Simplified contaminant source depletion models as analogs of multiphase simulators

Four simplified dense non-aqueous phase liquid (DNAPL) source depletion models recently introduced in the literature are evaluated for the prediction of long-term effects of source depletion under natural gradient flow. These models are simple in form (a power function equation is an example) but are shown here to serve as mathematical analogs to complex multiphase flow and transport simulators. The spill and subsequent dissolution of DNAPLs was simulated in domains having different hydrologic characteristics (variance of the log conductivity field=0.2, 1 and 3) using the multiphase flow and transport simulator UTCHEM. The dissolution profiles were fitted using four analytical models: the equilibrium streamtube model (ESM), the advection dispersion model (ADM), the power law model (PLM) and the Damkohler number model (DaM). All four models, though very different in their conceptualization, include two basic parameters that describe the mean DNAPL mass and the joint variability in the velocity and DNAPL distributions. The variability parameter was observed to be strongly correlated with the variance of the log conductivity field in the ESM and ADM but weakly correlated in the PLM and DaM. The DaM also includes a third parameter that describes the effect of rate-limited dissolution, but here this parameter was held constant as the numerical simulations were found to be insensitive to local-scale mass transfer. All four models were able to emulate the characteristics of the dissolution profiles generated from the complex numerical simulator, but the one-parameter PLM fits were the poorest, especially for the low heterogeneity case.     

Inverse modeling of BTEX dissolution and biodegradation at the Bemidji, MN crude-oil spill site

The U.S. Geological Survey (USGS) solute transport and biodegradation code BIOMOC was used in conjunction with the USGS universal inverse modeling code UCODE to quantify field-scale hydrocarbon dissolution and biodegradation at the USGS Toxic Substances Hydrology Program crude-oil spill research site located near Bemidji, MN. This inverse modeling effort used the extensive historical data compiled at the Bemidji site from 1986 to 1997 and incorporated a multicomponent transport and biodegradation model. Inverse modeling was successful when coupled transport and degradation processes were incorporated into the model and a single dissolution rate coefficient was used for all BTEX components. Assuming a stationary oil body, we simulated benzene, toluene, ethylbenzene, m,p-xylene, and o-xylene (BTEX) concentrations in the oil and ground water, respectively, as well as dissolved oxygen. Dissolution from the oil phase and aerobic and anaerobic degradation processes were represented. The parameters estimated were the recharge rate, hydraulic conductivity, dissolution rate coefficient, individual first-order BTEX anaerobic degradation rates, and transverse dispersivity. Results were similar for simulations obtained using several alternative conceptual models of the hydrologic system and biodegradation processes. The dissolved BTEX concentration data were not sufficient to discriminate between these conceptual models. The calibrated simulations reproduced the general large-scale evolution of the plume, but did not reproduce the observed small-scale spatial and temporal variability in concentrations. The estimated anaerobic biodegradation rates for toluene and o-xylene were greater than the dissolution rate coefficient. However, the estimated anaerobic biodegradation rates for benzene, ethylbenzene, and m,p-xylene were less than the dissolution rate coefficient. The calibrated model was used to determine the BTEX mass balance in the oil body and groundwater plume. Dissolution from the oil body was greatest for compounds with large effective solubilities (benzene) and with large degradation rates (toluene and o-xylene). Anaerobic degradation removed 77% of the BTEX that dissolved into the water phase and aerobic degradation removed 17%. Although goodness-of-fit measures for the alternative conceptual models were not significantly different, predictions made with the models were quite variable.     

A multi-objective optimization framework for surfactant-enhanced remediation of DNAPL contaminations

The occurrence of Dense Non-Aqueous Phase Liquid (DNAPL) contaminations in the subsurface is a threat for drinkwater resources in the western world. Surfactant-Enhanced Aquifer Remediation (SEAR) is widely considered as one of the most promising techniques to remediate DNAPL contaminations in-situ, be it with considerable additional costs compared to classical pump-and-treat remediations. A cost-effective design of the remediation set-up is therefore essential. In this work, a pilot SEAR test is executed at a DNAPL contaminated site in Belgium in order to collect data for the calibration of a multi-phase multi-component model. The calibrated model is used to assess a series of scenario-analyses for the full-scale remediation of the site. The remediation variables that were varied were the injection and extraction rate, the injection and extraction duration, and the surfactant injection concentrations. A constrained multi-objective optimization of the model was applied to obtain a Pareto set of optimal remediation strategies with different weights for the two objectives of the remediation: (i) the maximal removal of DNAPL and (ii) a total minimal cost. These Pareto curves can help decision makers to select an optimal remediation strategy in terms of cost and remediation efficiency. The Pareto front shows a considerable trade-off between the total remediation cost and the removed DNAPL mass.     

