Journal of Hydrologic Engineering

Forum papers are thought-provoking opinion pieces or essays founded in fact, sometimes containing speculation, on a civil engineering topic of general interest and relevance to the readership of the journal. The views expressed in this Forum article do not necessarily reflect the views of ASCE or the Editorial Board of the journal.

The USEPA’s Total Maximum Daily Load (TMDL) program under the Clean Water Act requires taking into account the critical conditions for stream flow, loading, and water quality parameters to develop TMDLs. However, properly defining a waterway’s critical conditions is one of the key challenges in developing a TMDL. Consequently, many current TMDLs address this aspect only in a cursory manner. This paper presents a review of critical condition modeling and analysis approaches and illustrates the strengths and weaknesses of available methodologies to define critical conditions with a view to provide guidance for TMDL development and implementation. Four major approaches for critical condition modeling and analysis are typically employed in TMDL studies: low-flow analysis using steady-state models, continuous simulation using dynamic models, statistically based load-duration curves, and the event-based critical flow-storm approach. By understanding better how to define critical conditions, water quality professionals can develop more-meaningful and improved estimates of TMDLs and formulate appropriate strategies to control pollution from point and nonpoint sources at a watershed level.

A novel wavelet-artificial neural network hybrid model (WA-ANN) for short-term daily inflow forecasting is proposed, using for the first time Tropical Rainfall Measuring Mission (TRMM) data together with inflow data, which were transformed using mother-wavelets to improve the model performance. The models were assessed using the inflow records to a Brazilian reservoir named Três Marias, located in the São Francisco River basin, and daily rainfall estimates from the TRMM both for the period of 1998–2012. Several combinations of inputs for both regular and hybrid artificial neural networks (ANN) were assessed to forecast inflows seven days ahead, and it was proved that the WA-ANN had a superior performance. Even the WA-ANN model, which uses only the approximation at level three of rainfall data, provided a higher performance than the regular ANN, which uses the raw inflow data [$r$ increase 16%, Nash–Sutcliffe model efficiency coefficient (NASH) increase 35%, and root-mean-square deviation (RMSD) decrease 47%]. It was also found the best model was the WA-ANN with transformed rainfall and inflow data as input ($r$ increase 20%, NASH increase 44%, and RMSD decrease 69%).

Water quality models are critical tools for the development of total maximum daily loads (TMDLs) and for the evaluation of water quality management alternatives by stakeholders and state environmental protection agencies. Currently there is a large availability of water quality models that can be used to support TMDL studies, and the selection of a particular model requires a good understanding of the model limitations, capabilities, and data requirements. The ASCE/Environmental and Water Resources Institute (EWRI) TMDL Analysis and Modeling Task Committee was established in part to produce guidance documentation to help modelers to identify and implement existing modeling approaches to address some of the most important causes of water quality impairment in the United States, including eutrophication, toxic chemicals, and metals. This paper presents a review of existing mathematical models to evaluate eutrophication processes, including carbon and nutrient cycling, phytoplankton dynamics, and dissolved oxygen availability, as well as a review of mathematical models to simulate the fate and transport of toxic chemicals and mercury. The paper also discusses the main capabilities and limitations of 11 widely used water quality models in the United States. Modelers can use this information to support more informed model selections and to facilitate an effective and successful application of models in TMDL studies.

Automatic rain gauges, such as tipping-bucket gauges, have generally replaced manual rain gauges in monitoring networks around the world. Both are susceptible to wind-induced errors. Shielding has been used to reduce these effects. The newly installed tipping-bucket rain gauges are not currently shielded in Cathedral Peak. The aim of this study was to improve understanding of the influence of shield and gauge design on rainfall measurement accuracy. The study was conducted at the Cathedral Peak Meteorological Station in the Drakensberg mountains, KwaZulu-Natal, South Africa. Three rain gauges, a historical shielded Snowdon, shielded, and ground-level tipping-bucket rain gauges, were installed alongside an unshielded tipping-bucket rain gauge. At the monthly scale, the Snowdon and unshielded rain gauges recorded 7%–8% and the shielded rain gauge 12% less than the ground-level rain gauge. At the event scale, the unshielded rain gauge recorded more rainfall for 58.47% of events than the shielded. Peak rainfall intensity was identified as the significant predictor variable affecting rainfall volume recorded by the rain gauges. From this study it was concluded that the use of a Nipher shielded rain gauge in the Cathedral Peak research catchments is ineffective.

