Evaluation of the spatial removal of nitrogen in the marshland upwelling system

jected into the system ranged from 45-143 mg-N/L, averaging 98 mg-N/L. Groundwater salinities within the system ranged from 0-13 ppt, averaging 4 ppt. Additionally, batch test pH (6.75) and temperature (21 °C) conditions were designed to emulate those observed in MUSs. The mean pH and temperature values observed within the Bayou Segnette system were 6.69 and 21.3 °C, respectively. Similar pH and temperature values have been recorded for all systems researched to date (Fontenot, 2003; Evans, 2005). 4.2.3 Sorption Isotherms Isotherms were constructed using the data obtained from the batch adsorption tests. The experimental isotherms were formed by plotting the mass of ammonium sorbed per mass of soil versus the ammonium concentration in solution at equilibrium. The experimental isotherms were then modeled using linear, Freundlich, and Langmuir sorption isotherms. Following is a general description of the sorption isotherms and associated retardation factors (refer to Table 4.3) used to describe the experimental data. Table 4.3 Equations used to model ammonium adsorption and retardation. Sorption Isotherm Isotherm Equation Retardation Equation Linear eqd CKq = (4.4) ddf KBR θ+=1 (4.5) Freundlich NeqKCq = (4.6) θ 1 1 − += N eqd ff KNCB R (4.7) Langmuir eq eq C C q α αβ += 1 (4.8) ( )     ++= 211 eq d fl C B R α αβ θ (4.9) q= mass of solute sorbed per dry unit weight of sorbent; Ceq = concentration of solute in equilibrium with the mass of solute sorbed; Kd = distribution coefficient, K = constant, N = constant, α = adsorption constant related to the binding energy; β = maximum amount of solute that can be sorbed by the solid; Bd = bulk density of the solid; θ = effective porosity of the solid 81 The linear sorption isotherm is mathematically the simplest of the sorption isotherms and is defined by Equation 4.4. The simplicity of the linear sorption isotherm is appealing from a modeling standpoint, but it also limits its applicability. Despite the verity that the mass of solute sorbed to a solid must be finite, the linear sorption isotherm theoretically implies that an infinite amount of solute can be sorbed onto a solid. Additional limitations occur when applying the linear sorption isotherm to a limited number of data points. Sorption isotherms fit using only a few data points may erroneously represent curvilinear data as a linear relationship. For these reasons, it is imperative to note that a linear relationship exists only within the experimental range. Values of interest should never be extrapolated beyond this range (Fetter, 1999). The distribution coefficient obtained from the linear sorption isotherm can be used to estimate the retardation factor, Rf (Equation 4.5). The Freundlich sorption isotherm is the most general of the nonlinear isotherms and is given by Equation 4.6. For values of N greater than 1, the plot of q versus Ceq is curvilinear with a spreading front; for values of N less than 1, the plot of q versus Ceq is curvilinear with a self- sharpening front; for a value of N equal to 1, the Freundlich sorption isotherm simplifies to the linear sorption isotherm. The Freundlich isotherm is similar to the linear isotherm in that the mass of solute sorbed does not approach an upper limit. Thus, this equation should not be extrapolated beyond the experimental range (Fetter, 1999). The sorption parameters obtained by fitting the Freundlich isotherm can be used to estimate the retardation factor, Rff (Equation 4.7). To address the limitations in applicability of the linear and Freundlich isotherms, the Langmuir sorption isotherm was developed. The Langmuir sorption isotherm, based on the verity that a finite number of sorption sites exist on a solid surface, can be expressed using Equation 4.8. Langmuir isotherms fit to the plot of Ceq as a function of q have a curved shape approaching a maximum value (Fetter, 1999). The sorption parameters estimated by the Langmuir isotherm can be used to estimate the retardation factor, Rfl. (Equation 4.9). 82 4.3 Results and Discussion 4.3.1 Ammonium Batch Adsorption Study With each combination of soil media (M), salinity (S), and initial ammonium concentration (C) being analyzed in triplicate, the ammonium batch tests provided a total of 270 data points. Prior to analyzing the data set, three data points were removed due to erroneous values. The data points were determined erroneous based on equilibrium concentrations which were at least 20 mg-N/L greater than the initial concentration. The abnormally high equilibrium concentrations were the result of human error during solution preparation. In all, 267 data points were used to analyze the batch shake tests: 88 from batch tests performed on BSM, 89 from batch tests performed on BSC, and 90 from batch tests performed on MPS. The 267 data points were applied to Equation 4.2 and ANOVA was used to determine the significance of the interaction terms. Of particular interests were the two- and three-factor fixed effect interactions. Statistical analysis (ANOVA) revealed that only the two-factor fixed effect interactions (SM, SC, and MC) were significant (p < 0.05). Response curves were generated by plotting the lsmeans versus one of the fixed effects forming the two-factor interaction term (Figures 4.1 (a), (b), and (c)). The generated response curves are not parallel, confirming the presence of a SM, SC, and MC interaction. In Figure 4.1 (a), the lsmeans averaged over initial ammonium concentration generally decrease with increasing salinity. Highest lsmeans are observed for BSC samples at all salinities, independent of initial ammonium concentration, except at 10 ppt. In Figure 4.1 (b), the lsmeans averaged over soil media increase with increasing initial ammonium concentration. Highest lsmeans are observed at samples analyzed at 0 ppt for all initial ammonium concentrations, independent of soil media. In Figure 4.1 (c), the lsmeans averaged over salinity increase with increasing initial ammonium concentration. Highest lsmeans are observed for BSC samples at all initial ammonium concentrations, independent of salinity. 