Ferdi L. Hellweger

Transport in the Hudson Estuary: A Modeling Study of Estuarine Circulation and Tidal Trapping

The effects of estuarine circulation and tidal trapping on transport in the Hudson estuary were investigated by a large-scale, high-resolution numerical model simulation of a tracer release. The modeled and measured longitudinal profiles of surface tracer concentrations (plumes) differ from the ideal Gaussian shape in two ways: on a large scale the plume is asymmetric with the downstream end stretching out farther, and small-scale (1–2 km) peaks are present at the upstream and downstream ends of the plume. A number of diagnostic model simulations (e.g., remove freshwater flow) were performed to understand the processes responsible for these features. These simulations show that the large-scale asymmetry is related to salinity. The salt causes an estuarine circulation that decreases vertical mixing (vertical density gradient), increases longitudinal dispersion (increased vertical and lateral gradients in longitudinal velocities), and increases net downstream velocities in the surface layer. Since salinity intrusion is confined to the downstream end of the tracer plume, only that part of the plume is effected by those processes, which leads to the largescale asymmetry. The small-scale peaks are due to tidal trapping. Small embayments along the estuary trap water and tracer as the plume passes by in the main channel. When the plume in the main channel has passed, the tracer is released back to the main channel, causing a secondary peak in the longitudinal profile.

Heterogeneity of intracellular polymer storage states in enhanced biological phosphorus removal (EBPR)--observation and modeling.

A number of agent-based models (ABMs) for biological wastewater treatment processes have been developed, but their skill in predicting heterogeneity of intracellular storage states has not been tested against observations due to the lack of analytical methods for measuring single-cell intracellular properties. Further, several mechanisms can produce and maintain heterogeneity (e.g., different histories, uneven division) and their relative importance has not been explored. This article presents an ABM for the enhanced biological phosphorus removal (EBPR) treatment process that resolves heterogeneity in three intracellular polymer storage compounds (i.e., polyphosphate, polyhydroxybutyrate, and glycogen) in three functional microbial populations (i.e., polyphosphate-accumulating, glycogen-accumulating, and ordinary heterotrophic organisms). Model predicted distributions were compared to those based on single-cell estimates obtained using a Raman microscopy method for a laboratory-scale sequencing batch reactor (SBR) system. The model can reproduce many features of the observed heterogeneity. Two methods for introducing heterogeneity were evaluated. First, biological variability in individual cell behavior was simulated by randomizing model parameters (e.g., maximum acetate uptake rate) at division. This method produced the best fit to the data. An optimization algorithm was used to determine the best variability (i.e., coefficient of variance) for each parameter, which suggests large variability in acetate uptake. Second, biological variability in individual cell states was simulated by randomizing state variables (e.g., internal nutrient) at division, which was not able to maintain heterogeneity because the memory in the internal states is too short. These results demonstrate the ability of ABM to predict heterogeneity and provide insights into the factors that contribute to it. Comparison of the ABM with an equivalent population-level model illustrates the effect of accounting for the heterogeneity in models.

