Journal of Hydrologic Engineering

Several types of precipitation models driven by radar data have been developed during the last few years. These models produce a spatially distributed rainfall forecasting suitable for hydrological needs. However, the assessment of their benefits and factors limiting their use in coupled hydrometeorological flood forecasting models (such as the response time of the basin, quality of the rainfall forecast, and rainfall forecast lead times) have not yet been analyzed in detail. This paper uses a case study to analyze the gain provided by a precipitation model for real-time flood forecasting in the case of a short response time basin (5-h lag time). The precipitation model, driven by radar and satellite data, is linked to a rainfall-runoff model that was previously set up and calibrated and is applied to the Gardon d'Anduze (545-km²) basin. Forecasts for lead times ranging from 1 to 4 h are used to assess the performance of the coupled model in flood forecasting from a hydrological perspective. The results from two rain events show that a 1-h lead-time rainfall forecast provides no additional improvement in flood forecasting compared with the simple persistence method. However, a real improvement is observed for the 2-h lead-time forecast. In the case of the 4-h lead time, the precipitation model essentially uses infrared satellite images and ground observations. The performance of the model for the 4-h forecast is not significantly better than the persistence method. However, for one of the two rain events, the model predicts the peak flow much better than the persistence method and the timing of the peak flow is also better. Because this work uses only a one-channel infrared satellite image, the 4-h forecast could be improved in the future when multiple-channel satellite estimates will be used.

Flood routing modeling of one-dimensional unsteady flow in open channels is sometimes based on the numerical solution of the diffusive wave equation. This paper presents a computational method for the solution of the diffusive wave problem with lateral inflow, based on the fractional-step technique. This method converts the diffusive wave problem into two single problems by utilizing separate equations for convection and diffusion. This separation is well adapted for computerization in distributed hydrological models. The applicability and the accuracy of this method are studied by mathematical analysis. Results show that this method provides an efficient and accurate resolution of the diffusive wave equation under some conditions on space and time steps and on spatial and temporal distribution of lateral inflow.

This paper presents a methodology for forecasting seasonal streamflow and is an extension of a previously developed categorical streamflow forecast model that used persistence (i.e., the previous season's streamflow) and El $\mathrm{Ni}ño$-Southern Oscillation (ENSO) indicators. This newly developed methodology takes persistence, an ENSO indicator, and several Pacific/Indian Ocean sea surface temperature (SST) series as the main predictor variables. Using linear discriminant analysis, the forecast is expressed as probability of exceedance of continuous streamflow amounts. An exceedance probability forecast is continuous and is useful for the design and operation of water resource systems, which require a high degree of system reliability. Application of the forecast model to five Australian catchments shows that persistence is the most important predictor of streamflow for the next season. The other predictors, SSTs and the Southern Oscillation Index, may be more useful for forecasts with longer lead times when the degree of persistence is less noticeable. Finally, it is noteworthy that this generic approach to making an exceedance probability forecast can be used on any predictors and predictands.

Wind effects on a rain gauge can cause a significant underestimation of rainfall depths and contribute to the inconsistency in rainfall data. To reconstruct rainfall data requires a consistent method to quantify the undercatch. Although some empirical adjustment factors have been recommended, they are not generalized enough to reflect the variations of wind speed with respect to height. This paper presents a model by which the undercatch at an elevated rain gauge can be estimated by the logarithmic wind velocity profile and raindrop terminal velocity. Deficiency percentages of rain catch predicted by this model agree with field observations and recommended adjustment factors. This approach was further expanded to predict rain undercatch percentage due to vegetal coverage or other obstructions.

A stochastic model that can simulate non-Fickian (scale-dependent) subsurface dispersion by imposing long-range correlation structure within a particle-based transport scheme is detailed here. The model uses increments from self-similar random fractal processes. These processes [fractional Brownian motions (FBMs)] are characterized by an index H, which is related to the correlation properties of the function. Temporally correlated FBMs that mimic the faster-than-Fickian growth observed in contaminant plumes in heterogeneous subsurface porous media are chosen. It is shown how FBMs can be incorporated within traditional Langevin-based formulations of the random-walk particle tracking models to describe mass displacements. A Fokker-Planck diffusion equation with a time-dependent tensor and its solution are presented to describe FBM anomalous dispersion. A relationship of the model's parameters to measurable porous media parameters is given.

Hydrologists and engineers need methods to disaggregate hourly rainfall data into subhourly increments for many hydrologic and hydraulic engineering applications. In the present engineering environment where time efficiency and cost effectiveness are paramount characteristics of engineering tools, disaggregation techniques must be practical and accurate. One particularly attractive technique for disaggregating long-term hourly rainfall records into subhourly increments involves the use of artificial neural networks (ANNs). A past investigation of ANN rainfall disaggregation models indicated that although ANNs can be applied effectively there are several considerations concerning the characteristics of the ANN model and the training methods employed. The research presented in this paper evaluated the influence on performance of several ANN model characteristics and training issues including data standardization, geographic location of training data, quantity of training data, number of training iterations, and the number of hidden neurons in the ANN. Results from this study suggest that data from rainfall-gauging stations within several hundred kilometers of the station to be disaggregated are adequate for training the ANN rainfall disaggregation model. Further, we found the number of training iterations, the limits of data standardization, the number of training data sets, and the number of hidden neurons in the ANN to exhibit varying degrees of influence over the ANN model performance.