83 BSM BSC MPS 0 100 200 300 400 500 600 700 800 -5 0 5 10 15 Salinity (ppt) Le as t S qu ar es M ea n 0 ppt 5 ppt 10 ppt 0 100 200 300 400 500 600 700 800 0 20 40 60 80 100 120 140 160 180 200 220 Cinitial (mg-N/L) Le as t S qu ar es M ea n BSM MPS BSC 0 100 200 300 400 500 600 700 800 0 20 40 60 80 100 120 140 160 180 200 220 Cinitial (mg-N/L) Le as t S qu ar es M ea n Figure 4.1 Response curves for the a) SM interaction, b) SC interaction, and c) CM interaction. Bonferroni pairwise comparisons were conducted on data points sharing common x-axis values in Figure 4.1 (a), (b), and (c). Table 4.4 summarizes the Bonferroni pairwise comparisons conducted on the points illustrating the SM interaction. The BSM and BSC lsmeans averaged a) b) c) 84 over initial ammonium concentration are not significantly different at any of the salinity levels. The BSM, BSC, and MPS lsmeans averaged over initial ammonium concentration are not significantly different at the 10 ppt salinity level. The MPS lsmeans averaged over the initial ammonium concentrations are significantly different from those determined for the BSM and BSC at the 0 and 5 ppt levels. Table 4.5 summarizes the Bonferroni pairwise comparisons conducted on the points illustrating the SC interaction. The BSM and BSC lsmeans averaged over salinity level are not significantly different for any of the initial ammonium concentrations. The BSM and MPS lsmeans averaged over salinity are not significantly different for initial ammonium concentrations less than 120 mg-N/L. The BSC and MPS lsmeans averaged over salinity are not significantly different for initial ammonium concentrations less than 100 mg-N/L. All other data points sharing common x-axis values in Figure 4.1 (b) are significantly different. Table 4.6 summarizes the Bonferroni pairwise comparisons conducted on the points illustrating the MC interaction. The 0 ppt and 5 ppt lsmeans averaged over soil media are not significantly different for initial ammonium concentrations of 20 and 40 mg-N/L. The 0 ppt and 10 ppt lsmeans averaged over soil media are not significantly different for an initial ammonium concentration of 20 mg-N/L. The 5 ppt and 10 ppt lsmeans obtained by averaging over the soil media are not significantly different for initial ammonium concentrations less than 120 mg-N/L. All other data points sharing common x-axis values in Figure 4.1 (c) are significantly different. Table 4.4 Pairwise comparison of the SM interaction data points. Bonferroni Pairwise Comparison (α = 0.05) BSC MPS Salinity ppt 0 5 10 0 5 10 0 NS -- ---- S -- ---- 5 -- NS -- -- S -- BSM 10 -- -- NS -- -- NS 0 -- -- -- S -- -- 5 ---- ---- -- ---- S -- BSC 10 -- -- -- -- ---- NS NS= not significant (p>0.05), S= significant (p<0.05), --= comparison not of interest 85 Table 4.5 Pairwise comparison of the SC interaction data points. Bonferroni Pairwise Comparison (α = 0.05) BSC MPS Cinitial mg-N/L 20 40 60 80 100 120 140 160 180 200 20 40 60 80 100 120 140 160 180 200 20 NS -- --- -- -- -- -- -- -- -- NS -- -- -- -- -- -- -- -- -- 40 -- NS -- -- -- -- -- -- -- -- -- NS -- -- -- -- -- -- -- -- 60 -- -- NS -- -- -- -- -- -- -- -- -- NS -- -- -- -- -- -- -- 80 -- -- -- NS -- -- -- -- -- -- -- -- -- NS -- -- -- -- -- -- 100 -- -- -- -- NS -- -- -- -- -- -- -- -- -- NS -- -- -- -- -- 120 -- -- -- -- -- NS -- -- -- -- -- -- -- -- -- S -- -- -- -- 140 -- -- -- -- -- -- NS -- -- -- -- -- -- -- -- -- S -- -- -- 160 -- -- -- -- -- -- -- NS -- -- -- -- -- -- -- -- -- S -- -- 180 -- -- -- -- -- -- -- -- NS -- -- -- -- -- -- -- -- -- S -- BSM 200 -- -- -- -- -- -- -- -- -- NS -- -- -- -- -- -- -- -- -- S 20 -- -- -- -- -- -- -- -- -- -- NS -- --- -- -- -- -- -- -- -- 40 -- -- -- -- -- -- -- -- -- -- -- NS -- -- -- -- -- -- -- -- 60 -- -- -- -- -- -- -- -- -- -- -- -- NS -- -- -- -- -- -- -- 80 -- -- -- -- -- -- -- -- -- -- -- -- -- NS -- -- -- -- -- -- 100 -- -- -- -- -- -- -- -- -- -- -- -- -- -- S -- -- -- -- -- 120 -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- S -- -- -- -- 140 -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- S -- -- -- 160 -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- S -- -- 180 -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- S -- BSC 200 -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- S NS= not significant (p>0.05), S= significant (p<0.05), --= comparison not of interest 86 Table 4.6 Pairwise comparison of the CM interaction data points. Bonferroni Pairwise Comparison (α = 0.05) 5 ppt 10 ppt Cinitial mg-N/L 20 40 60 80 100 120 140 160 180 200 20 40 60 80 100 120 140 160 180 200 20 NS -- --- -- -- -- -- -- -- -- NS -- -- -- -- -- -- -- -- -- 40 -- NS -- -- -- -- -- -- -- -- -- S -- -- -- -- -- -- -- -- 60 -- -- S -- -- -- -- -- -- -- -- -- S -- -- -- -- -- -- -- 80 -- -- -- S -- -- -- -- -- -- -- -- -- S -- -- -- -- -- -- 100 -- -- -- -- S -- -- -- -- -- -- -- -- -- S -- -- -- -- -- 120 -- -- -- -- -- S -- -- -- -- -- -- -- -- -- S -- -- -- -- 140 -- -- -- -- -- -- S -- -- -- -- -- -- -- -- -- S -- -- -- 160 -- -- -- -- -- -- -- S -- -- -- -- -- -- -- -- -- S -- -- 180 -- -- -- -- -- -- -- -- S -- -- -- -- -- -- -- -- -- S -- 0 ppt 200 -- -- -- -- -- -- -- -- -- S -- -- -- -- -- -- -- -- -- S 20 -- -- -- -- -- -- -- -- -- -- NS -- --- -- -- -- -- -- -- -- 40 -- -- -- -- -- -- -- -- -- -- -- NS -- -- -- -- -- -- -- -- 60 -- -- -- -- -- -- -- -- -- -- -- -- NS -- -- -- -- -- -- -- 80 -- -- -- -- -- -- -- -- -- -- -- -- -- NS -- -- -- -- -- -- 100 -- -- -- -- -- -- -- -- -- -- -- -- -- -- NS -- -- -- -- -- 120 -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- S -- -- -- -- 140 -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- S -- -- -- 160 -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- S -- -- 180 -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- S -- 5 ppt 200 -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- S NS= not significant (p>0.05), S= significant (p<0.05), --= comparison not of interest 87 Bonferroni pairwise comparisons (α = 0.05) were also conducted on each of the fixed effects independent of interaction. All salinity (0, 5, and 10 ppt) lsmeans averaged over initial ammonium concentrations and soil media were found to be significantly different from one another (p < 0.05). All initial ammonium concentration (20, 40, 60, 80, 100, 120, 140, 160, 180, and 200 mg-N/L) lsmeans averaged over salinity and soil media were found to be significantly different from one another (p < 0.05). Some media lsmeans averaged over salinity and initial ammonium concentration were found to be significantly different (p < 0.05). Specifically, the BSM and MPS lsmeans were significantly different as well as the BSC and MPS lsmeans. The BSM and BSC lsmeans obtained by averaging over the salinity and initial ammonium concentration were not significantly different (p > 0.05). 4.3.2 Ammonium Sorption Isotherms Ammonium sorption parameters were estimated using Langmuir, linear and Freundlich isotherms fit to the plots of experimental data. The plots were constructed by charting mean q values versus mean Ceq values. Ceq values were determined during the batch shake tests and used to calculate q values. The mean Ceq and q values (refer to Tables 4.7 and 4.8, respectively) are an average of the Ceq and q values measured in triplicate for each SCM combination. Table 4.7 Summary of ammonium equilibrium concentrations determined by batch shake tests. Mean Ceq (mg-N/L) BSM BSC MPS Cinitial (mg-N/L) 0 ppt 5 ppt 10 ppt 0 ppt 5 ppt 10 ppt 0 ppt 5 ppt 10 ppt 20 13 15 16 8 14 16 14 15 16 40 23 27 30 17 25 29 27 30 31 60 36 40 43 26 37 43 40 46 47 80 45 53 59 37 49 58 54 61 60 100 56 67 71 48 64 71 69 77 78 120 69 81 85 59 79 87 85 94 95 140 81 93 99 71 90 100 98 108 113 160 93 107 115 82 104 116 114 127 127 180 107 121 131 96 117 132 125 145 143 200 119 136 142 110 133 142 146 159 163 88 Table 4.8 Summary of ammonium adsorption capacities determined by batch shake tests. Mean q (mg/kg) BSM BSC MPS Cinitial (mg-N/L) 0 ppt 5 ppt 10 ppt 0 ppt 5 ppt 10 ppt 0 ppt 5 ppt 10 ppt 20 66 53 37 119 65 44 65 46 44 40 166 130 104 234 147 107 128 98 93 60 245 199 165 340 232 169 200 144 134 80 353 267 213 433 307 217 260 188 198 100 437 328 287 524 361 285 310 232 222 120 510 390 352 611 414 329 354 265 252 140 593 473 410 695 500 396 420 323 272 160 668 535 452 780 556 443 463 333 333 180 727 587 495 837 625 480 546 350 371 200 813 642 578 896 669 581 536 407 369 The ammonium sorption isotherms fit to the BSM, BSC, and MPS batch tests data are provided in Figures 4.2, 4.3, and 4.4, respectively. Parameters describing the fitted ammonium sorption isotherms as well as their associated retardation factors are available in Table 4.9. Figure 4.2 includes three plots corresponding to BSM samples equilibrated in solutions of 0, 5, and 10 ppt. These plots were effectively modeled using the linear isotherm. However, the Langmuir isotherm better approximated q at higher aqueous ammonium concentrations. The BSM sorption capacities (q) decreased with respect to increasing salinity. This may have been caused by Na+ ions from the saline solution competing with NH4+ ions for available exchange sites. Despite NH4+ having a greater exchange site affinity, the incremental increases in Na+ may have caused the preferential sorption of Na+. Alternatively, this may have been the result of interactions with cations previously sorbed to the soil. Figure 4.3 includes the sorption isotherms fit to the data obtained from the batch tests performed on the BSC. The plot corresponding to the 0 ppt salinity level was non-linear and best approximated by the Langmuir isotherm. The plots corresponding to the 5 and 10 ppt salinity levels were approximately linear, but the leading fronts were better approximated by the Langmuir isotherm. The BSC sorption capacities decreased with respect to increasing salinity. 89 This may have been due to the preferential sorption of Na+ caused by incremental increases in salinity or ammonium interactions with cations previously sorbed to the soil. Figure 4.4 includes the sorption isotherms fit to the MPS data. The plots corresponding to the 0, 5, and 10 ppt salinity levels were approximately linear with the leading fronts being better approximated by the Langmuir isotherm. The MPS sorption capacities decreased when salinity was increased from 0 to 5 ppt, but showed little change when salinity was increased from 5 to 10 ppt. This may have been due to the competitive sorption of Na+ or interactions with cations previously sorbed to the soil. The negligible change in q, noted as salinity increased from 5 to 10 ppt, may be indicative of Na+ precipitation, making it unavailable for adsorption. Alternatively, the Na+ adsorption capacity of the MPS may be less than that of the BSM and BSC causing Na+ to equilibrate at lower aqueous concentrations. A comparison of the r-square values determined for each of the sorption isotherms confirms that the Langmuir isotherm provided a better fit than the linear isotherm in most instances. The importance of proper isotherm selection is illustrated by the sensitivity of the retardation factors. Retardation factors calculated using isotherms with lower r-square values were smaller than those calculated using isotherms with higher r-square values. For modeling purposes, the Bayou Segnette and Moss Point adsorption capacities could be estimated using a linear isotherm. However, given the limitations of the linear isotherm and the better fit provided by the Langmuir isotherm at higher equilibrium concentrations, the Langmuir isotherm was selected as the best suited representation. This is consistent with previous research which has illustrated that most ammonium isotherms are non-linear at higher aqueous concentrations. Some are non-linear at aqueous concentrations below 50 mg-N/L (Bus et al, 2003). Langmuir isotherms fit to data obtained through adsorption studies conducted using Queenston Shale, Fonthill Sand, and Niagara Shale samples provided maximum adsorption capacities (β) of 835 mg/kg, 746 mg/kg, and 443 mg/kg, respectively (Rozema, 1997). These values are substantially lower than those estimated for the Bayou Segnette and Moss Point soil samples. 