A database joining individual earthen dam breach failure studies is assembled and reanalyzed across all aggregate observations. Conventional regression methods are employed along with newer predictive approaches to estimating peak discharges resulting from an earthen dam failure. Goodness of fit is quantified through relative standard error and relative bias. These measures are computed and presented for previous predictive equations. Numerical optimization techniques are used to calibrate power law functions of one, two, and three predictors to estimate peak discharge from the aggregate database. Findings show that equations calibrated from the aggregate database have better goodness-of-fit metrics than those determined from their earlier, individual data sets. Improvement in relative standard error varies from essentially zero to as much as 50%. Two similar innovative techniques are applied to the aggregate database: region of influence (ROI) and k-nearest neighbor (kNN). Both of these approaches identify a subset of most similar observations from the database, given a specific test location. The ROI approach performs poorly in prediction mode, uniformly producing relative standard errors that are greater than the originally calibrated equations and that often exceed 100% of the standard deviation of the observations. Smaller relative standard errors are obtained as ROI size increases, contrary to the spirit of this approach. In contrast, the kNN approach performs well, with best results obtained for a simple numerical average of the $k$ nearest observations. The size of the optimum $k$ neighborhood varied from 3 to 29, with 12 being the median value among the cases examined. Regression equation calibration via logarithmic transformation is briefly explored, and the need to limit predictions to the test space within the convex hull of the observations is discussed.

The solution of interactions between groundwater and surface water provides good insight into the mechanism of these water resources. Considering groundwater as a significant source of irrigation and a main freshwater supply for agricultural purposes, a stream–aquifer system is studied to investigate the groundwater level and flow direction. The solution is obtained and applicable for streams fully penetrating the aquifer. The system is solved and a set of expressions for computing level and exchange in the aquifer are derived by means of the Laplace transformation technique, wherein the right end of the aquifer is a constant head boundary and the stream stage at the left boundary decreases exponentially with time. A broad range of sensitivity analysis is carried out to investigate the effects of variations in aquifer parameters on groundwater head. Aquifer response for exponential and sudden drops in stream stage is investigated. A useful expression describing the relation of stream stage variation rate and volumetric exchange is found, which clarifies whether the final volumetric exchange is negative, zero, or positive. For a sudden drop in stream stage, a water divide takes place in the aquifer and therefore flow reversal occurs at the water divide position. It is shown that how and how much the position of the water divide changes depends on changes in aquifer parameters. In addition, it is observed that groundwater level stabilization occurs relatively faster in aquifers with a greater hydraulic conductivity, smaller specific storage, or smaller length.

Probable maximum precipitation (PMP) is used for computing probable maximum flood (PMF), which is then used to design large hydraulic structures. A basin-scale model, based on the Hershfield method, was developed for calculating PMP and PMP uncertainty in this study. It showed that PMP from the proposed method was more consistent with the values reported by the Texas Commission on Environmental Quality (TCEQ). PMP uncertainties were more reliable than those from the previous studies because of the improvement in assumptions and methodology. Utilizing basin-wide data, the proposed method developed equations for frequency factor ($K$) enveloping curves and calculated and mapped 1-h, 6-h, and 24-h duration PMPs for the Brazos River basin, Texas, United States. The PMP values computed by the basin-scale method were smaller than those computed by the Hershfield method but were more consistent with the values reported by TCEQ (TCEQ results were smallest among the three). The average improvements of the difference percentages for 1-h, 6-h, and 24-h duration PMP were 53.84%, 81.04%, and 72.60%, respectively. The longer-duration PMP map showed a declining trend from east coast to west (inland) and was consistent in spatial distribution with the TCEQ map. The PMP uncertainty was determined using the delta and bootstrap methods, which produced consistent results. The uncertainty differences between the delta values and bootstrap values were 10.05%, 10.35%, and 19.02%, and between the Salas values and bootstrap values for each duration were 25.59%, 20.06%, and 27.95%, respectively, which means the delta method results were more supported by the bootstrap method results. Comparison showed that the uncertainty by other studies was smaller because uncertainty caused by $K$ and normality assumption was not considered. Using bootstrap resampling, the distribution of PMP was explored, and then PMPs with 95% confidence intervals were obtained.

