Technical Memorandum. United States Army Corps of Engineers Savannah District

Transcription

1 Technical Memorandum United States Army Corps of Engineers Savannah District Savannah Harbor Expansion Project Updated Saltwater Intrusion Modeling to Support the Groundwater Monitoring Plan January 29, 2016

2 Section 1 Introduction 1.1 Purpose and Scope This technical memorandum documents the refinement and application of the Savannah Harbor Expansion Project (SHEP) groundwater flow model to project future chloride concentrations at sentry monitoring wells and selected production wells. The modeling results will be used for the purpose of developing monitoring benchmarks to evaluate the potential impacts to the Floridan aquifer by the proposed dredging of the Savannah Harbor channel. The US Army Corps of Engineers (ACOE) will develop the monitoring benchmarks in consultation with Georgia Environmental Protection Division (GaEPD). This work was completed under ACOE Contract No. W D-0010 Savannah Harbor Expansion Project Groundwater Modeling. 1.2 Project Background The Savannah Harbor Expansion Project (SHEP) includes deepening approximately 35 miles of the Savannah Harbor navigation channel (ACOE, 2002). As part of a series of investigations to determine whether deepening of the Savannah Harbor channel has the potential to impact the groundwater quality in the upper permeable zone of the Floridan aquifer within the project area, CDM Smith developed a three-dimensional coupled flow and transport groundwater model of the proposed harbor dredging (CDM, 2005). The model simulated the downward migration of salt water through the river/harbor sediments and underlying Miocene age confining unit down to the Floridan aquifer. The Floridan aquifer is the largest source of freshwater in the coastal area of Georgia, and the potential for saltwater intrusion due to heavy groundwater pumping in the Savannah area and Hilton Head Island, South Carolina, is a growing concern among the coastal communities and State and Federal agencies. The SHEP groundwater monitoring plan developed by the ACOE in coordination with Georgia Environmental Protection Division (GaEPD) Division of Natural Resources (DNR) (ACOE, 2012) includes the installation and sampling of a network of background and sentry monitoring wells in the Floridan aquifer downgradient of the harbor dredging to monitor for changes in water quality that could be due to dredging. This assessment will be made based on comparison of chloride concentrations measured in the sentry wells to benchmark chloride concentrations developed with the use of the SHEP groundwater model. Specifically, the benchmark concentrations will be developed based on model projections of future chloride concentrations at the sentry wells. SHEP groundwater model refinements and simulations projecting future chloride concentrations in the Floridan aquifer are presented in this technical memorandum. 1-1

3 Section 1 Introduction 1.3 Study Approach The overall approach adopted by CDM Smith for this study consisted of the following steps: Review of New Monitoring Well Data Soil boring logs for the new sentry monitoring wells were reviewed and compared to existing model layering. Groundwater level and chloride concentration data were compared to previous (2005) model simulation results. Review of Recent Pumping Data - Recent total pumping data reported for the Savannah area Red Zone were provided from GaEPD for this study. Model Refinement Model layering was revised and a reasonable range of Miocene confining unit hydraulic properties was determined consistent with previous modeling and new well data from the Groundwater Monitoring Report, Savannah Harbor Expansion Project (ACOE, 2015). Model input of recent pumping for the Savannah area ( Red Zone ) was updated for some simulations based on data received from GaEPD. Model Application The refined SHEP coupled groundwater flow and solute transport model was applied to make projections of chloride concentrations at the sentry wells and at selected Savannah area production wells. Simulations were made for a reasonable range of Miocene confining unit hydraulic properties to bracket a range of projected chloride concentrations. Development of Data Analysis Protocols/Tools Observed chloride concentrations at the sentry wells, in conjunction with results from groundwater model simulations, were used to develop candidate benchmark concentrations for sentry well monitoring. The ACOE will develop benchmark concentrations in consultation with GaEPD. 1-2