90 0 200 400 600 800 1000 1200 0 25 50 75 100 125 150 Ceq (mg-N/L) q (m g/ kg ) Langmuir Linear Freundlich 0 200 400 600 800 1000 1200 0 25 50 75 100 125 150 Ceq (mg-N/L) q (m g/ kg ) Langmuir Linear Freundlich 0 200 400 600 800 1000 1200 0 25 50 75 100 125 150 Ceq (mg-N/L) q (m g/ kg ) Langmuir Linear Freundlich Figure 4.2 Ammonium sorption isotherms fit to data obtained from batch shake tests using BSM samples equilibrated in saline solutions of (a) 0 ppt, (b) 5 ppt, and (c) 10 ppt. a) b) c) 91 0 200 400 600 800 1000 1200 0 25 50 75 100 125 150 Ceq (mg-N/L) q (m g/ kg ) Langmuir Linear Freundlich 0 200 400 600 800 1000 1200 0 25 50 75 100 125 150 Ceq (mg-N/L) q (m g/ kg ) Langmuir Linear Freundlich 0 200 400 600 800 1000 1200 0 25 50 75 100 125 150 Ceq (mg-N/L) q (m g/ kg ) Langmuir Linear Freundlich Figure 4.3 Ammonium sorption isotherms fit to data obtained from batch shake tests using BSC samples equilibrated in saline solutions of (a) 0 ppt, (b) 5 ppt, and (c) 10 ppt. a) b) c) 92 0 200 400 600 800 1000 1200 0 25 50 75 100 125 150 175 Ceq (mg-N/L) q (m g/ kg ) Langmuir Linear Freundlich 0 200 400 600 800 1000 1200 0 25 50 75 100 125 150 175 Ceq (mg-N/L) q (m g/ kg ) Langmuir Linear Freundlich 0 200 400 600 800 1000 1200 0 25 50 75 100 125 150 175 Ceq (mg-N/L) q (m g/ kg ) Langmuir Linear Freundlich Figure 4.4 Ammonium sorption isotherms fit to data obtained from batch shake tests using MPS samples equilibrated in saline solutions of (a) 0 ppt, (b) 5 ppt, and (c) 10 ppt. a) b) c) 93 Table 4.9 Parameters and retardation factors describing the ammonium sorption isotherms in Figures 4.2, 4.3, and 4.4. BSM BSC MPS Sorption Isotherm Isotherm Parameters 0 ppt 5 ppt 10 ppt 0 ppt 5 ppt 10 ppt 0 ppt 5 ppt 10 ppt α (L/mg) 0.0026 0.0004 0.0004 0.0078 0.0027 0.0004 0.0033 0.0029 0.0038 β (mg/kg) 3333 14286 10000 2000 2500 10000 1667 1250 1000 R2 0.9930 0.9971 0.9887 0.9968 0.9956 0.9914 0.9895 0.9921 0.9855 Langmuir Rfl 2.64 1.99 1.83 43.37 21.21 13.49 17.31 11.65 12.11 Kd (L/kg) 7.09 4.88 3.96 9.28 5.35 3.87 4.13 2.69 2.54 R2 0.9914 0.9967 0.9928 0.9340 0.9884 0.9928 0.9732 0.9697 0.9657 Linear Rf 2.39 1.96 1.78 30.18 17.80 13.15 13.99 9.44 9.00 K 4.63 3.26 1.53 26.00 5.72 2.33 6.51 4.21 3.98 N 1.10 1.09 1.21 0.77 0.99 1.11 0.91 0.91 0.91 R2 0.9905 0.9863 0.9896 0.9905 0.9863 0.9896 0.9841 0.9815 0.9773 Freundlich Rff 2.27 1.86 1.59 37.97 18.44 11.53 15.94 10.80 10.29 94 Quantifying ammonium adsorption capacities of wetland soils can provide insight into the nitrogen removal capabilities of MUSs. However, numerous factors including the total cation concentrations within the soil, the ionic strength of the wastewater and the groundwater, as well as the CEC of the soil play critical rolls in ammonium retardation. It is important to note that the application of isotherm parameters is specific to the lithology, test solution, and experimental conditions under which they were determined. In MUSs, nitrogen removal is historically expressed in terms of distance traveled from the point of injection. The Moss Point system was estimated to reduce TAN concentrations to an effluent standard of 10 mg-N/L at distances ranging from 1.93-3.34 m (Fontenot, 2003). The Bayou Segnette system required substantially longer distance estimates (3.73-4.14 m). When comparing the two systems, the Moss Point system treated TAN to effluent standards at shorter distances despite having smaller ammonium adsorption capacities at the investigated salinity levels. This suggests that factors in addition to those experimentally simulated are influencing TAN reduction within the MUS. 4.3.3 Design Implications The Langmuir isotherm parameters selected to model the ammonium adsorption of Bayou Segnette and Moss Point soils are specific to the test solutions used to determine their values. Test solutions consisting of only ammonium or mixtures of ammonium and synthetic solutions can yield overly optimistic results due to the absence of or lesser abundance of additional cations. The use of such results for risk assessment is strongly discouraged (Bus et al, 2003). Prior to conducting risk assessments, source solutions should be analyzed and compared against test solutions to ensure any measurements used are comparable with site conditions. To date, such comparisons have not been conducted using MUS source solutions. Despite the current inability to conduct formal risk assessments, the Langmuir isotherm parameters can be used to provide valuable information about the adsorptive capabilities of wetland soils. The Langmuir isotherm parameters measured for the BSM, BSC, and MPS samples 95 were used to determine the q values needed to approximate ammonium saturation times (refer to Table 4.10). The ammonium saturation times were determined for two nitrogen loading rates. The nitrogen loading rates were calculated assuming a constant influent ammonium concentration of 98 mg-N/L and hydraulic loading rates of 224 L/day and 393 L/day. The initial ammonium concentration is equal to the mean ammonia concentration injected into the Bayou Segnette MUS. The hydraulic loading rates are equivalent to the maximum and minimum rates estimated for the Bayou Segnette MUS. A bulk density of 0.1 g/cm3 was assumed representative of the BSM samples and a bulk density of 1.6 g/cm3 was assumed representative of the BSC and MPS samples. These values were selected based on the typical ranges of bulk densities for organic and mineral soils being 0.05 to 1.0 g/cm3 and 1.2 to 1.7 g/cm3, respectively (USDA, 1983). A soil volume of 100 m3 was arbitrarily selected for demonstrative purposes and does not represent the actual volume of soil within either of the MUSs. It is also important to note that ammonium saturation has not been reached within any of the MUSs Table 