Simple screening models of NAPL dissolution in the subsurface

Three simple screening models of nonaqueous phase liquid (NAPL) dissolution in the subsurface are proposed based on the NAPL mass conservation and the assumption of proportionality between the residual NAPL source zone concentration and the remaining residual NAPL mass. The purpose of the proposed models is to predict the solute concentration in the zone of the residual NAPL as a result of dissolution. The predicted source zone concentration decrease is used to simulate and account for the decrease of dissolution rate with time. The proposed simple NAPL dissolution models enable the pseudo-equilibrium formulation to be used and therefore the numerical simulations for field application problems can be simplified compared to the non-equilibrium counterpart. With proper choice of empirical parameters, the proposed simple screening models can work as well as more complex dissolution rate correlation models, such as that of Imhoff et al. [Water Resour. Res. 30 (1994) 307-320]. It is found that the proposed models are very good for quantifying non-equilibrium dissolution, which is characterized by tailing of breakthrough curves. The models are especially useful for situations of small residual NAPL saturation, which are typical for many field applications.     

Influence of mass transfer characteristics for DNAPL source depletion and contaminant flux in a highly characterized glaciofluvial aquifer

The transfer of contaminant mass between the nonaqueous- and aqueous-phases is a process of central importance for the remediation of sites contaminated by dense nonaqueous-phase liquids (DNAPLs). This paper describes a comparison of the results obtained with various alternative DNAPL-aqueous-phase mass transfer models contained in the literature for predicting DNAPL source-zone depletion times in groundwater systems. These dissolution models were largely developed through laboratory column experiments. To gain insight into the implications of various representations of the local-scale kinetic as well as equilibrium DNAPL dissolution processes, aquifer heterogeneity and the complex architecture of a DNAPL source-zone, the aqueous-phase contaminant concentrations and mass fluxes arriving at a down-gradient compliance boundary are analyzed in a conditional stochastic framework. The hydrogeologic setting is a heterogeneous fluvial aquifer in Southwest Germany, referred to as the aquifer analog dataset, that was intensively characterized in three dimensions for hydrogeological parameters that include permeability, effective porosity, grain size, mineralogy and sorption coefficients. By embedding the various dissolution models into the compositional, multiphase flow model, CompFlow, the relative times predicted for complete depletion of a released DNAPL source due to natural dissolution are explored. Issues related to achieving environmental benefits through, for example, partial DNAPL-zone source removal via enhanced remedial technologies are also discussed. In this context, performance metrics in the form of peak aqueous-phase contaminant concentrations and mass fluxes arriving at a down-gradient compliance boundary are compared to each other. This is done for each of the alternative mass transfer models. A significant reduction in the fractional flux at a downstream location from the DNAPL source can be achieved by partial source-zone mass reduction; however, peak concentration levels at the same location remain much higher than the United States Environment Protection Agency (US-EPA) drinking water limits. Although groundwater quality was found to improve more rapidly for the equilibrium dissolution model, it is also shown that dissolution models that promote rapid DNAPL disappearance produce greater prediction uncertainty in the aqueous-phase flux reduction.     

