Journal of Hydrologic Engineering

A nonparametric Bayesian simulation procedure, Pólya resampling, is presented for comparing statistical distributions D and estimation methods E for at-site flood frequency analysis. The methodology does not rely on the hypothesis of a parent distribution and compares the models on the basis of expected cost. It is applied to the Québec-Ontario region. The statistical distribution minimizing expected cost depends mainly on the ratio of underestimation to overestimation costs. When these costs are similar, the gamma distribution performs well. When overestimation costs are more important, the normal distribution minimizes expected cost. And when underestimation costs are larger, the Gumbel distribution is preferred. However, this only tells us that the systematic use of these distributions is preferable to the systematic use of other distributions considered in the study. A dynamic procedure, which selects a D/E combination on the basis of some sample statistics, may give better results. In particular, a simple dynamic procedure based on the observed coefficients of variation and skewness is shown to outperform the gamma and Gumbel distributions for extrapolation when underestimation and overestimation costs are similar.

The flood frequency analysis has been widely conducted where gauging stations are available and statistical distributions can be used to analyze historical data. On the other hand, flash floods usually occur at unexpected locations having no gauging stations, and have hardly been discussed. Based on the available information from surrounding precipitation gauges, potential flash floods were defined for a watershed and their regional distributions were investigated. Daily truncation levels of 99.9 and 99.99% were used to derive potential flash floods, where a 99.9% level means that 0.1% of daily precipitations are larger than or equal to the truncated value. The higher the truncation level, the greater the magnitude of a potential flash flood. Records from 21 precipitation gauging stations were selected for the studied region. The kriging method based on the minimum variance unbiased estimation was introduced to calculate spatial interpolation. These kriging estimates form a regional distribution of potential flash floods, which are expressed by contour lines. The result shows that the estimation errors associated with the regionalization were reasonably small.

The U.S. Geological Survey, finite-difference flow model (MODFLOW) is the most commonly applied numerical code in ground-water hydrology. This flow model is frequently used in the design of pump-and-treat systems for aquifer remediation. In this paper we modified the MODFLOW well package to improve on the approximation of multilayer well effects when simulating vertical wells that penetrate several layers and adjacent cell effects when simulating horizontal wells. This was done by simulating flow through a well as flow through a closed conduit pipe, to incorporate flow between adjacent model cells that contain the well. Thus, water exchange between adjacent layers or cells containing the same well as dependent on the flow within the well and flow outside the well of the adjacent cells containing the well. Examples are presented involving ground-water extraction to demonstrate the applicability of the approach to simulate horizontal and vertical wells. It was proposed that the shape of the capture zone of pumping was closer to that of a pear than that of an ellipsoid.

In the study of solute transport through laboratory columns packed with natural porous media, tracers are commonly applied as a step input or as a square-wave pulse of several pore volumes duration. However, tracer applications such as a short-duration pulse (effectively simulating a Dirac input) should provide a more sensitive means of determining flow characteristics. We describe an experimental confirmation of this expectation using studies of tritiated water transport through soil columns. The porous media studied include both homogeneously packed sands and media with clearly defined paths of preferential flow. For all of the systems studied, the duration of the solute input pulse dramatically influenced our ability to discern the adequacy of a simple convection-dispersion model. In particular, the use of Dirac inputs was most effective at revealing situations in which physical/chemical processes other than convection and dispersion were present.

The increasing practice of infiltration in disposal of storm-water runoff leads to concerns over the long-term quality of receiving ground waters. To ensure that best management practices are employed, understanding the fate and transport of applied trace pollutants within the vadose zone must be improved. This study is directed at expanding the knowledge base governing facilitated transport of heavy metals by dissolved or particulate natural organic matter (NOM). Experimental studies were conducted to evaluate the effects of NOM on the mobility of trace quantities of cadmium, copper, lead, and zinc dissolved in water infiltrating through near-surface soils. Large-scale soil columns (on order of 1 m) were tested under periodic loading conditions thought to be similar to those experienced under actual infiltration scenarios. The corresponding metals transport was observed as function of depth and volume eluted. Increased concentrations of NOM yielded greater metal attenuation within the soils tested. Explanation of these results focus on the possible presence of organo-metallic complexes that exhibit a high affinity for the soil matrix.

In practice, runoff peak flow rates are typically estimated by the rational method, a design storm unit hydrograph (UH) method, or a statistical regression equation. In this technical note, the balanced design storm UH procedure is used to derive a rational method peak flow rate equation that, in turn, is used to derive a regression equation. This new mathematical linkage across these three widely used peak flow rate estimation techniques provide a foundation as to how these approaches differ or agree, and may also provide an answer as to which method is “best”; specifically, the methods are essentially the same for many practical conditions, and where they differ, the underpinnings of their mathematical structures are illuminated.

The Spearman correlation test that was developed for use with two random variables has been used incorrectly for analyzing univariate data for serial correlation. This occurs where measurements of the second random variable are incomplete. Data for assessing watershed change, such as for urbanization and deforestation, are commonly incomplete. Although the Spearman coefficient can be applied to detect serial correlation, the probability density function of the bivariate test statistic is not correct for univariate analyses of serial correlation. This research has produced critical values of the test statistic appropriate for use when the Spearman test is applied to serial correlation testing. These will enable more accurate evaluations of the effects of watershed change, as shown by the applications provided in this technical note.

For an overland plane with rainfall as the lateral inflow, the flow regime is complicated by the varying flow depth and velocity along the plane. The flow regime becomes variable. For a plane that is sufficiently long, from the upstream to the downstream end of the plane, the flow regime may change from laminar through transitional to turbulent. In this paper, by coupling the kinematic wave equations with the Darcy-Weisbach friction formula, a time of concentration formula for a portion of a plane subject to uniform rainfall excess and a constant upstream inflow is derived. This formula is applicable to a portion of plane with a single flow regime only. For practical application, this formula is further developed for a plane of uniform slope subject to uniform rainfall excess, with flow regime varying from laminar through transitional to turbulent. An application of the formula for varying flow regime shows that its time of concentration estimate is longer than those based on a single flow regime covering the entire plane.

Once the soil surface reaches saturation, rain excess is generated. This rain excess may run over and infiltrate in other areas of the watershed. In this study, we consider three scenarios of distributing rain excess/run-on to examine how it affects infiltration over a watershed. For these scenarios, we introduce simple solutions that allow the users to (1) estimate the ponded area-fraction of the watershed, and (2) determine the area-averaged infiltration. The solutions are practical because they require trivial computations. The solutions of ponding time and area-averaged infiltration account for the influence of initial water content and contain parameters with precise physical meaning. We find that area-averaged infiltration depends strongly on the distribution of rain excess at the surface. Thus we question the use of many stochastic area-averaged infiltration models that ignore the influence of rain excess.