It is common knowledge that the optimal values of the calibrated parameters of a rainfall-runoff model for one model response may not be the optimal values for another model response. Thus, it is highly desirable to derive a Pareto front or trade-off curve on which each point represents a set of optimal values satisfying the desirable accuracy levels of each of the model responses. This paper presents a new genetic algorithm (GA) based calibration scheme, accelerated convergence GA (ACGA), which generates a limited number of points on the Pareto front. A neural network (NN) is then trained to compliment ACGA in the derivation of other desired points on the Pareto front by mimicking the relationship between the ACGA-generated calibration parameters and the model responses. The calibration scheme, ACGA, is linked with HydroWorks and tested on a catchment in Singapore. Results show that ACGA is more efficient and effective in deriving the Pareto front compared to other established GA-based optimization techniques such as vector evaluated GA, multiobjective GA, and nondominated sorting GA. Verification of the trained NN forecaster indicates that the trained network reproduces ACGA generated points on the Pareto front accurately. Thus, ACGA-NN is a useful and reliable tool to generate additional points on the Pareto front.

A mixed population of flood flows was shown to cause quality-of-fit problems if a single-population flood-frequency distribution was used to describe the flood data. The three populations in this mix were “ordinary,” tropical cyclone, and ice-jam–release floods. Parametric descriptions of the single and separated flood populations were evaluated using probability-plot correlation-coefficient tests. These tests quantified how well the flood-probability distributions agreed with plotting-position descriptions of the data and quantified the differences due to the mixed-population analysis. High outliers caused the high skewness found in the single-population analyses. The tropical cyclone component was underestimated by single-population analyses at gauging stations in Massachusetts that had little data.

A single equation for estimating the peak flows of annual floods at ungauged sites in Massachusetts was developed by combining the conditional probabilities of floods caused by tropical cyclones and ice-jam releases with the conditional probability of “ordinary” floods. Regression equations for these three flood populations demonstrated that two basin characteristics, drainage-basin area and basin head, and frequency of flooding determined the corresponding peak-flow rates.

Recently developed integrated circuit technology and new reliable load-cell scales have made it possible to develop a battery-operated weighing system to measure and record precipitation electronically on a continuous basis. This system was developed to replace the spring scale and chart and/or potentiometer system that is generally used in the Universal Recording Rain gauge. The load-cell scale system records and stores the depth of water in the collecting receiver at least every 15 min, or as often as every 1 min if 0.25 mm of precipitation has fallen during the 1-min period. These data are later telemetered to a central data storage facility. A laboratory calibration of the load-cell scale showed a very good linear relationship between load-cell output voltage and the weight of water in the receiver. There was no apparent hysteresis shown in the output between loading and unloading the receiver. This load-cell scale is ideal for remote locations because the circuitry draws very little quiescent current and a small solar panel keeps the system's 12-V battery fully charged.

A simple method is proposed for reconstructing the discharge hydrograph at a river section where only water level is monitored and discharge is recorded at another upstream section. The method, which is based on the assumption that lateral inflows are negligible, has two parameters linked to remotely observed discharge and permits, without using a flood routing procedure and without need of a rating curve at a local site, relating the local river stage with the hydraulic condition at a remote upstream section. Calibration and validation of the method were carried out by comparing its results with those of a hydrodynamic model called MIKE11. The method was applied to simulate five flood events that have occurred along an instrumented reach of the Tiber River in central Italy. The shape characteristics of the reconstructed hydrographs at a gauged section where only local stage was available were similar to those of the hydrographs computed by MIKE11. A comparison between the Muskingum model and proposed method showed also that the latter is more reliable if a hypothetical rating curve must be given at the downstream section.

Depth-duration-frequency (DDF) or intensity-duration-frequency (IDF) curves are developed using a single series of maximum rainfall data for the desired duration from a single station. If these curves are developed using the series for all durations from the same station, in theory they should never intersect. However, during the development of sets of IDF curves for different durations and return periods to develop a rainfall frequency atlas for a region in Venezuela, South America, it was found that these curves often intersected. A technique to generate these curves that would prevent their intersection is proposed. This new method entails the collective examination of all series of extreme rainfall depths, and therefore, guarantees consistent results. This method is able to detect series with insufficient or doubtful measurements through the value of the exponent of an equation that relates rainfall depth and duration, and through the statistics of their series. This method also allows the generation of new DDF and/or IDF curves for a virtual rainfall gauge strategically located in areas under study.

Ten-stage operation policies for routing of flood hydrographs from very small magnitudes up to the probable maximum flood (PMF) for any dam having a gated spillway are suggested. The gate opening rules are determined based on the recent pool level. Without having to predict the magnitude of the incoming flood hydrograph beforehand, the fixed rules of the ten-stage operation policy will provide quasi-optimum routing for any floods, which are classified into 10 different groups. 0.1 PMF, 0.2 PMF, . . . , and PMF are the upper limits of the ten groups. When the PMF is routed, the rising and falling limbs of the outflow hydrograph make 10 sudden jumps and sudden drops at definite times and smoothly vary between steps. The developed flood routing model is applied sequentially to Catalan and Seyhan dams on the Seyhan River in Turkey.