Modelling of silica diffusion experiments with 32Si in Boom Clay

A mathematical model describing the dissolution of nuclear glass directly disposed in clay combines a first-order dissolution rate law with the diffusion of dissolved silica in clay. According to this model, the main parameters describing the long-term dissolution of the glass are etaR, the product of the diffusion accessible porosity eta and the retardation factor R, and the apparent diffusion coefficient D(app) of dissolved silica in clay. For determining the migration parameters needed for long-term predictions, four Through-Diffusion (T-D) experiments and one percolation test have been performed on undisturbed clay cores. In the Through-Diffusion experiments, the concentration decrease after injection of 32Si (radioactive labelled silica) was measured in the inlet compartment. At the end of the T-D experiments, the clay cores were cut in thin slices and the activity of labelled silica in each slice was determined. The measured activity profiles for these four clay cores are well reproducible. Since no labelled silica could be detected in the outlet compartments, the Through-Diffusion experiments are fitted by two In-Diffusion models: one model assuming linear and reversible sorption equilibrium and a second model taking into account sorption kinetics. Although the kinetic model provides better fits, due to the sufficiently long duration of the experiments, both models give approximately similar values for the fit parameters. The single percolation test leads to an apparent diffusion coefficient value about two to three times lower than those of the Through-Diffusion tests. Therefore, dissolved silica appears to be strongly retarded in Boom Clay. A retardation factor R between 100 and 300 was determined. The corresponding in situ distribution coefficient K(d) is in the range 25-75 cm(3) g(-1). The apparent diffusion coefficient of dissolved silica in Boom Clay is estimated between 2 x 10(-13) and 7 x 10(-13) m(2) s(-1). The pore diffusion coefficient is in the range from 6 x 10(-11) to 1 x 10(-10) m(2) s(-1).     

Evaluation of simplified mass transfer models to simulate the impacts of source zone architecture on nonaqueous phase liquid dissolution in heterogeneous porous media

Nonaqueous phase liquid (NAPL) dissolution was studied in three-dimensional (3D) heterogeneous experimental aquifers (25.5 cm x 9 cm x 8.5 cm) with two different longitudinal correlation lengths (2.1 cm and 1.1 cm) and initial spill volumes (22.5 ml and 10.5 ml). Spatial and temporal distributions of NAPL during dissolution were measured using magnetic resonance imaging (MRI). At high NAPL spill volume, average effluent concentrations initially increased during dissolution, as NAPL pools transitioned to NAPL ganglia, and then decreased as the total NAPL-water interfacial area decreased over time. Experimental results were used to test six dissolution models: (i and ii) a one-dimensional (1D) model using either specific NAPL-water interfacial area values estimated from MR images at each time step (i.e., 1D quasi-steady state model), or an empirical mass transfer (Sh') correlation (i.e., 1D transient model), (iii and iv) a multiple analytical source superposition technique (MASST) using either the NAPL distribution determined from MR images at each time step (i.e., MASST steady state model), or the NAPL distribution determined from mass balance calculations (i.e., MASST transient model), (v) an equilibrium streamtube model, and (vi) a 3D grid-scale pool dissolution model (PDM) with a dispersive mass flux term. The 1D quasi-steady state model and 3D PDM captured effluent concentration values most closely, including some concentration fluctuations due to changes in the extent of flow reduction. The 1D transient, MASST steady state and transient, and streamtube models all showed a monotonic decrease in effluent concentration values over time, and the streamtube model was the most computationally efficient. Changes during dissolution of the effective NAPL-water interfacial area estimated from imaging data are similar to changes in effluent concentration values. The 1D steady state model incorporates estimates of the effective NAPL-water interfacial area directly at each time point; the 3D PDM does so indirectly through mass balance and a relative permeability function, which causes reduced water flow through high saturation NAPL regions. Hence, when model accuracy is required, the results indicate that a surrogate of this effective interfacial area is required. Approaches to include this surrogate in the MASST and streamtube models are recommended.     

Entrapment and dissolution of DNAPLs in heterogeneous porous media

Two-dimensional multiphase flow and transport simulators were refined and used to numerically investigate the entrapment and dissolution behavior of tetrachloroethylene (PCE) in heterogeneous porous media containing spatial variations in wettability. Measured hydraulic properties, residual saturations, and dissolution parameters were employed in these simulations. Entrapment was quantified using experimentally verified hydraulic property and residual saturation models that account for hysteresis and wettability variations. The nonequilibrium dissolution of PCE was modeled using independent estimates of the film mass transfer coefficient and interfacial area for entrapped and continuous (PCE pools or films) saturations. Flow simulations demonstrate that the spatial distribution of PCE is highly dependent on subsurface wettability characteristics that create differences in PCE retention mechanisms and the presence of subsurface capillary barriers. For a given soil texture, the maximum and minimum PCE infiltration depth was obtained when the sand had intermediate (an organic-wet mass fraction of 25%) and strong (water- or organic-wet) wettability conditions, respectively. In heterogeneous systems, subsurface wettability variations were also found to enhance or diminish the performance of soil texture-induced capillary barriers. The dissolution behavior of PCE was found to depend on the soil wettability and the spatial PCE distribution. Shorter dissolution times tended to occur when PCE was distributed over large regions due to an increased access of flowing water to the PCE. In heterogeneous systems, capillary barriers that produced high PCE saturations tended to exhibit longer dissolution times.     