4.10 Predicted ammonium saturation times of Bayou Segnette and Moss Point soil samples. Ammonium saturation time (yr) at varying nitrogen loading rates (NLR) Soil Sample Salinity (ppt) q (mg/kg) NLR = 22.0 kg-N/day NLR = 38.5 kg-N/day 0 684 0.85 0.49 5 481 0.60 0.34 BSM 10 401 0.50 0.29 0 869 17.35 9.89 5 525 10.49 5.98 BSC 10 378 7.54 4.30 0 409 8.18 4.66 5 274 5.48 3.12 MPS 10 272 5.43 3.09 The purpose of these calculations was to illustrate that while the degree of ammonium storage within a MUS depends heavily on the ammonium adsorption capacity, additional parameters including nitrogen loading rate and soil bulk density can strongly influence the life of the system. Increasing the nitrogen loading rate increases the mass of ammonium entering the 96 system and thus decreases the time to saturation. A comparison of the BSM and BSC shows that the time to saturation drastically increases for soils with larger bulk densities. Where possible, MUSs should be installed in soils characterized by large sorption capacities and high bulk densities. Systems should be operated under the lowest hydraulic loading rate required to meet household demands. The lowest hydraulic loading rate, assuming a constant influent nitrogen concentration, would produce the lowest nitrogen loading rate. MUSs installed and operated under such conditions will provide longer ammonium saturation times. 4.4 Conclusions The ammonium batch shake tests revealed the following findings for soil samples collected from the Bayou Segnette and Moss Point project sites: 1. All of the two-factor fixed effect interactions (SM, SC, and MC) used in the experimental design are significant (p < 0.05). The three-factor (SCM) fixed effect interaction is not significant (p > 0.05) 2. The ammonium adsorption capacities of the BSM and BSC are not significantly different (p > 0.05). The effect of salinity on the ammonium adsorption capacities of the BSM is not significantly different from the effect of salinity on the ammonium adsorption capacities of the BSC (p > 0.05). The effect of initial ammonium concentration on the ammonium adsorption capacities of the BSM is not significantly different from the effect of initial ammonium concentration on the ammonium adsorption capacities of the BSC (p > 0.05). 3. The ammonium adsorption capacities of the MPS are significantly different from the ammonium adsorption capacities of the BSM and BSC (p < 0.05). The effect of salinity on the ammonium adsorption capacities of the MPS is significantly different from the effect of salinity on the ammonium adsorption capacities of the BSM and BSC at the 0 and 5 ppt levels (p < 0.05). The effect of initial ammonium concentration on the 97 ammonium adsorption capacities of the MPS is significantly different from the effect of initial ammonium concentration on the ammonium adsorption capacities of the BSM at initial concentrations greater than 100 mg-N/L. The effect of initial ammonium concentration on the ammonium adsorption capacities of the MPS is significantly different from the effect of initial ammonium concentration on the ammonium adsorption capacities of the BSC at initial concentrations greater than 80 mg-N/L. 4. For modeling purposes, the linear sorption isotherm can be used to approximate ammonium adsorption in the BSM, BSC, and MPS. However, under most instances the Langmuir sorption isotherm provided a better estimate than the linear sorption isotherm at higher aqueous concentrations. The mean ammonia concentration injected into the Bayou Segnette MUS is 98 mg-N/L. The Langmuir sorption isotherm should be used to model ammonium adsorption at locations near the point of injection. The linear sorption isotherm can be used to model ammonium adsorption at dilute locations further from the point of injection. 5. Salinity greatly impacts the degree of ammonium sorption in each of the analyzed soils. At 0 ppt, soil samples in order of greatest ammonium adsorption capacity are BSM>BSC>MPS. At 5 ppt, soil samples in order of greatest ammonium adsorption capacity are BSM>BSC>MPS. At 10 ppt, soil samples in order of greatest ammonium adsorption capacity are BSC>BSM>MPS. The ammonium adsorption capacities of MPS at 5 and 10 ppt were approximately equal. 98 Chapter 5: Global Discussion and Conclusion The purpose of this research study was to investigate the nitrogen reduction capabilities of MUSs operating under low saline background conditions. The objectives of this research study were to: 1) determine the removal constants necessary for the future development of nitrogen transport equations, 2) explore the spatial dependencies of nitrogen concentrations within the Bayou Segnette system, and 3) determine the nitrogen adsorptive capacities for the Bayou Segnette and Moss Point soil matrices. The research presented in this thesis was divided into three main sections (refer to Chapters 2-4), with each section focusing on a specific objective. The first section discussed the removal capabilities of a MUS operating under low saline background conditions and varying flow regimes. The flow regimes evaluated during this research were 0.95 L/min for 15 min/hr, 1.89 L/min for 15 min/hr, and 1.89 L/min for 30 min/hr. Nitrogen removal constants were estimated for each of the flow regimes using an area-based first- order model. TAN and TKN removal constants exhibited little change in value with response to altering flow regimes. TAN and TKN removal constants were used to predict the travel distances required to meet the assumed effluent regulatory standard of 10 mg-N/L. Flow regimes producing lower hydraulic loading rates resulted in shorter travel distances. Travel distances ranging from 3.73-4.14 m and 4.29-4.89 m were predicted to provide TAN and TKN concentrations equal to the effluent standard, respectively. Observed TAN and TKN concentrations within the Bayou Segnette system were reduced to levels below the effluent standard at vector distances greater than 4.58 m. Overall nitrogen removal efficiencies of the Bayou Segnette