4 Section 2 Model Refinement Model geometry was adjusted based on SHEP monitoring well boring logs. Model hydraulic parameters were adjusted so that simulated concentrations representing recent conditions are reasonably consistent with measured concentrations at the SHEP monitoring wells. These data were not available at the time of the previous model calibration. 2.1 Initial Data from the SHEP Groundwater Monitoring Program The groundwater monitoring network is documented by the ACOE in the Groundwater Monitoring Report, Savannah Harbor Expansion Project (ACOE, 2015). The network consists of eight Floridan aquifer monitoring wells at four sites, shown in Figure 1 together with simulated Floridan aquifer potentiometric surface contours. Two of the wells are existing wells, and six additional wells were installed in Two wells, MW-1 and MW-2, are located upgradient of the channel and are included for background monitoring. The other six wells are located downgradient of the channel dredging area for sentry monitoring. At all well locations there is one well screened just below the Miocene confining unit/floridan aquifer contact, at depths ranging from 119 to 135 feet, and a deeper Floridan aquifer monitoring well at depths ranging from 221 to 348 feet. Figure 1. Area Production Wells and SHEP Monitoring Wells 2-1

5 Section 2 Model Refinement Table 1 lists the monitoring well depths and the depths and elevations of the top and bottom of the Miocene confining unit encountered at each well location. Well Total Depth (feet BGS) Table 1. SHEP Monitoring Wells Top of Miocene Depth (feet BGS) Bottom of Miocene Depth (feet BGS) Top of Miocene Elevation (feet BGS) Bottom of Miocene Elevation (feet BGS) MW MW MW MW Not Available Not Available Not Available Not Available MW MW MW Not Available 120 Not Available MW The wells have been sampled quarterly since August 2012 for chloride and other salt ions (chloride, bromide, sulfate, sodium, magnesium, potassium, calcium) as well as total dissolved solids (TDS) and alkalinity. The monitoring report includes results through January The first four sampling events were conducted between August 2012 and June 2013, a period of time when wells were still equilibrating following installation; as such these results are not believed to be representative of existing Floridan aquifer quality at the well sites. Results from MW-2 are not considered valid due to casing leakage which allows surficial aquifer water to enter the well (ACOE, 2015). Measured chloride concentrations from October 2013 to January 2015 for each well except MW-2 are plotted in Figure 2. Typical chloride concentrations for the shallower wells ranged from 7 to 14 mg/l. Chloride concentrations for the deeper wells typically ranged from 5 to 11 mg/l. These data are assumed to represent baseline Floridan aquifer chloride concentrations under pre-dredging conditions, including the impact of downward migration of saltwater in the area of a cone of depression in the Floridan aquifer due to heavy groundwater pumping in the Savannah area and Hilton Head Island, South Carolina. 2.2 Model Layering Adjustment The elevation of the top of the Miocene confining unit in the model was adjusted based on the data shown in Table 1. Specifically, at the MW-5/6 location, the top of the confining unit was lowered from -48 to -79 feet mean sea level (MSL), and at the MW-7/8 location the top of confining unit was lowered from -50 to -71 feet MSL. The existing model elevations of the top of the confining unit at the other two monitoring sites were consistent with the sentry well boring data at these locations. Figure 3 shows the existing and revised model layering in cross section with the boring data at wells MW-1/2 and MW-5/6 superimposed. The existing model representations of the bottom of the confining unit were consistent with the boring data at all four sites, so no adjustment was made. 2-2

6 Section 2 Model Refinement Figure Sentry Well Chloride Concentrations Figure 3. Original and Revised Model Layering Near MW-1/2 and MW-5/6 2-3