Design of aquifer remediation systems: (1) describing hydraulic structure and NAPL architecture using tracers

Aquifer heterogeneity (structure) and NAPL distribution (architecture) are described based on tracer data. An inverse modelling approach that estimates the hydraulic structure and NAPL architecture based on a Lagrangian stochastic model where the hydraulic structure is described by one or more populations of lognormally distributed travel times and the NAPL architecture is selected from eight possible assumed distributions. Optimization of the model parameters for each tested realization is based on the minimization of the sum of the square residuals between the log of measured tracer data and model predictions for the same temporal observation. For a given NAPL architecture the error is reduced with each added population. Model selection was based on a fitness which penalized models for increasing complexity. The technique is demonstrated under a range of hydrologic and contaminant settings using data from three small field-scale tracer tests: the first implementation at an LNAPL site using a line-drive flow pattern, the second at a DNAPL site with an inverted five-spot flow pattern, and the third at the same DNAPL site using a vertical circulation flow pattern. The Lagrangian model was capable of accurately duplicating experimentally derived tracer breakthrough curves, with a correlation coefficient of 0.97 or better. Furthermore, the model estimate of the NAPL volume is similar to the estimates based on moment analysis of field data.     

Design of aquifer remediation systems: (1) Describing hydraulic structure and NAPL architecture using tracers

Aquifer heterogeneity (structure) and NAPL distribution (architecture) are described based on tracer data. An inverse modelling approach that estimates the hydraulic structure and NAPL architecture based on a Lagrangian stochastic model where the hydraulic structure is described by one or more populations of lognormally distributed travel times and the NAPL architecture is selected from eight possible assumed distributions. Optimization of the model parameters for each tested realization is based on the minimization of the sum of the square residuals between the log of measured tracer data and model predictions for the same temporal observation. For a given NAPL architecture the error is reduced with each added population. Model selection was based on a fitness which penalized models for increasing complexity. The technique is demonstrated under a range of hydrologic and contaminant settings using data from three small field-scale tracer tests: the first implementation at an LNAPL site using a line-drive flow pattern, the second at a DNAPL site with an inverted five-spot flow pattern, and the third at the same DNAPL site using a vertical circulation flow pattern. The Lagrangian model was capable of accurately duplicating experimentally derived tracer breakthrough curves, with a correlation coefficient of 0.97 or better. Furthermore, the model estimate of the NAPL volume is similar to the estimates based on moment analysis of field data.     

A new approach for modelling potential effects in cation adsorption onto binary (hydr) oxides

The current approach for modelling ion adsorption onto binary (hydr)oxides using homogeneous surface complexation models involves the assumption of either an ideal mixture of the two surfaces (i.e. two surface sites on one surface) or a patchwise surface (i.e. two surfaces with one surface site on each surface). As the physical truth should be between these two limiting cases, a model which assumes a patchwise surface constituted of three patches is proposed. Two patches represent the distinct (hydr)oxides, and the third one a mixture of these distinct (hydr)oxides. Using the diffuse layer model, the three approaches are applied to literature data for Cd adsorption onto binary mixtures of alumina-coated silica at total constant Cd concentration and varying amounts of alumina coatings. For Cd adsorption onto these binary (hydr)oxide systems, the new approach explains the observed potential effects. The proposed model, which contains two additional adjustable parameters in terms of site concentrations or one adjustable parameter in terms of specific surface area, is more successful than the two limiting cases. The new model is then validated by predicting Ca and Zn behaviour on the same binary (hydr)oxide system.     

Development of a fuzzy-stochastic nonlinear model to incorporate aleatoric and epistemic uncertainty

Site uncertainties significantly influence groundwater flow and contaminant transport predictions. Aleatoric and epistemic uncertainty are both identified in site characterization and represented using proper uncertainty theories. When one theory best represents one parameter whereas a different theory may be more suitable for another parameter, the hybrid propagation of aleatoric (random) and epistemic (nonrandom) uncertainties will occur. The computational challenges of joint propagation of aleatoric and epistemic uncertainty through groundwater flow and contaminant transport models are significant. A fuzzy-stochastic nonlinear model was developed in this paper to incorporate these two types of uncertain site information and reduce the computational cost. The results show that (1) the computational cost using the nonlinear model is reduced compared with that of using the sparse grid algorithm and Monte Carlo methods; (2) the uncertainty of hydraulic conductivity (K) significantly influences the water head and solute distribution at the observation wells compared to other uncertain parameters, such as the storage coefficient and the distribution coefficient (K_d); and (3) the combination of multiple uncertain parameters substantially affects the simulation results. Neglecting site uncertainties may lead to unrealistic predictions.     

