Deborah J. Mossman

One-dimensional unsteady flow and unsteady pesticide transport in a reservoir

A one-dimensional unsteady contaminant transport model employing an unsteady hydraulic transport field was used to model four pesticides (alachlor, atrazine, metolachlor, metribuzin) in Coralville Reservoir, Iowa, USA. Eulerian chemistry data were collected during April–July 1986 along three longitudinal points in the impoundment. The water quality model uses a split-operator format for solving the advective-dispersive-reactive equation, and is coupled with a fixed bed through particulate settling and diffusive exchange. The velocity, reach volume and average depth were generated by an unsteady open channel code (CARIMA) and input to the water quality program. The time-step used in the water quality program was one day, and the space-step was 250 m. The Courant number ranged up to 64 in the water quality model while avoiding significant numerical dispersion. Longitudinal dispersion was held at a constant 0.5 m2/s. The upstream concentration boundary condition was reconstructed by reaching agreement between downstream data and model output. The frequency of data collection suggested that some high concentration events could go unmeasured. Good model calibrations were reached for the modeled chemicals. The system was dominated by advective processes. Calibration of the water quality model was more sensitive to transport parameters than to reaction kinetics.

Field observations of longitudinal dispersion in a run-of-the-river impoundment

Abstract Dye studies were conducted on a run-of-the-river impoundment to observe longitudinal dispersion in a lacustrine/riverine environment. Field data were analyzed in the context of a streamtube model which accounts for dispersion by the lateral distribution of longitudinal velocity. The differential advection associated with the vertical distribution of longitudinal velocity [i.e. Elders equation ( D x = α x u ∗ h )] is incorporated in this stream-tube model providing longitudinal dispersion within each tube. Model calibration attempts required α values on the order of 100, greatly in excess values expected from fluvial experience. A diffusive stall was observed twice when the dye plume moved into a sudden expansion of channel width. The stall is predicted by Fischers K x equation. The stream-tube model reasonably reproduced the field data in a variety of initial-period conditions, caused by the highly variable geometry of Coralville Reservoir.

Added flexibility for Holly-Preissmann advection operator

Abstract The Holly-Preissmann advection operator has been adapted for use on an uneven computational grid while maintaining very low numerical dispersion. This adaptation allows longer computational time-steps, while retaining high numerical accuracy. The longer time-steps are preferable for modeling slow reactions, such as biological transformations and uptake of pollutants. The Holly-Preissmann method has been incorporated into numerous finite-difference pollutant transport models, and has previously been restricted to local Courant number maxima of 1. The adapted operator shows good numerical performance at Courant numbers in excess of 20, and included testing the transport of a signal through an uneven velocity field. The adapted scheme can easily be integrated into existing model codes.

Plane source injector design for river dispersion studies

A device that injects a planar, spike input of tracer and accommodates irregular stream cross-sections was designed and field tested. The injector is based on a modular support and electrical system that can easily be modified in the field. The tracer is distributed to the water from self-emptying packets made of latex tubing. These packets resist premature breakage and contribute no litter to the stream. The packets empty after applying a current to the electrical system. Instream assembly of the tracer injector is possible from a boat. A two-dimensional input of tracer significantly shortens the initial mixing period for streams that are wide and deep. During the initial period, the tracer cloud becomes well-mixed vertically and laterally. After the initial period is satisfied, the variance of the tracer cloud grows linearly with time, and longitudinal dispersion can be measured easily. The use of a plane source injector significantly reduces the duration of the initial period.