7 Section 2 Model Refinement As a result of the model layering revisions, the thickness of the confining unit in the vicinity of wells MW-5/6 and MW-7/8 was reduced approximately 30 and 21 feet, respectively, in the model. This reduces the overall hydraulic resistance and travel distance between the surficial aquifer and the Floridan aquifer in the model, thereby increasing the rate of model-simulated transport of chloride to the Floridan aquifer in the vicinity of these locations. 2.3 Savannah Area Pumping Assignments For the projection simulations used to establish sentry well benchmark concentrations, constant pumping at year 2000 rates was assigned throughout, consistent with the previous (2005) projection simulations. For the historical simulations used for refining Miocene confining unit hydraulic properties as described below in Section 2.4, an adjustment to Savannah area pumping assignments was made. The previous SHEP model historical simulations represented the period from 1900 to For this project, the historical simulation period was extended through For the period from 2000 to 2015, year 2000 pumping rates were applied everywhere except in the Savannah area ( Red Zone ) where reported groundwater pumping has decreased since Savannah Red Zone pumping assigned in the model for year 2000 was approximately 68 million gallons per day (mgd). That pumping rate was maintained through By 2015, simulated Red Zone pumping was 50 mgd. 2.4 Miocene Confining Unit Hydraulic Properties The previous SHEP groundwater model s ability to reasonably reproduce both the historical temporal behavior of Floridan groundwater heads and the measured chloride levels from pore water samples in the Miocene confining unit below the Savannah River was demonstrated with a historical transient simulation starting with pre-development conditions (1900) and extending through the year 2000 (CDM, 2005). The simulation results were especially sensitive to the vertical hydraulic conductivity (Kv) assigned to the Miocene confining unit. Simulation results were presented for a range of Kv values considered to bracket the likely range of Miocene confining unit properties. As the previous model report noted (CDM, 2005), [t]he true system response lies somewhere in between the two simulations. At the higher end of the range, Kv = feet/day, the simulated Floridan aquifer heads agreed well with measured data, but the simulated rate of chloride penetration into the confining unit and the simulated confining unit concentrations were generally observed to exceed the observed data. Using the lower end of the range, Kv = feet/day, the rate of simulated chloride penetration tended to be reasonably consistent with observed data, though typically slightly less than the rate of penetration indicated by the pore water data. However, simulated Floridan aquifer water levels were approximately 25 feet lower than the measured data using the lower Miocene Kv value. The historical simulations were repeated using the revised layer model, and extended through Simulated heads and concentrations at the SHEP monitoring wells were compared with 2-4

8 Section 2 Model Refinement the field measured data for the purpose of refining the estimated range of Miocene confining unit hydraulic properties. Head Comparison While a formal model calibration process was not part of the scope of work for this project, a generalized comparison of measured and simulated heads at the sentry monitoring wells was developed. Table 2 shows the computed 2015 head at the sentry wells for different simulations compared with the range of measured heads over the August 2012 to January 2015 monitoring period. The model simulated 2015 heads are representative of the entire 2012 to 2015 time period because pumping and recharge are constant in the model over that period. Similar to the previous 2005 SHEP groundwater modeling, the simulated heads using the higher Miocene Kv = feet/day are reasonably consistent with the measured heads, though generally a few feet lower than observed. Using the lower bound Miocene Kv = feet/day, simulated heads are lower than measured heads by approximately 25 feet, also similar to the previous 2005 SHEP groundwater modeling. Table 2. Simulated and Measured Head at SHEP Monitoring Wells Observed Head (feet MSL) Aug-2012 Jan-2015 Well Minimum Maximum Simulated Head (feet MSL) Jan-2015 Miocene Kv= feet/day Miocene Kv= feet/day MW MW MW MW MW MW MW MW Concentration Comparison For the lower Miocene Kv simulation, there was no impact of downward chloride migration simulated at the sentry wells, consistent with the observed data. However, the historical simulation using the upper limit Kv of feet/day resulted in significant chloride concentrations by 2015 at a number of the sentry wells, which is inconsistent with the observed data. The simulated 2015 concentrations at the sentry wells were as high as 2,000 mg/l at MW- 6. To maintain reasonable consistency with the recent sentry well monitoring data, the Miocene confining unit effective porosity assignment was increased from 0.1 to 0.3 for simulations using the upper limit Kv value. Increasing the effective porosity assignment slows the velocity of simulated chloride migration through the Miocene confining unit. The value of 0.3 is within a reasonable range of effective porosity values for the confining unit. Figure 4 shows the simulated 2-5

9 Section 2 Model Refinement concentrations at the sentry wells using the upper limit Kv value and increased effective porosity. In this case, the highest simulated 2015 chloride concentration is approximately 21 mg/l which is much closer to the observed condition than the previous result using Kv = feet/day and an effective porosity of 0.1. Using the range of Miocene hydraulic properties described above (Kv from feet/day to feet/day and effective porosity from 0.1 to 0.3), the model projected chloride concentrations are considered representative of a reasonable range of expected concentrations. 2-6