Capability and limitations of first-order and diffusion approaches to describe long-term sorption of chlortoluron in soil

This paper compares the capability of a first-order and a spherical diffusion model to describe and predict long-term sorption and desorption processes of chlortoluron in two soils. Chlortoluron sorption was investigated at different time scales utilizing one rate experiment (120 days) and two sorption/desorption experiments. Experimental periods for sorption and desorption were set to 1 day (five desorption steps) and 30 days (three desorption steps), respectively. Upon fitting, the two models satisfactorily described the whole set of data. The spherical diffusion model performed better than the first-order model. We then tested the predictive capability of the models by predicting 30-day sorption/desorption data using kinetic parameters fitted on 1-day sorption/desorption data only. While the spherical diffusion model was able to predict the 30-day data set, the first-order model failed completely. Fitting both models to subsets of the data corresponding to different experimental time scales revealed that the rate parameter as well as the Freundlich coefficient of the first-order model are strongly time-dependent--a property that is not shared by parameters of the spherical diffusion model. The apparent stability of the spherical diffusion model with regard to time dependency of its parameters indicates that sorptive uptake may be diffusion-controlled. This also explains the models greater predictive power across different time scales compared to the first-order model. Finally, we investigate the suitability of solute class specific log-linear relationships between the first-order rate parameter and the Freundlich coefficient presented by earlier researchers in the light of the time dependency observed for the parameters of the first-order model.     

Modelling of silica diffusion experiments with Si in Boom Clay

A mathematical model describing the dissolution of nuclear glass directly disposed in clay combines a first-order dissolution rate law with the diffusion of dissolved silica in clay. According to this model, the main parameters describing the long-term dissolution of the glass are ηR, the product of the diffusion accessible porosity η and the retardation factor R, and the apparent diffusion coefficient D_app of dissolved silica in clay.For determining the migration parameters needed for long-term predictions, four Through-Diffusion (T-D) experiments and one percolation test have been performed on undisturbed clay cores. In the Through-Diffusion experiments, the concentration decrease after injection of ³²Si (radioactive labelled silica) was measured in the inlet compartment. At the end of the T-D experiments, the clay cores were cut in thin slices and the activity of labelled silica in each slice was determined. The measured activity profiles for these four clay cores are well reproducible.Since no labelled silica could be detected in the outlet compartments, the Through-Diffusion experiments are fitted by two In-Diffusion models: one model assuming linear and reversible sorption equilibrium and a second model taking into account sorption kinetics. Although the kinetic model provides better fits, due to the sufficiently long duration of the experiments, both models give approximately similar values for the fit parameters. The single percolation test leads to an apparent diffusion coefficient value about two to three times lower than those of the Through-Diffusion tests.Therefore, dissolved silica appears to be strongly retarded in Boom Clay. A retardation factor R between 100 and 300 was determined. The corresponding in situ distribution coefficient K_d is in the range 25–75 cm³ g⁻¹. The apparent diffusion coefficient of dissolved silica in Boom Clay is estimated between 2×10⁻¹³ and 7×10⁻¹³ m² s⁻¹. The pore diffusion coefficient is in the range from 6×10⁻¹¹ to 1×10⁻¹⁰ m² s⁻¹.     