MUS were in excess of 98% for TAN reduction and 96% for TKN reduction. Based on the aforementioned findings, it was concluded that the nitrogen removal capabilities of the Bayou Segnette MUS were not adversely affected by the low groundwater salinities (averaging 6 ppt) native to the project site. 99 The second section discussed the spatial trend and modeling of nitrogen concentrations within the MUS. The spatial trend was identified by applying exploratory and spatial data analyses to the mean monitoring well TAN concentrations observed during each of the flow regimes. Both the exploratory and spatial analyses unveiled a similar spatial trend. The spatial trend existed during each of the flow regimes and was defined by increasing TAN concentrations in a northwest direction. The trend is thought to be explained by the influence of the Bayou Segnette Canal and the Gulf of Mexico on the native groundwater flows. Variogram modeling performed on general linear regression residuals revealed that the regression residuals were spatially correlated in the northing-easting plane at Euclidean distances up to 2.24, 3.55, and 3.33 m for the 0.95 L/min for 15 min/hr, 1.89 L/min for 15 min/hr, and 1.89 L/min for 30 min/hr flow regimes, respectively. Additionally, regression-kriging was used to estimate the TAN concentrations along the leading edge boundaries of the modeled MUS. Of the 120 points estimated along the leading edge boundaries, approximately 100% of the estimated points representing the 0.95 L/min for 15 min/hr flow regime and 75% of the estimated points representing the remaining flow regimes were treated to levels below the assumed effluent standard of 10 mg-N/L. The leading edge boundaries were estimated at a depth of 2.7 m below the surface of the marsh, thus providing an ultra-conservative estimate of effluent concentrations. Based on the aforementioned findings, it was concluded that the nitrogen concentrations observed within the Bayou Segnette system exhibited spatial trend and were spatially dependent. The third section discussed the impact of salinity on the ammonium adsorptive capacities of three soil samples collected from the Bayou Segnette and Moss Point project sites. The ammonium adsorptive capacities of each soil were quantified during a series of batch shake tests using saline solutions of 0, 5, and 10 ppt. The data resulting from the batch shake tests were modeled using both statistical methods and sorption isotherms. Bonferroni pairwise comparisons revealed that increasing salinity had the same effect on the ammonium adsorptive capacities of the Bayou Segnette muck (BSM) and Bayou Segnette clay (BSC) soil samples (p > 0.05). 100 Additionally, Bonferroni Pairwise comparisons revealed that increasing salinity from 0 to 5 ppt had a significantly different effect on the ammonium adsorption capacities of the Moss Point soil (MPS) sample than the BSM and BSC samples (p < 0.05). Ammonium adsorptive capacities were modeled using linear, Freundlich, and Langmuir sorption isotherms. In most instances, the Langmuir isotherm provided better data estimates at higher aqueous concentrations. However, the linear sorption isotherm could be used to model ammonium adsorption at dilute locations away from the point of injection. Based on the aforementioned findings, it was concluded that increasing solution salinity adversely affected the degree of ammonium attenuation in both Bayou Segnette and Moss Point soils. The ammonium adsorption capacities of the BSM and BSC were more sensitive to increases in solution salinity than the MPS, particularly at the 10 ppt level. 101 Chapter 6: Design Implications and Research Recommendations The research presented in this thesis provides valuable insight into the nitrogen removal capabilities of MUSs installed in low (<10 ppt) saline groundwaters. The Bayou Segnette system effectively reduced nitrogen concentrations to levels below the self-imposed effluent standard of 10 mg-N/L. The maximum predicted travel distance required to obtain this level of treatment was 4.89 m. The shortest travel distance within the MUS is a straight line from the primary injection wellhead to the marsh surface. This distance is 4.27 m in the Bayou Segnette system. Though the actual travel trajectories are likely much greater than 4.27 m, the primary injection well should be installed at a depth greater than 4.89 m to ensure adequate nitrogen removal. Previous research on nitrogen removal within the Moss Point MUS (installed in saline groundwater >30 ppt) resulted in a maximum predicted travel distance of 3.2 m (Fontenot, 2003). Fontenot (2003) recommended a safety factor be included in the injection well depth and suggested a depth of 5.0 m for future MUSs. Nitrogen research on the Bayou Segnette system confirms that this injection depth would likely provide adequate nitrogen removal in low saline groundwaters as well. Of the three flow regimes evaluated during this research study (0.95 L/min for 15 min/hr, 1.89 L/min for 15 min/hr, and 1.89 L/min for 30 min/hr) those producing lower hydraulic loading rates treated nitrogen to effluent standards at shorter travel distances. MUSs installed under similar groundwater and soil conditions as the Bayou Segnette system should be operated using the lowest hydraulic loading rate capable of meeting household demands. The monitoring wells sampled throughout the course of this research ranged in depth from 2.7 to 4.6 m. The installation of monitoring wells at depths shallower than 2.7 m would extend the depth over which nitrogen concentrations could be estimated. Variogram modeling conducted on the regression residuals suggests that additional monitoring wells be installed no more than 3.33 m apart in the northing-easting plane. At distances greater than 3.33 m, the regression residuals are no longer spatially dependent. In addition to extending the depth over 102 which nitrogen concentrations could be estimated, the installation of shallower monitoring wells would strengthen variogram