10 Section 3 Model Application The refined SHEP model was used to develop projected future chloride concentration versus time plots at SHEP monitoring wells MW-3 to MW-8. Simulations were completed using both the upper limit and lower limit Miocene confining unit Kv values presented in Section 2. Simulations were also completed to represent conditions with harbor dredging and with no harbor dredging. The effect of dredging was represented in the model by lowering the channel bed elevation and increasing the chloride source concentrations assigned at the base of the dredged channel. The projection simulation period was 30 years. 3.1 Projection Simulation Parameters The predictive simulation parameters are generally similar to those assigned for the previous SHEP groundwater modeling, except that the predictive period begins in Initial groundwater heads: The simulated groundwater levels at the end of the historical simulation (2015) were used as the starting condition for the predictive simulations. Initial chloride concentrations: The simulated distributions of groundwater chloride concentrations at the end of the historical simulation (2015) were used as the starting condition for the predictive simulations. Predictive simulations using the upper limit confining unit Kv value were started with the heads and concentrations simulated at the end of the historical simulation using the upper limit confining unit Kv value. Similarly, initial conditions for predictive simulations using the lower limit Kv value were derived from the end condition of the historical simulation for the lower limit Kv value simulations. Groundwater pumping: Groundwater pumping was kept constant at year 2000 levels. Savannah River salinity: The Savannah River nodes, including the South Channel were assigned a constant chloride concentration to create a source of chloride in the model. No other chloride source was simulated. The chloride concentrations used for the dredging simulations (4,100 to 8,100 mg/l) are higher than the no-dredging scenario (2,600 to 7,400 mg/l) and were obtained from the surface water modeling conducted for the SHEP by Tetra Tech (Tetra Tech, November 2004). For the dredging scenario, the higher values were applied at the beginning of the projection simulation (year 2015). The simulated Savannah River source chloride concentrations were the same as the previous SHEP groundwater modeling (CDM, 2005). Miocene thickness and dredged depths: Data provided by the ACOE for the previous study were used to determine the change in elevation of the top of the Miocene confining unit as a result of the dredging. Dredging depths range from 55 to 59 feet below Mean Low Water 3-1

11 . Section 3 Model Application (MLW), or elevations range from -58 to -62 feet MSL. These dredging depths are the same as applied in the previous SHEP groundwater modeling (CDM, 2005) and are considered the maximum depths that could occur. Transport parameters: Table 3 shows the transport parameters utilized in the simulations. Parameter Longitudinal Dispersivity Transverse Dispersivity Floridan Aquifer Vertical Dispersion Anisotropy Effective Porosity Retardation Table 3. Transport Modeling Parameters 30 feet 3 feet 0.1 (dimensionless) 0.1 (dimensionless) Value 0.3 assigned to Miocene with upper limit Kv 1 = no retardation (dimensionless) Salt water density. The ratio of salt water density to fresh water density was varied linearly from 1.0 for zero chloride concentration to for a chloride concentration of 10,000 mg/l. 3.2 Model Projected Concentration Trend Lines at the Sentry Wells Figures 4 and 5 show projected chloride concentrations at the SHEP monitoring plan sentry wells, MW-3 through MW-8. The simulated projection period starts at the end of 2015 and extends to Background wells MW-1 and MW-2 are upgradient of the river; hence, the simulations indicated no chloride impact at these wells from the modeled Savannah River source, although these wells may be impacted from downward vertical migration of seawater from other shallow sources as a result of lower Floridan aquifer heads and the cone of depression in the Savannah and Hilton Head, SC vicinity. Figure 4 shows the projection simulation sentry well chloride concentrations for both the dredging and no-dredging conditions using the higher Kv value for the Miocene confining unit. There is very little difference between the dredging and no-dredging simulation results. Simulations indicate potential chloride impacts at wells MW-3, MW-6 and MW-7, the shallower wells in each sentry well cluster. The highest simulated concentrations are seen at well MW-6, reaching approximately 140 mg/l by MW-6 is a shallow monitoring well in the area where the model thickness of the Miocene confining unit was reduced based on the monitoring program boring data. No chloride impact was simulated at wells MW-4, MW-5, and MW-8, the deeper wells in each cluster; hence, the lines for these wells in Figure 4 are superimposed on each other at concentration zero. Figure 5 shows the projection simulation sentry well chloride concentrations using the lower Kv value for the Miocene confining unit. Again, results are shown for both dredging and no-dredging conditions. In this case, simulated chloride concentration impacts are minor at all of the sentry wells. 3-2

