James N. Carleton

Factors affecting the performance of stormwater treatment wetlands.

Data from 35 studies on 49 wetland systems used to treat stormwater runoff or runoff-impacted surface waters were examined and compared in order to identify any obvious trends that may aid future stormwater treatment wetland design efforts. Despite the intermittent nature of hydrologic and pollutant inputs from stormwater runoff, our analysis demonstrates that steady-state first-order plug-flow models commonly used to analyze wastewater treatment wetlands can be adapted for use with stormwater wetlands. Long-term pollutant removals are analyzed as functions of long-term mean hydraulic loading rate and nominal detention time. First-order removal rate constants for total phosphorus, ammonia, and nitrate generated in this fashion are demonstrated to be similar to values reported in the literature for wastewater treatment wetlands. Constituent removals are also demonstrated via regression analyses to be functions of the ratio of wetland area to watershed area. Resulting equations between these variables can be used as preliminary design tools in the absence of more site-specific details, with the understanding that they should be employed cautiously.

Damköhler number distributions and constituent removal in treatment wetlands

Abstract Using hypothetical wetland simulations and data from the literature, Kadlec [Eco. Eng. 15 (2000) 105] recently demonstrated that plug-flow models commonly used to quantify treatment wetland performance fail to describe conditions other than those under which calibration data are collected. Parameters of these models (removal rate constants ( k ) and background concentrations ( C *)) demonstrate apparent dependence on inlet concentration and hydraulic loading rate which is not alleviated by including dispersion to address non-ideal flow. The phenomenon can be understood as resulting from an interdependence between k and local flow velocity, due to the functional dependence of each on drag-inducing surfaces (and attached biofilms) associated with submerged vegetation and litter. This paper presents a simple method, based on theoretical considerations, for determining C * using inlet–outlet data, independent of the degree of mixing or the nature of the removal processes. This paper also expands upon the hypothetical multi-channel example to suggest a modeling approach in which a wetland is treated conceptually as an ensemble of parallel, non-interacting stream tubes in plug-flow, characterized by a continuous distribution of Damkohler numbers. The Damkohler number distribution (DND) can be estimated from the residence time distribution (RTD) under the assumption of uniform flow path length. For such a wetland, under steady state conditions and with constant inlet concentration, the fraction of a removable pollutant remaining as a function of normalized distance from inlet to outlet is given by the Laplace transform of the DND. Similarly, the DND can be derived from the inverse Laplace transform of the normalized concentration versus normalized distance curve. Given both an RTD and a DND, it is possible to investigate the relationship between k and residence time, and the mechanistic nature of the removal process. Employing these concepts makes it possible to generate an expression for normalized concentration as a function of fractional distance that is unaffected by changes in inlet concentration, and inherently takes into account changes in hydraulic loading rate.

Implementation of a probabilistic curve number method in the PRZM runoff model

Abstract Runoff from agricultural fields, which is a major factor influencing ecological and human exposure assessments, is often estimated by models that use the curve number method. One such model, PRZM, adjusts curve number on a daily basis according to calculated soil moisture in a manner that does not usually capture the full range of variability in rainfall–runoff relationships. Under the assumption that soil moisture alone cannot account for the large observed variability in event-to-event curve number, we modified PRZM by decoupling the curve number from any soil moisture relationship. Instead, we varied the daily curve number by randomly selecting the days initial abstraction from a distribution. Using field data, we showed that the modified PRZM better characterized the variability in the rainfall–runoff relationship than the method based on soil moisture.

An analysis of performance models for free water surface wetlands

Although treatment wetlands are intended to attenuate pollutants, reliably predicting their performance remains a challenge because removal processes are often complex, spatially heterogeneous, and incompletely understood. Although initially popular for characterizing wetland performance, plug flow reactor models are problematic because their parameters exhibit correlation with hydraulic loading. One-dimensional advective-dispersive-reactive models may also be inadequate when longitudinal dispersion is non-Fickian as a result of pronounced transverse gradients in velocity (preferential flow). Models that make use of residence time distributions have shown promise in improving wetland performance characterization, however their applicability may be limited by certain inherent assumptions, e.g. that transverse mixing is nil. A recently-developed bicontinuum (mobile-mobile) model that addresses some of these weaknesses may hold promise for improving wetland performance modeling, however this model has yet to be tested against real-world wetland data. This paper examines the state of the science of free water surface wetland hydrodynamics and transport modeling, discusses the strengths and weaknesses of various steady state models, and compares them to each other in terms of each models ability to represent data sets from monitored wetlands.

Analysis and Solution of a Bicontinuum Transport Model

A solute transport model applicable to systems where the heterogeneity is too small to be considered explicitly and too large to be treated asymptotically is developed from both stochastic averaging and random walk perspectives. Both perspectives are found to produce equivalent models that can be reversibly transformed to multicontinuum form. The eventual outcome of the development is a 4-equation multicontinuum model applicable to transport in three-dimensional heterogeneous flowfields. The model has the expected form in the Markovian (Fickian) case and a nonlocal interzonal mixing term in the non-Markovian (fractional, super-dispersive) case. All model parameters are physically-based and independently measurable. The developed formulation is expected to help biological engineers design better treatment and remediation strategies that foster the health of tissues, organisms, soils, wetlands and aquifers. Extensions to anisotropic fields, reactive transport and unsaturated flows will be pursued in future studies.

Combining GIS, AI and Modeling to Analyze Wetland Functions in Maryland Watersheds

A wetlands-focused Decision Support System (DSS) integrating Expert Systems (ES) and a hydrologic model is developed within a raster-based GIS and applied to two watersheds of the Maryland eastern shore. The DSS is used to develop a wetlands enhancement plan for atrazine export control by first identifying export hot spots and then delineating optimized locations for potential wetland construction and restoration. Hydrologic modeling is used to quantify the expected reduction in agrichemical export load with the proposed expanded wetland acreage in place. Simulation results indicate a 46% reduction in atrazine export.