100 Years since Streeter and Phelps: It Is Time To Update the Biology in Our Water Quality Models

T development of the dissolved oxygen sag model by Streeter and Phelps 100 years ago (the study was completed in 1915, but publication was delayed to 1925 due to World War I) marks the beginning of water quality modeling. Since then, water quality managers have relied on models to relate point and nonpoint pollutant discharges to ambient conditions and criteria as part of waste load allocations (WLA) and total maximum daily load (TMDL) analyses. Unfortunately, unlike physical models, the predictive skill of our microbial water quality models is generally very low. Given sufficient input data I can set up a hydraulic model of a river that produces a reasonable prediction “out of the box”, but microbial (e.g., eutrophication) models typically require sitespecific calibration and even then have very low predictive ability. As a result we continue to be surprised and puzzled by how our ecosystems respond to change. Lake Erie is a poster child for nutrient reduction and has initially responded with significant improvements in eutrophic symptoms (e.g., hypoxia, algal biomass), but now is experiencing a resurgence of eutrophy, including frequent toxic cyanobacteria blooms. Although there are various hypotheses for why this is happening (e.g., changes in the reactivity of P loading, zebra mussels) there certainly is no consensus. We have to conclude that we do not know. I think Lake Erie illustrates one of the grand challenges faced by the field of environmental engineering: We do not have a predictive understanding of our ecosystems, sufficient for making effective management decisions. A worthwhile goal for the environmental science and engineering field is to develop modeling technology that can explain what is happening to Lake Erie. I propose that one area where significant progress toward this goal can be made is in the biology of our models. There is progress and water quality models do improve, but most advances have occurred in the physics (e.g., highresolution, 3-D hydrodynamics), land-side integration (e.g., watershed models), forecasting and uncertainty analysis. The biology in our models has and continues to change very little. For example, for nutrient limitation of phytoplankton growth, models typically still use a saturation equation originally proposed by Monod in 1942. Water quality modelers know this, and a simple analysis of the literature confirms that this equation is a persistent (and maybe even increasing?) feature of our models (Figure 1). Jacques Monod would be pleased! Of course, using an old equation is not a problem, if it still represents the state of the science. The Saint-Venant Equations are even older than Monod, yet still constitute the basis of shallow water hydraulics. However, the biological sciences have seen so much progress since 1942 (including by Monod), that it is increasingly difficult to ignore the disconnect with our models. It is hard to even begin to describe the advances in biology over the past century. We now know the detailed intracellular makeup of many microbes, have resolved the structure of many enzymes, and are even designing and making bacteria from scratch. For example, we have sequenced the genome and mapped out the complete intracellular metabolic network of the cyanobacteria Synechocystis. We almost have a mechanistic understanding of life! Along with these advances come insights into biological and ecological mechanisms that go way beyond what’s in our models. It is been known for some time that some phytoplankton can grow on organics, and now we also learned that some heterotrophic bacteria can use light to power membrane pumps. Bacteria can learn and exhibit anticipatory behavior. For example, E. coli turn on anaerobic respiration genes when the temperature goes up, because they “think” they just entered the oral cavity of a mammal and the next stop is the anaerobic gastrointestinal tract. Although our models typically assume that microbial populations are made up of identical individuals (like chemicals, as in the “Redfield molecule”), we now know that they are very heterogeneous in properties and behaviors, like nutrient content. Microbes not only interact passively via resources (nutrients and light), but are territorial and combat each other using various attack and defense mechanisms. For example, the cyanobacteria Microcystis produces compounds that inhibit the growth of

Accounting for Nitrogen Fixation in Simple Models of Lake Nitrogen Loading/Export

Coastal eutrophication, an important global environmental problem, is primarily caused by excess nitrogen and management efforts consequently focus on lowering watershed N export (e.g., by reducing fertilizer use). Simple quantitative models are needed to evaluate alternative scenarios at the watershed scale. Existing models generally assume that, for a specific lake/reservoir, a constant fraction of N loading is exported downstream. However, N fixation by cyanobacteria may increase when the N loading is reduced, which may change the (effective) fraction of N exported. Here we present a model that incorporates this process. The model (Fixation and Export of Nitrogen from Lakes, FENL) is based on a steady-state mass balance with loading, output, loss/retention, and N fixation, where the amount fixed is a function of the N/P ratio of the loading (i.e., when N/P is less than a threshold value, N is fixed). Three approaches are used to parametrize and evaluate the model, including microcosm lab experiments, lake field observations/budgets and lake ecosystem model applications. Our results suggest that N export will not be reduced proportionally with N loading, which needs to be considered when evaluating management scenarios.

A Three-Dimensional Model for Cohesive Sediment Dynamics in Shallow Bays

The fate and transport of pollutants in aquatic systems is strongly influenced by cohesive sediment dynamics. This paper describes the development and validation of a high-resolution, three-dimensional, time-variable, cohesive sediment transport model of Green Bay. The model incorporates state-of-the-art cohesive sediment resuspension and deposition methodology, including bed armoring and flocculation. Resuspension and deposition are functions of bed shear stress due to wave-current interaction. To compute shear stresses and water column transport, hydrodynamic and wind-wave models of the bay were integrated with the sediment transport model. Boundary conditions at the Green Bay-Lake Michigan interface were derived from a coarser resolution hydrodynamic model of Lake Michigan. The model accounts for all significant sources of solids into the system including major tributaries, direct runoff, shoreline erosion and internal solids (i.e., algae) production. The model was simulated for a period of 516 days. Results indicated that the model was capable of reproducing the temporal variation of observed inorganic suspended sediment concentrations at 25 locations and annual sedimentation rates at 49 other locations.