12 . Section 3 Model Application Figure 4. Simulated Chloride Concentrations at Sentry Wells - Higher K Miocene Scenario Figure 5. Simulated Chloride Concentrations at Sentry Wells - Lower K Miocene Scenario 3-3

13 . Section 3 Model Application Based on these simulation results, actual future concentrations at the sentry wells would be expected to be somewhere between the concentrations indicated for higher Kv Miocene representation and the concentrations indicated for the lower Kv Miocene representation. As such, the simulated concentrations for the higher Kv Miocene could reasonably form the basis for establishing sentry well monitoring benchmark concentrations. 3.3 Model Projected Concentrations at Selected Production Wells Projected concentrations were also computed for four selected Floridan aquifer production wells. The selected production wells, shown in Figure 1, include: International Paper Well #2 Savannah Main Well #11 Kemira Well ( ) Whitemarsh Island Well #28 These wells are relatively close to the Savannah River, and are distributed north-south over a roughly seven-mile reach of the river. The selected wells also exhibited some of the highest simulated future production well concentrations presented in the previous model report (CDM, 2005) for a 200-year simulation period. For the 30-year projection period specified for this study, simulated chloride concentrations at the productions wells were all zero using the revised model. 3-4

14 Section 4 Supporting the SHEP Groundwater Monitoring Program The objective of this study is to develop model-based and/or statistical chloride concentration limits that can be established as benchmarks to which future sentry well monitoring data will be compared. If a benchmark concentration has been exceeded at a sentry well, a determination must be made as to whether the exceedance is due to dredging or the known progression of salt water towards the pumping cone of depression in the Savannah area. A possible impact does not necessarily signify an actual impact: an evaluation or assessment is typically necessary to declare an actual impact. Proposed monitoring benchmarks based on model simulation results and sentry well data are presented below. For this study, CDM Smith also developed a Microsoft Excel workbook application to help track sentry well groundwater monitoring data over time. A description of this tool is presented in Appendix A. 4.1 Benchmark Limits for Sentry Wells with Simulated Chloride Impacts CDM Smith conducted simulations of chloride transport using both the upper limit and lower limit Miocene confining unit Kv values presented in Section 2. Simulation results with the higher Miocene Kv showed chloride impacts at some of the sentry wells. Simulated impacts to the sentry wells in the projection simulation with lower Miocene Kv were minimal or none. As such, model simulation results from the higher Miocene Kv simulation were used to develop benchmark limits for monitoring. In both the dredging and no-dredging scenarios, for four of the six wells the simulations predicted chloride increases (MW-3, MW-4, MW-6, MW-7); no chloride increases were predicted for the other two wells (MW-5, MW-8). For the four increasing chloride cases, three-parameter exponential model curves were fit to the predicted data: y = a exp(bx) + c where y = chloride concentration, x = date, and a, b and c are the exponential model parameters. The plots showing the best-fit model curves for the four wells (no dredging simulation) are provided in Figures 6 to

15 Section 4 Supporting the SHEP Groundwater Monitoring Program Figure 6. Best-fit Exponential Model for MW-3 Figure 7. Best-fit Exponential Model for MW-4 4-2

16 Section 4 Supporting the SHEP Groundwater Monitoring Program Figure 8. Best-fit Exponential Model for MW-6 Figure 9. Best-fit Exponential Model for MW-7 4-3

