter three ($\mathrm{NSE}=0.83$) during validation. Lumped models were found to have comparable, albeit better in simulating low, median, and high quantiles of flows during both calibration and validation periods, compared to other models of varying complexity set for the study basin. The hydrometeorological coupling also revealed that GR4J yielded the least while HEC-HMS yielded the highest biases (up to 30-fold at some stations) in daily flow forecasting. The analysis suggested that while process-based and relatively complex models may exhibit better performance in data-rich basins, simple conceptual models like GR4J are useful for daily flow simulation and forecasting in data-constrained basins of the region.

Small dams are infrastructures that regulate water supply for multiple users and play a key role in the agricultural development of the Brazilian savannah region known as the Cerrado. Evaporation is one of the major components of the hydrological cycle of small reservoirs, and should be better quantified. Studies based on machine learning techniques usually adjust models based on large datasets, which are frequently unavailable in developing countries. This study adjusted and evaluated the performance of different evaporation machine learning models that were regressed on a very small dataset and for restrictive scenarios. The performance of each model was assessed with five climatic input combinations. The performance of the random forest models was one of the better for the input combinations, and was considered to be one of the more robust machine learning techniques among those assessed for estimating evaporation from a small reservoir in the region. The Penman (benchmark) equation performed worse, as it overestimated evaporation by 14.7% on average. Strategies for improving the performance and applicability of models and overcoming data scarcity in remote areas are further discussed.

In frequency analysis of point precipitation extremes, one approach that has been suggested is to segregate precipitation data by storm type. Thus, for short-duration (e.g., 2-h) extremes, only observations from convective storms would be considered in the analysis. The supposition informing this choice is that only convective events are capable of generating truly extreme 2-h precipitation depths. In this paper, annual maxima of 2-h point precipitation are evaluated for a region in eastern Tennessee using a storm-type database that identifies convective events. The use of storm typing contrasts with the standard procedure of using data from all storms to develop annual maxima series. This paper presents comparative frequency analyses of convective event and overall annual maxima series. For extremely low-probability events, predicted regional quantiles are slightly lower for convective event maxima than for annual maxima. However, differences between quantiles at individual sites vary widely. The convective event maxima series tend to have lower means, higher coefficients of variation, and slightly lower skewness (with great variability) than annual maxima series. Differences between extreme quantiles for convective event and annual maxima are correlated with differences in skewness. Overall, the differences between the regional results obtained from the convective event and annual maxima used in this study of 2-h precipitation are small, but differences at individual sites vary widely.

A probabilistic approach to flood susceptibility determination is investigated in this paper. Nonclimatic flood risk factors are considered and used to develop a base for flood potential prior to investigating the potential impacts due to climate change, which is mapped within Currituck County, North Carolina, using a multivariate logistic regression. Several site characteristics (elevation, slope, curvature, land cover, impervious surface, distance to water, tree density, surficial materials, and soil drainage) were identified as potential flood risk factors and divided into classes. An attempt was made to correlate these flood risk factors to the spatial extent of the FEMA 100-year Special Flood Hazard Area (SFHA). Site characteristics of an equal number of locations ($\sim \mathrm{43,000}$) from within and outside of the SFHA were used to train the statistical model. Logistic regression was then used to estimate coefficients for each class of each flood risk factor. It was found that elevation, land slope, and soil drainage class have the greatest correlation to flood inundation in the county. The coefficients were then used to create a logistic regression equation from which probabilities of flooding were estimated for the entire county at a 30-m resolution. Large areas of the central portion of the county and the northern Outer Banks were classified as very high risk (80%–100%) and high risk (60%–80%). While all other high-risk areas corresponded to areas already located within the SFHA, the Outer Banks encompassed areas that are not included in the SFHA. Critical infrastructure related to basic human needs, such as education, safety, and health, located in this high-risk area warrant additional attention regarding potential flood mitigation efforts.

Statistics indicate that granular rock mass frequently acts as the flow channel and cause water inrush disasters under high water pressure in underground mining and tunnel engineering. The flow velocity of water inrush is so high that the flow behavior shows nonlinearity and cannot be described by Darcy’s law. During the water inrush, the non-Darcy flow behavior associated with the porosity of granular rock mass is constantly changing due to seepage erosion. Therefore, this paper aims at the non-Darcy flow behavior of the granular limestone with a wide range of porosity (from 0.242 to 0.449). The non-Darcy flow properties were experimental investigated under steady-state condition by a self-designed apparatus. Based on the experimental data, an empirical hydraulic conductivity-porosity relation together with the expression of inertial coefficient was proposed. Then the non-Darcy flow regime was quantitative studied by the hydraulic conductivity-based Reynolds number. And the Forchheimer number associated with porosity was used to determine the critical flow velocity of the non-Darcy flow. From the point of view of seepage, two reasons for water inrush frequently occurred through granular rock mass with high porosity were finally discussed.