Introduction Site Description Experimental Methodology Math Model Results >Discussion Acknowledgments & References Figures & Equations PDF Version

6. Discussion

[21] For the low-Re_s flow observed in this study, the stem-wake structure is laminar and particle-spreading rates reflect contributions of Brownian diffusion, bed-induced shear, and mechanical dispersion. Brownian diffusion played a negligible role in dispersive mixing within the channel, as the Brownian diffusion coefficient for the TiO₂ particles (= 5 x 10^-9 m² h^-1) is several orders of magnitude lower than the best-fit estimates of D_Lon, D_Lat, and D_V. Like Brownian diffusion, boundary-induced shear flow was not an important contributor to particle dispersion because, within aquatic vegetation, gradients in velocity attributable to retardation of flow near the bed surface are restricted to a narrow region (i.e., 1-2 cm) adjacent to the bed [Nepf et al., 1997b]. This boundary layer lies well below the portion of the water column sampled by the TiO₂ tracer cloud. Mechanical dispersion, or mixing caused by local variations in the direction and velocity of flow around the mean velocity (V), represented the dominant mechanism of TiO₂-particle dispersal. The local variations in advective transport that promote mechanical dispersion are not caused by turbulence, but arise from small-scale heterogeneity in the density of vegetation and resulting nonuniformities in flow resistance and tortuosity of particle-transport pathways [see Nepf et al., 1997b].

[22] Mechanical dispersion was anisotropic in our experiment, with spreading in the longitudinal and lateral directions exceeding dispersion in the vertical direction by more than two orders of magnitude. The comparatively small vertical dispersion is consistent with observations of solute transport through geologic environments, which, like wetland systems, are composed of tortuous transport pathways, and suggests that vertical variation in vegetative structure is considerably less than that in the horizontal directions. These results also indicate that vertical mixing of particles is exceedingly slow in the absence of strong winds, thermal overturn, or other conditions that could promote turbulence. In our experiment, where the water depth (dw) equaled 0.6 m, the time scale for complete vertical mixing (=dw² /D_V) was 360 h.

[23] Model calculations made with the best-fit parameter values show that peak breakthrough concentrations were 60 times lower than those calculated assuming conservative advective-dispersive transport (i.e., = 0). The TiO₂ particles were too small to be removed by settling, and the particles did not diffuse to the channel bottom, where they would be susceptible to removal by adsorption to the sediments. There is also no evidence to support particle trapping within stagnation zones as a significant particle removal mechanism because breakthrough- curve tailing, a diagnostic feature of this reversible mass-transfer process, was not observed in our experiment. Interception of particles by aquatic vegetation represented the primary mechanism of particle removal within the surface-water channel.

[24] The effectiveness of the plant stems in scavenging TiO₂ from the water column can be quantified in terms of a single-stem collection efficiency (_s), which expresses the ratio that particles stick to a single stem to the rate that particles approach a single stem from upstream. Particles are removed from a unit volume of surface water at the rate C and thus are collected by (stick to) a single stem at the rate C(h/P-hd²/4), where P is the stem density (stems/m²) and the quantity in parentheses is the volume of water associated with a single (cylindrical) emergent stem of diameter d. The rate at which particles approach the stem is VChd, giving

  D

The _s value computed with the best-fit estimates of and V and with measurements of P (=1150 stems/m²) equals 0.29, indicating that a single stem was capable of scavenging 29% of the particles that approached its projected cross-sectional area from the upstream direction. We suspect that the majority of intercepted particles were bound to periphyton that formed thin coatings on the plant stems. These periphyton sweaters likely provide a substrate favorable for the attachment of most natural inorganic particles of colloid size (i.e., 1 nm to 10 µm) that, like the TiO₂ used in this study, have a net negative surface charge in waters of circumneutral pH.

[25] Our results suggest that migration of colloid-sized mineral particles may be limited to a few tens of meters in wetlands characterized by laminar flow regimes. Under turbulent flow, particle trapping by wetland vegetation may be comparatively less effective (due to low particle-substrate contact times) and may be reversible (due to elevated shear forces on attached particles). Although the TiO₂ particles should serve as a good analog for other mineral colloids, we caution against using these results to draw inferences about the transport of larger inorganics (e.g., silt), which are susceptible to removal by settling, or colloidal organic matter, which has a relatively lower density and different surface properties. In light of the complex effects that interactions between particle type and flow regime have on particulate-matter transport, we recommend that future research focus on additional model evaluations against field data collected outside the range of conditions studied here.