17 Section 4 Supporting the SHEP Groundwater Monitoring Program The best-fit graphs were adjusted to reflect measured well-specific baseline chloride concentrations measured at the sentry wells. The well-specific baseline chloride concentrations (referred to as Baseline Adjustment on the y-axes in the Figures 6-9) were set equal to the average baseline values for the currently available pre-dredging data. The average baseline values for the sentry and background wells are shown in Table 4 below. Table Average Baseline and 95% Upper Prediction Limit (UPL) Chloride Concentrations Monitoring Well Average Baseline (mg/l) Baseline 95% UPL (mg/l) MW MW MW MW MW MW MW MW Well to be replaced; data are not to be used. The best-fit curves presented in Figures 6 9 can be used as benchmarks for future monitoring at sentry wells MW-3, MW-4, MW-6, and MW Benchmark Limits for Sentry Wells MW-5 and MW-8 For sentry wells MW-5 and MW-8, where model simulation results suggest no change in chloride concentrations from a Savannah River source over the 30-year simulation period, benchmark limits for monitoring can be developed from the baseline chloride concentrations at these wells. For these wells, an upper prediction limit (UPL) was calculated using the baseline data, such that there is 95-percent confidence that the subsequent measurement will be below the limit. The UPL baseline chloride concentrations for well MW-5 and MW-8 are 5.9 mg/l and 4.5 mg/l, respectively. Calculation of these values, which can be completed with statistical analysis workbook tool developed by CDM Smith, is presented in Appendix A. 4-4

18 Section 5 Summary CDM Smith refined and applied the SHEP groundwater flow model to project future chloride concentrations at sentry monitoring wells and selected production wells. Model simulation results were used to propose sentry monitoring well benchmark concentrations for future monitoring to evaluate the potential impacts to the Floridan aquifer by the proposed dredging of the Savannah Harbor channel. The following benchmark concentrations are proposed for consideration to track potential chloride impacts to the Floridan aquifer from Savannah Harbor channel dredging: 1. Use the Best-Fit Model Benchmark line (Figures 6, 7, 8 and 9) to represent the benchmark limit in cases where simulations predict increasing concentrations from the modeled Savannah River source for the well. This benchmark will change with time, based on the model-projected values. This recommendation applies to wells MW-3, MW-4, MW-6, and MW Use the 95-percent UPLs calculated for baseline data to represent the benchmark limit in cases where simulations suggest no change in chloride concentrations from a Savannah River source over the 30-year simulation period. This recommendation applies to wells MW-5 and MW-8. Per the SHEP Groundwater Monitoring Plan, if a benchmark concentration is exceeded at a sentry well, both wells in the well cluster will be resampled to confirm the benchmark exceedance. If the exceedance is confirmed, further evaluation will be completed to determine whether the exceedance is associated with dredging operations or the known progression of saltwater towards the pumping cone of depression in the Savannah area. Evaluation may include chloride trend analysis at sentry and other monitoring wells in the area and groundwater model simulation analysis using the SHEP groundwater model developed for the study. 5-1

19 Section 6 References CDM, Savannah Harbor Expansion Three-Dimensional Salt Water Intrusion Modeling. Fanning, J. L. and V.P. Trent, Water Use in Georgia by County for 2005; and Water-Use Trends, United States Geological Survey Scientific Investigations Report Gibbons, R.D., Bhaumik, D.K and S. Aryal, Statistical Methods for Groundwater Monitoring, Second Edition, John Wiley & Sons. Lawrence, S.J., Water Use in Georgia by County for 2010 and Water Use Trends, United States Geological Survey Open File Report US Army Corps of Engineers, Groundwater Monitoring Report Savannah Harbor Expansion Project. April US Army Corps of Engineers, Groundwater Monitoring Plan Savannah Harbor Expansion Project. September USEPA, Statistical Analysis of Groundwater Monitoring Data at RCRA Facilities, Unified Guidance, EPA 530/R March