The analysis of the effects of human activities and climate variability on the structural characteristics of streamflow series facilitates further exploring the regularity of streamflow change and revealing the process of hydrological cycle response. The principal goal of this study is to establish an analysis system to diagnose the change in structural characteristics of streamflow series via employing suitable complexity measurement methods for specific cases. Meanwhile, an application scope table has been built based on testing and comparing the logicality and antinoise robustness through using logistic mapping series, random series, white noise, and $1/f$ noise for five commonly used complexity measurement methods (approximate entropy, permutation entropy, sample entropy, fractal, and Lempel-Ziv). According to the application scope table, sample entropy is the most suitable method for this specific case. Moreover, the results illustrate that in the prechange period, the complexity of the streamflow series is 0.5176; in the postchange period, the complexity value is 0.3722 if the streamflow is only impacted by climate variability; and its value is 0.2571 if it is impacted by both human activities and climate variability. Climate variability reduces the complexity of streamflow series, and human activities intensify this reduction. A decrease of complexity indicates that the streamflow series becomes more stable and regular.

Estimation of reference evapotranspiration (${\mathrm{ET}}_{\mathrm{o}}$) using air temperature is particularly attractive for places where solar radiation, wind speed, and air humidity data are not readily available. In this study, seven temperature-based (TET) models and the standardized reference evapotranspiration equation for short canopies method were compared. Using only temperature data from the Yucatán Peninsula, México, the Food and Agriculture Organization of the United Nations (FAO)-Penman-Monteith (PMT) model was used to estimate ${\mathrm{ET}}_{\mathrm{o}}$. Results from the temperature-based models are compared with FAO-56 daily ${\mathrm{ET}}_{\mathrm{o}}$ calculations using the Nash-Sutcliffe model efficiency coefficient (NSE), the coefficient of determination ($R^2$), mean absolute bias error (MAE), mean absolute percentage error (MAPE) and root mean square (RMSE). The results show that the noncalibrated PMT expression using temperatures alone produced the best results, with $\mathrm{RMSE}=0.7\mathrm{mm}{\mathrm{d}}^{-1}$. The Hargreaves-Samani calibrated ($\mathrm{RMSE}=0.74\mathrm{mm}·{\mathrm{d}}^{-1}$) and Camargo calibrated ($\mathrm{RMSE}=0.78$) models exhibited the next best performance. RMSE values were as high as $1.56\mathrm{mm}·{\mathrm{d}}^{-1}$ for the other models.

The coincidence of floods on a mainstem and its tributaries may cause significant flood damage downstream of the confluence. In this study, the coincidence risks of annual maximum (AM) floods on Jinsha River and Min River, China, were analyzed using multivariate copulas based on both systematic records and available historical information. Archimedean and elliptical copulas were selected to establish the joint distributions of flood magnitudes and occurrence dates of the two rivers. The coincidence probabilities of AM floods were investigated and compared with monthly maximum floods. The results show that the coincidence probabilities of AM flood magnitudes and occurrence dates are much less than those of monthly maximum floods. Higher coincidence probabilities of AM flood occurrence dates occur in the period from mid-June to early September with three coincidence peaks, while there is only one coincidence peak in each month for monthly maximum floods. Ignoring historical flood information would underestimate the coincidence probabilities of flood magnitudes. The results can provide decision support for cascade reservoir operation on Jinsha River and flood mitigation downstream.

In order to reassess the hydrological drought in the Yangtze River under changing environment, a nonstationary standardized streamflow index was proposed in this study to fit the streamflow series at Yichang station; time and a modified reservoir index were introduced as covariates to assess the effect of reservoir regulation. The copula method was applied for bivariate modeling of drought duration and severity, in which joint and conditional return periods were considered for drought risk assessment. The results indicated that the monthly streamflow series at Yichang station have undergone great changes and the stationary assumption is no longer valid. The drought severity was more severe with its marginal distribution changed from the generalized extreme value to gamma when considering nonstationary properties. The Joe copula was selected for bivariate frequency analysis, and the correlation coefficient between drought characteristics increased in nonstationary models. Using the conditional return period, the model with reservoir index as covariate reported a worse drought condition compared with the stationary model, implying that the reservoirs may deteriorate the downstream hydrological drought at the Yichang station. By contrast, the time covariate may underestimate the drought risk. The nonstationary index is capable for drought modeling in the Yangtze River, and can be a useful tool in further research.