Evaluating non-equilibrium solute transport in small soil columns

Displacement studies on leaching of bromide and two pesticides (atrazine and isoproturon) were conducted under unsaturated steady state flow conditions in 24 small undisturbed soil columns (5.7 cm in diameter and 10 cm long) each collected from two sites differing in soil structure and organic carbon content in North Germany. There were large and irregular variabilities in the characteristics of both soils, as well as in the shapes of breakthrough curves (BTCs) of different columns, including some with early breakthrough and increased tailing, qualitatively indicating the presence of preferential flow. It was estimated that one preferential flow column (PFC) at site A, and four at site B, contributed, respectively to 11% and 58% of the accumulated leached fraction and to more than 80% of the maximum observed standard deviation (SD) in the field-scale concentration and mass flux of pesticides at two sites. The bromide BTCs of two sites were analyzed with the equilibrium convection-dispersion equation (CDE) and a non-equilibrium two-region/mobile-immobile model. Transport parameters of these models for individual BTCs were determined using a curve fitting program, CXTFIT, and by the time moment method. For the CDE based equilibrium model, the mean values of retardation factor, R, considered separately for all columns, PFCs or non-preferential flow columns (NPFCs) were comparable for the two methods; significant differences were observed in the values of dispersion coefficients of two sites using the two estimation methods. It was inferred from the estimated parameters of non-equilibrium model that 5-12% of water at site A, and 12% at site B, was immobile during displacement in NPFCs. The corresponding values for PFCs of two sites were much larger, ranging from 25% to 51% by CXTFIT and from 24% to 72% by the moment method, suggesting the role of certain mechanisms other than immobile water in higher degrees of non-equilibrium in these columns. Peclet numbers in PFCs of both sites were consistently smaller than five, indicating the inadequacy of the non-equilibrium model to incorporate the effect of all forms of non-equilibrium in PFCs. Overall, the BTCs of individual NPFCs, PFCs and of field average concentration at the two sites were better reproduced with parameters obtained from CXTFIT than by the moment method. The moment method failed to capture the peak concentrations in PFCs, but tended to describe the desorption and tail branches of BTCs better than the curve fitting approach.     

Comparison of RBCA and CalTOX for setting risk-based cleanup levels based on inhalation exposure

Risk-based corrective action (RBCA) and CalTOX (California EPA) are often used to develop risk-based soil cleanup levels. The determination of the entry parameters, including slope factors, degradation assumption, methodologies, and dispersion models for these two approaches greatly affect the onsite/offsite cleanup levels, risk distribution, and ranking of the influential factors. The subsurface soil-to-ambient air was considered as the only significant exposure pathway in this study. RBCA and CalTOX apply analytical equations and multimedia fugacity model, respectively, to simulate the transport of contaminants from subsurface soil to ambient air. Nine carcinogenic organic contaminants were selected as the target compounds. Environmental monitoring data collected from a contaminated site in southern Taiwan was used as model inputs. In this study, degradation assumption had greater influence on CalTOX evaluation than slope factors. The cleanup soil levels of all target chemicals developed by both models were close under the same slope factors and degradation assumptions, except for vinyl chloride and hexachlorobenzene. Furthermore, RBCA generally had larger offsite dispersion ratios than CalTOX, especially for long distances. The risk distribution obtained by RBCA was much board than by CalTOX. When 95th percentile was considered as the starting point, the SSTLs derived by RBCA were much stricter than by CalTOX. The ranking of influential factors in the onsite risk assessment for these two models were completely different because of their distinct model methodologies.     

Modeling air/sea flux parameters in a coastal area: A comparative study of results from the TOGA COARE model and the NOAA Buoy model

Because estuaries and coastal regions are particularly susceptible to nutrient over-enrichment due to their close proximity to source-rich regions, a goal of the BRACE study was to improve estimates of nitrogen air/sea transfer rates in the Tampa Bay Estuary. Our objective was to critically evaluate two air/sea gas exchange models to determine their efficacy for use in a coastal region, with the ultimate goal of improving nitrogen exchange estimates in Tampa Bay. We used meteorological data and oceanographic parameters collected hourly at an instrumented tower located in Middle Tampa Bay, Florida. The data was used to determine the friction velocity and the turbulent flux of heat and moisture across the air/sea interface and then compared with modeled parameters at the same offshore site. On average both models underpredicted sensible heat flux and there was considerable scatter in the data during stable conditions, indicating that nitrogen gas exchange rates may also be underestimated. Model improvement, however, was observed with friction velocity comparisons. Model inter-comparisons of sensible heat flux and friction velocity suggest excellent agreement between the TOGA COARE and the NOAA Buoy models, but model estimated heat transfer coefficients and latent heat fluxes did not agree as well. Based on our analysis, we conclude that both models are suitable for use in a coastal environment to estimate nitrogen air/sea gas exchange, although the NOAA Buoy model requires fewer meteorological inputs. However, if the purpose is to conduct more sophisticated microscale modeling of air/sea interactions, we recommend the TOGA COARE model.