models by providing more pairs of points and increasing the number of lags. Strengthening the variogram models would in turn strengthen the kriging estimates. Regression-kriging was used to predict nitrogen concentrations at unsampled points within the system. Though this kriging method provided good results and can be applied to any MUS site, alternative kriging methods including indicator kriging and three-dimensional kriging should be explored. Indicator kriging is often used in risk assessment and imparts a binary system of zeros and ones to predict whether a point is above or below a regulatory standard. Three-dimensional kriging could be used to visualize the wastewater plume in three-dimension rather than in two- dimensional slices in depth. Ammonium adsorption is believed to be the primary nitrogen reduction mechanism within the Bayou Segnette MUS. Denitrification may be responsible for removing nitrogen at locations close to the marsh surface. However, at the current monitoring well depths the redox potential, groundwater pH, and negligible nitrate concentrations make denitrification an unlikely removal mechanism. The ammonium adsorption capacities of two Bayou Segnette soils and one Moss Point soil were quantified and modeled in an effort to assess the degree of ammonium adsorption within MUSs. The resulting sorption isotherms cannot be readily applied to model MUSs. The isotherm parameters are specific to the test solutions used to determine their values and, if used in risk assessment, may provide overly optimistic results. Prior to conducting risk assessments, wastewater diluted with site-specific groundwater should be analyzed and compared against test solutions to ensure laboratory and site conditions are comparable. Despite the inability to conduct formal risk assessments, isotherm parameters were used to evaluate the effect of soil bulk density, sorption capacity, and nitrogen loading rate on the time to ammonium saturation. Where possible, MUSs should be installed in soils characterized by large sorption capacities and high bulk densities and operated under low nitrogen loading rates. Time to ammonium saturation will be longest in systems installed and operated under such conditions. 103 References Ache, W. Brent, and Wenger, J. Seth, 1999. A survey of onsite wastewater treatment systems: Identifying alternatives appropriate for coastal Louisiana based on performance and cost data. Battelle coastal resources and ecosystems management. A publication of the Barataria- Terrebonne National Estuarine Program, Louisiana, Nov. 1999. Addo, B.K., 2004. Bacterial Removal in the Marshland Upwelling System Employed in the Near Freshwater Conditions. MS thesis. Department of Civil and Environmental Engineering, Louisiana State University, Baton Rouge, Louisiana. APHA (American Public Health Association), 1998. Standard Methods for the Examination of Water and Wastewater, 20th Edition. American Water Works Association, Water Pollution control Federation, and American Public Health Association, Washington D.C., Virginia. ASTM (American Society for Testing and Materials), 1995. Book of Standards, Soil and Rock (I), 04.08:D420-5579, Philadelphia, Pennsylvania. ASTM (American Society for Testing and Materials), 2004. Standard Guide for Analysis of Spatial Variation in Geostatistical Site Investigation. D5523-5596, Philadelphia, Pennsylvania. Bus, S.R., A.W. Herbert, P. Morgan, and S.F. Thornton, 2003. Review of ammonium attenuation in soil and groundwater. National Groundwater and Contaminated Land Centre. Environment Agency, Almondsbury, Bristol. Caezan, M.L., E.M. Thurman, and R.L. Smith, 1989. Retardation of Ammonium and Potassium Transport through a Contaminated Sand and Gravel Aquifer: the Role of Cation Exchange. Environmental Science and Technology 23(1):1402-1408. Chapman, H.D., 1965. Cation exchange capacity. In: C.A. Black (ed.) Methods of soil analysis- Chemical and microbial processes. Agronomy 9:891-901. American Society of Agronomy, Inc., Madison, Wisconsin. Charbeneau, R.J., 2000. Groundwater Hydraulics and Pollutant Transport. Prentice Hall, Upper Saddle River, New Jersey, Chapters 5-7, pp. 247-362. Corbett, D.R., K. Dillon, W. Burnett, G. Schaeffer, 2002. The Spatial Variability of Nitrogen and Phosphorus Concentration in a Sand Aquifer Influenced by Onsite Sewage Treatment and Disposal Systems: a Case Study on St. George Island, Florida. Environmental Pollution 117: 337- 345. DeBusk, William. 1999. Wastewater Treatment Wetlands: Contaminant Removal Processes. Soil and Water Science Department, Florida Cooperative Extension Service, Institute of Food and Agricultural Sciences, University of Florida, Gainesville, Florida. DNREC (Department of Natural Resources and Environmental Control). Nutrient Reducing Alternative Septic Systems, Whole Basin Team. Dover, Delaware. Domenico, P.A. and F.W. Schwartz, 1998. Physical and Chemical Hydrogeology, 2nd Edition. Wiley, New York. 104 Esmail, O.J. and Kimbler, 1967. Investigation of the Technical Feasibility of Storing Fresh Water in Saline Aquifers. Water Resources Research 3:3:683-695. EEC (European Economic Community), 1991. Council Directive of May 21, 1991 Concerning Urban Wastewater Treatment, 91/271/EEC. Evans, D.A., 2005. Assimilation and Removal of Phosphorus in the Marshland Upwelling System. MS thesis. Department of Civil and Environmental Engineering, Louisiana State University, Baton Rouge, Louisiana. Fetter, C.W., 1999. Contaminant Hydrogeology. Prentice Hall, Upper Saddle River, New Jersey, Chapter 3, pp. 120-129. Fontenot, J.P., 2003. The Fate of Nitrogen in the Marshland Upwelling System. MS thesis. Department of Civil and Environmental Engineering, Louisiana State University, Baton Rouge, Louisiana. Gambit Weekly, 2001. New Orleans Know-it-All. Hengl, T., G.B.M. Heuvelink, and A. Stein, 2004. A Generic Framework for Spatial Prediction of Soil Variables Based on Regression-Kriging. Geoderma 120: 75-93. Hudson, G. and H. Wackernagel, 1994. Mapping Temperature Using Kriging