20 Appendix A SHEP Groundwater Monitoring Program Statistical Analysis Workbook Tool

21 Appendix A Workbook Tool The workbook tool contains numerous features designed to assist the SHEP groundwater monitoring program. The tool: (1) stores and maintains all groundwater data as it is collected over the course of the monitoring program; (2) performs the benchmark calculations pertaining to the monitoring program (as described in Section 4); (3) displays time-series plots showing the data as it is collected; (4) performs various statistical calculations useful for detection monitoring and assessment purposes; and (5) contains a tabulation feature for summary and reporting purposes. Detailed instructions for operating the tool are provided in this appendix. Instructions are also provided on the Start sheet of the workbook tool. Opening the Workbook When the workbook is opened the Start sheet is displayed with the title: The Start sheet contains instructions for operating the tool. Opening the workbook also creates three custom toolbar controls on the Add-Ins tab of the ribbon: The left-most control is a dropdown control for selecting the monitoring well. The middle control is a dropdown control for selecting the analyte; since chloride is the only analyte currently included in the monitoring program, only Chloride appears on the dropdown control. The rightmost control is a dropdown control for selecting display and calculation options and tabulation features. Clicking Select and then Options opens a dialog box where various display and calculation options are selected. The dialog box is titled Select Options and contains two tabs, Display and Calculation : A-1

22 Appendix A Workbook Tool Display Options The Display tab on the Select Options dialog box controls what will be displayed on the timeseries plot. On the Display tab are three grouped controls, Standard, Type of Interval Limit and Basis of Interval Limit. The Standard group contains two checkbox controls: The upper checkbox labeled Simulation Limit refers to the best-fit curves for the simulations as described in Section 4 of Updated Saltwater Intrusion Modeling to Support the Groundwater Monitoring Plan (Figures 6, 7, 8 and 9). These are the best-fit curves for the four sentry wells (MW-3, MW-4, MW-6 and MW-7) that have predicted cone of depression related increasing chloride concentrations. The model simulated concentrations are the basis for the benchmark limits at these wells. When this checkbox is checked, the best-fit lines are calculated for the current data and displayed on the time-series plot. The lower checkbox labeled GWPS (groundwater protection standard) refers to a particular standard that may be entered into the workbook, and which may be applicable for additional assessment purposes. Entering a particular standard (and displaying it) is entirely optional. When this checkbox is checked, the GWPS line (if entered) will be displayed on the time-series plot. The Type of Interval Limit group on the Display tab contains two option controls, Prediction and Tolerance": Both of these controls refer to the upper limit calculated on background data (either baseline or upgradient, as discuss further below). For detection monitoring purposes, the limit is specifically applicable to the two sentry wells (MW-5 and MW-8) that exhibited no simulated cone of A-2

23 Appendix A Workbook Tool depression related impact.. However, the upper limits are calculated and displayed for all wells for possible additional assessment purposes (if applicable). Selecting Prediction (the default) results in calculation of a 95% confidence upper prediction limit for the next observation (95% UPL; 95% confidence that the next background observation will fall below the limit). Selecting Tolerance results in calculation of a 95% confidence, 95% coverage upper tolerance limit (95% UTL; 95% of all possible background data are expected to fall below this limit with 95% confidence). Per Section 4 and 5, the recommended option for the SHEP monitoring program is the Prediction option. For details on calculation of tolerance and prediction limits, see Gibbons et al (2009) and USEPA (2009). Note that in the above discussion the term background refers to either baseline in the case of calculating an upper limit for comparison of baseline data (pre-dredging) to detection data (post-dredging) for an individual well, or to upgradient in the case of calculating an upper limit for comparison of detection data (post-dredging) to upgradient data (no dredging impact). In the case of the SHEP detection monitoring program, baseline comparison is the applicable option (as discussed in Section 4). However, for certain additional assessment purposes, comparison to upgradient data may be useful. Thus, in order to have this option in the workbook tool, the Basis of Interval Limit group is provided on the Display tab: In the case of selecting Upgradient Data, the upgradient and downgradient data are set via incell dropdown controls located on the Locations sheet of the workbook: A-3

