Journal of Hydrologic Engineering

Effective storm water management plans depend on reliable rainfall intensity-duration-frequency (IDF) relationships. Due to the perception of high-intensity rainfall events as occurring more frequently than expected, the Michigan Dept. of Transportation (MDOT) commissioned a study to update rainfall IDF estimates for each of seven durations (1, 2, 3, 6, 12, 18, and 24 h) and six recurrence intervals (2, 5, 10, 25, 50, and 100 years). In contrast to a traditional at-site frequency analysis using method of moments estimators, this study applied a regional frequency analysis approach based on $L$ -moments. Data were compiled from 76 hourly recording stations and 152 daily recording stations, and trend and outlier analyses were conducted on both annual maximum series (AMS) and partial duration series (PDS) data. With the entire state considered a homogeneous region, two regional index flood models were applied: a generalized Pareto distribution fit to PDS data (PDS/GPA model), and a generalized extreme value distribution fit to AMS data (AMS/GEV model). Verification of results indicated that the revised rainfall IDF estimates provide more reliable values than those previously used.

Intensity-duration-frequency (IDF) estimates are important statistical summaries of precipitation records used for hydrologic engineering design. To make use of recent data and modern statistical methods, the IDF estimates for the state of Michigan have been updated using a regional frequency analysis procedure. Since short duration data (less than one hour) are only available in the Detroit metropolitan area, scale-invariance assumptions are used to estimate short-duration IDF values throughout the state. Both simple scaling and multiscaling models are considered, and the simple scaling model is selected for practical purposes. Verification results indicate that the short-duration estimates are consistent with historical observations.

Splitting the Nash single linear reservoir into two serially connected reservoirs of unequal storage coefficients (one hybrid unit) for a physically realistic response, a hybrid model is introduced for derivation of a synthetic unit hydrograph. Empirical relations are given for estimation of the two storage coefficients from known peak flow $\left(q_p\right)$ and time to peak $\left(t_p\right)$ . The hybrid model with two serially connected units is found to work significantly better than the most widely used methods such as those of Snyder, the Soil Conservation Service (SCS), and the two-parameter gamma distribution when tested on synthetically generated data and the data from four catchments from India and one from Turkey. The workability of the proposed approach was also tested for partial and no data availability situations.

A theoretical analysis derives that master recession curves on log paper indicate the physical occurrence of stream flows. The master recession curves in this study are generated by one-dimensional overland flow without and with lateral flow and groundwater flow. One-dimensional overland flow is also classified by assumption of the energy slope and kinematic or diffusion wave models. These theoretical results are coincident with the slopes of the observed master recession curves of the hydrographs. The slopes of the master recession curve of water depth and the discharge hydrographs on the log paper designate the predominancy of river discharge. These results predict the governing causes of river discharge occurrence during the master recession period. When the lateral flow is predominant, the master recession curve is not definitely determined. The observation data indicate that the slopes of the master recession curves are found to be independent of the initial conditions of channel flows, spatially varied rainfalls, moving rainstorms, etc.

A solution is obtained for unsteady flow to a continuous point sink in a homogeneous, anisotropic, leaky aquifer. Then superposition is used to distribute sinks with constant strength along straight-line segments to model flow to either partially penetrating vertical wells or nonvertical wells with finite length. The result is shown to duplicate the Hantush solution for flow to a vertical, partially penetrating well in a fully confined aquifer. However, differences occur when the corresponding result for a leaky aquifer is compared with an approximate solution obtained by Hantush, who assumed that inflow leakage is distributed uniformly over the full aquifer thickness. These differences are believed to result from inaccuracies in the Hantush approximation. Additional examples are used to illustrate the method, including flow to a single horizontal well, flow to two horizontal wells crossed at right angles to each other, and flow to a “blind” well formed by joining inclined and horizontal wells. A particle tracking exercise is carried out to illustrate use of this solution for pump-and-treat groundwater remediation.

In a flood-prone region, quick and accurate flood forecasting is imperative. It can extend the lead time for issuing disaster warnings and allow sufficient time for habitants in hazardous areas to take appropriate action, such as evacuation. In this paper, two hybrid models based on recent artificial intelligence technology, namely, the genetic algorithm-based artificial neural network (ANN-GA) and the adaptive-network-based fuzzy inference system (ANFIS), are employed for flood forecasting in a channel reach of the Yangtze River in China. An empirical linear regression model is used as the benchmark for comparison of their performances. Water levels at a downstream station, Han-Kou, are forecasted by using known water levels at the upstream station, Luo-Shan. When cautious treatment is made to avoid overfitting, both hybrid algorithms produce better accuracy in performance than the linear regression model. The ANFIS model is found to be optimal, but it entails a large number of parameters. The performance of the ANN-GA model is also good, yet it requires longer computation time and additional modeling parameters.

A new saturated-unsaturated three-dimensional (3D) groundwater flow model (SU3D) has been developed that calculates the pressure distribution over the entire groundwater flow domain in response to rainfall and evapotranspiration. Recent advances in solving the saturated-unsaturated groundwater flow equation are incorporated into SU3D, which solves the nonlinear, 3D, modified mixed form of the Richards equation continuously throughout the groundwater flow domain. The block-centered, finite-difference method with a variably sized grid is employed to solve the governing partial differential equation. The non-linear terms of the governing equation are linearized using a modified Picard iteration scheme, and the preconditioned conjugate gradient method is used to solve the linearized system of equations. SU3D can simulate the effects of pumping from an aquifer, and it has an option to calculate potential evapotranspiration (PET) from meteorological data. The PET is partitioned into potential evaporation and potential transpiration as a function of the leaf area index, and then the actual evaporation and transpiration are calculated. Overland flow and seepage face calculations are not included in SU3D.

Verification and application results are presented for a new saturated-unsaturated 3D groundwater flow model (SU3D) developing in Dogan and Motz, which can be used to calculate the pressure distribution over the entire groundwater flow domain in response to rainfall and evapotranspiration. SU3D solves the nonlinear 3D, modified mixed form of the Richards equation continuously throughout the groundwater flow domain, including both the unsaturated and saturated zones. The block-centered finite-difference method, a modified Picard iteration scheme, and the preconditioned conjugate gradient method are used to solve the governing partial differential equations in the new model. SU3D can simulate evaporation from land surface and transpiration from the root zone. Potential evapotranspiration is partitioned into potential evaporation and potential transpiration as a function of the leaf area index; actual evaporation and actual transpiration are then calculated individually. As shown in this paper, SU3D has been verified by reproducing the results of five analytical and numerical studies and a 3D unconfined-aquifer pumping test from the published literature.

Series expansions are obtained for a contour integral that appears in some solutions for flow in aquifers that have a bottom impermeable boundary and are capped on top with an aquitard containing a shallow standing water table. These solutions describe drawdown for delayed-yield flow to a well and the stream depletion and drawdown that occur when water is pumped from a well beside a stream. Hunt in 2003 previously gave a series expansion for the calculation of this integral for a somewhat more limited range of its parameters. However, the series expansions obtained herein give a more accurate and efficient way to compute values for this integral over the entire permissible range of its parameters.