with External Drift: Theory and an Example from Scotland. International Journal of Climatology 14: 77-91. Jetten, M.S.M., 2001. New Pathways for Ammonia Conversion in Soil and Aquatic Systems. Plant and Soil 230: 9-19. Juang, K.W. and D.Y. Lee, 1998. Simple Indicator Kriging for Estimating the Probability of Incorrectly Delineating Hazardous Areas in a Contaminated Site. Environmental Science and Technology 32: 2487-2493. Kadlec, R.H. and R.L. Knight (1996). Treatment Wetlands. Lewis Publishers, Boca Raton, Florida, Chapter 13, pp. 373-440. Kaluzny, S.P., S.C. Vega, T.P. Cardoso, and A.A. Shelly, 1998. S+SPATIAL STATS. User's Manual for Windows and UNIX. York: Maple-Vail Book. Kilgen, M.B., and R.H. Kilgen, 1990. State of Louisiana Lower Atchafalaya-Barataria Basin Sanitary Sewer Survey Report: Basin 01 and 12. Louisiana Department of Health and Hospitals, Office of Public Health, Oyster Water Monitoring Program. LaGrega, M.D., P.L. Buckingham, J.C. Evans, and Environmental Resource Management, 2001. Hazardous resource Management, 2nd Edition. McGraw-Hill, Boston Massachusetts. Little, L.S., D. Edwards, and D.E. Porter, 1997. Kriging in Estuaries: as the Crow Flies, or as the Fish Swims?. Journal of Experimental Marine Biology and Ecology 213: 1-11. Metcalf & Eddy, Inc., 2003. Wastewater Engineering: Treatment and Reuse, 4th Edition. Irwin McGraw-Hill, Boston, Massachusetts. 105 NOAA (National Oceanic and Atmospheric Administration), 1997. The 1995 National Shellfish Register of Classified Growing Waters. U.S. Department of Commerce, Strategic Assessment Branch, Office of Oceanography and Marine Assessment, National Oceanic and Atmospheric Administration. NOAA (National Oceanic and Atmospheric Administration), 1998. NOAA Economic Statistics. U.S. Department of Commerce, Office of Policy and Strategic Policy. NOAA (National Oceanic and Atmospheric Administration), 1999. Estuarine Eutrophication Survey, National Assessment of Nutrient Enrichment Conditions in Estuaries. Office of Ocean Resources Conservation Assessment, Silver Spring, Maryland. NOAA (National Oceanic and Atmospheric Administration), 2002. NOAA Economic Statistics. U.S. Department of Commerce, Office of Policy and Strategic Policy. NOAA (National Oceanic and Atmospheric Administration), 2003. Draft Environmental Impact Statement. Gulf of Mexico Fishery Management Council. Tampa, Florida. NOAA (National Oceanic and Atmospheric Administration), 2004. Population Trends along the Coastal United States: 1980-2008. U.S. Department of Commerce, National Ocean Service, Management and Budget Office. Netter, J., and W. Wasserman, 1974. Applied Linear Statistical Models: Regression, Analysis of Variance, and Experimental Designs. Homewood, Illinois. Orhon, D., O. Tünay, F. Germirli, N. Artan, S. Sözen, R. Tasli, E. Ubay Çokgör, E. Görgün, 1996. Wastewater Management and Appropriate Treatment Technologies in Sensitive Areas, Technical Report. Turkish Technology Development Fund. pp. 208. Ouyang, Y., J. Higman, D. Campbell, and J. Davis, 2003. Three-Dimensional Kriging Analysis of Sediment Mercury Distribution: A Case Study. Journal of the American Water Resources Association 39(3): 689-702. Patrick, W.H., R.P. Gambrell, and S.P. Faulkner, 1996. Methods of Soil Analysis, Part 3. Soil Science Society of America and American Society of Agronomy, Madison, Wisconsin, Chapter 42, pp. 1255-1273. Reddy, K.R., R. Khaleel, and M.R. Overcash, 1981. Behavior and Transport of Microbial Pathogens and Indicator Organisms in Soils Treated with Organic Wastes. Journal of Environmental Quality 10(3): 225-256. Richardson, S.D., 2002. Bacterial Plume Dynamics in the Marshland Upwelling System. MS thesis. Department of Civil and Environmental Engineering, Louisiana State University, Baton Rouge, Louisiana. Richardson, S.D. and K.A. Rusch, 2004. Use of Rhodamine Water Tracer in the Marshland Upwelling System. Ground Water, 42(4): 678-688. Richardson, S.D. and K.A. Rusch, 2005. Fecal Coliform Removal within a Marshland Upwelling System Consisting of Scatlake Soils. Journal of Environmental Engineering 131(1): 60-70. 106 Rozema, L., 1997. MS Thesis. Department of Land Resource Science, University of Guelph, Ontario, Canada. Stremleau, H.T., 1994. Feasibility Study on the Use of Shallow Upwelling Systems in Coastal Areas as a Polishing Treatment to Remove Bacterial contamination from Wastewater, MS thesis. Louisiana State University, Baton Rouge, Louisiana. USDA (United States Department of Agriculture), 1983. Soil Survey of Jefferson Parish, Louisiana Soil Conservation Service. USEPA (United States Environmental Protection Agency), 2001. Nutrient Criteria Technical Guidance Manual: Estuarine and Coastal Marine Waters. EPA-822-B-01-003. Office of Water, Washington D.C., Virginia. USEPA (United States Environmental Protection Agency), 2002. Onsite Wastewater Treatment Systems Manual. EPA-625-R-00-008. Office of Water, Washington D.C., Virginia. USEPA (United States Environmental Protection Agency), 2004. National Coastal Condition Report II. EPA-620-R-03-002. Office of Water, Washington D.C., Virginia. W&PE (Woods and Poole Economics, Inc.), 2003. 2003 Desktop Data Files, Washington D.C., Virginia. WSDE (Washington State Department of Ecology), 2002. Water Quality Program Permit Writer’s Manual. Publication Number 92-109. Watson Jr., R.E., 2000. Evaluation of a Marshland Upwelling System for the Treatment of Raw Domestic Wastewater, MS thesis. Louisiana State University, Baton Rouge, Louisiana. 107 Vita Andrea Reneé Turriciano was born on February 19, 1980, in Redwood City, California. In May of 1998, she graduated from Mandeville High School in Mandeville, Louisiana. Immediately following high school, she moved to Baton Rouge and attended Louisiana State University where she earned the Bachelor of Science in Environmental Engineering degree. Andrea became a full- time graduate student at Louisiana State University in December of 2002 and is currently a candidate for the Master of Science in Civil Engineering degree.