24 Appendix A Workbook Tool Calculation Options The Calculation tab on the Select Options dialog box controls the statistical calculations for detection and assessment purposes. On the Calculation tab are two grouped controls, Limit Type and Detection Date for Assessment. The Limit Type group contains one textbox control: The value entered into the textbox is the normality test criterion (alpha) value used to determine whether a parametric limit is calculated. The default value is Normality testing is conducted using the Shapiro-Wilk (SW) method (see USEPA 2009 for further details on this method). If the original (non-transformed) background data (baseline or upgradient depending on the selection) pass the normality test (p-value > alpha) then parametric limits are calculated. If the original data do not pass the normality test but the log-transformed data pass the normality test then parametric limits are calculated based on the log-transformed data (indicated with "LN"). If neither case passes the normality test then non-parametric limits are calculated. The Detection Date for Assessment group on the Calculation tab contains two option controls: Select the Most Recent Event option (the default) to select the most recent Detection value (regardless of date) for detection or assessment purposes. Select the Only Events Post Cutoff Date option and enter the cutoff data to select the most recent Detection value post an entered cutoff date. This option may be useful in cases where, for example, data were not collected for a particular well during a particular event. Baseline versus Detection Data Baseline versus Detection data are distinguished in the tool by entering either Baseline or Detection under the Use_Flag field on the Data sheet: A-4

25 Appendix A Workbook Tool The Data sheet is where the monitoring data are stored. Only records with either "Baseline" or "Detection" entered into a Use_Flag cell are used in the analysis. The Baseline Rules are a reminder that "Baseline" should correspond to pre-dredging data and "Detection" to postdredging data. However, the setting of baseline and detection data is purposely designed to be flexible, as it may be desirable to periodically adjusted baseline if it is deemed appropriate to do so. Furthermore, leaving a Use_Flag cell blank will result in ignoring the record for statistical calculation purposes. For example, an outlier may be ignored by leaving a Use_Flag cell blank. The tool contains a feature for testing for statistical outliers in the baseline and upgradient data using the Grubbs-Rosner (GR) method (see USEPA 2009 for further details on this method). Also, it may be desirable to exclude the first value of a two-value resample by leaving the value blank. Trend Analysis The workbook tool also provides trend analysis for detection data using the Mann-Kendall (MK) method (see USEPA 2009 for further details on this method). Trend analysis is an additional feature that may be useful for assessment purposes. This analysis, along with all detection and assessment results, is reported on the Analysis sheet. Note that there is no automatic consequence of the trend analysis; the results are simply provided as an additional feature to assist with assessment should it be deemed useful or necessary. Analysis Sheet The workbook tool contains several individual worksheets. The Analysis sheet provides all detection and assessment results for the current selection, including the time-series plot. When a selection is made from the Add-Ins controls, the Analysis sheet is automatically displayed. Results on the Analysis sheet are displayed in sections that can be collapsed or expanded by toggling, i.e., clicking, the plus or minus symbols: Collapsed Expanded A-5

26 Appendix A Workbook Tool Tabulation The workbook tool contains a tabulation feature. From the Select dropdown control select the Tabulation item to open the Select for Tabulation dialog box: Drag or hold down the Ctrl key to make multiple selections. Click the OK button to initiate tabulation. The tabulation results are displayed on the Table sheet. Time-Series Plots and Benchmarks Time-series plots on the Analysis sheet provide a graphical display of the monitoring data along with the various limit options as selected. As described in Section 4, the recommended benchmarks for the SHEP are either the simulation limit, adjusted based on average baseline concentrations or the upper prediction limit (UPL) for baseline data, depending on the monitoring well. The following is an example of a time-series plot for a case where the simulation limit is the benchmark (MW-06): A-6

27 Appendix A Workbook Tool Note that in this case, although both limits are displayed, the simulation limit is the applicable benchmark. The UPL is shown only for possible assessment purposes. Note also that only baseline data (blue squares) are plotted since this is the only data currently available; as detection data (post-dredging) are collected, they will appear on the time-series plot as detection data (black diamonds) to the right of the baseline data, and the limits will be extended to encompass the detection data. The following is an example of a time-series plot for a case where the upper prediction limit is the benchmark (MW-05): A-7

28 Appendix A Workbook Tool Note that in this case, only the UPL is displayed because the simulation limit is not applicable, i.e., the simulation did not predict a cone of depression related impact for this well. Thus, in this case the UPL determined for the baseline data is the applicable benchmark. A-8