Ft. Drum Wells Field Final D-B RFP - Appendix C - Volume 2.pdf

Appendix C Vol. 1

Groundwater Model Update Report

U.S. Army Corps of Engineers – New York District

Groundwater Model Update

Report ERCIP Well Field Expansion Project

Wheeler-Sack Army Airfield Bedrock Potable Water

Wellfield – Area 4

United States Army Garrison, Fort Drum, New York

Contract: W912DS20D0002

December 2025

This document is intended only for the use of the individual or entity for which it was prepared and may contain information that is privileged, confidential and exempt from disclosure under applicable law. Any dissemination, distribution or copying of this document is strictly prohibited.

www.arcadis.com Final-GWModelReport_10222025

Groundwater Model Update Report ERCIP Well Field Expansion Project

Wheeler-Sack Army Airfield Bedrock Potable Water Wellfield – Area 4

United States Army Garrison, Fort Drum, New York

Contract: W912DS20D0002

December 2025

Prepared By: Prepared For:

Arcadis U.S., Inc. U.S. Army Corps of Engineers – New York District

One Lincoln Center, 110 West Fayette Street, Suite 300

Syracuse, NY 13202

United States

Phone: 315 446 9120

Fax: 315 449 0017

Our Ref:

30244799

Andrew R. Vitolins, PG

Vice President

Jennifer Wahlberg

Senior Environmental Specialist www.arcadis.com Final-GWModelReport_10222025

Version Control Issue Revision No. Date Issued Description

1 0 9/19/2025 Draft Report

2 1 10/22/2025 Final Report

Final-GWModelReport_10222025 i

Contents Acronyms and Abbreviations ................................................................................................................................. v

1 Introduction

2 History of Groundwater Model Development

2.1 2011 Groundwater Flow Model

2.2 2020 Groundwater Flow Model Update

2.3 2024 Groundwater Flow Model Update

3 Conceptual Site Model

3.1 Geology

3.2 Hydrogeology

3.2.1 Assessment of Hydraulic Conductivity in Area 4

3.3 Groundwater Recharge

3.4 Conceptual Site Model Summary

4 Groundwater Flow Model Update

4.1 Boundary Conditions

4.1.1 Constant Head Cells

4.1.2 No Flow Boundaries

4.1.3 Recharge

4.1.4 Production Wells

4.2 Hydraulic Conductivity

5 Groundwater Flow Model Calibration

5.1 Steady-State Calibration

5.2 Sensitivity Analysis

5.3 Transient Calibration

5.3.1 Transient Model Set-Up

5.3.2 Transient Model Results

6 Production Scenarios

6.1 Scenario 1 – Area 4 Estimated Sustainable Rates

6.1.1 Steady-State – Drought Conditions

6.1.2 Transient

6.2 Scenario 2 – Area 4 Simulated Pumping Rates with Existing Well Fields at Current Rates

6.2.1 Steady State

Steady State – Drought Conditions

Final-GWModelReport_10222025 ii

6.2.2 Transient

6.3 Scenario 3 – Area 4 Simulated Pumping Rates with Existing Well Fields at Permitted Rates

6.3.1 Steady State

Steady State – Drought Conditions

6.3.2 Transient

7 Summary and Conclusions

8 Uncertainties and Limitations

9 References

Tables

Table 4-1 Model Layer Description (in text)

Table 4.2-1 Overall Model Hydraulic Conductivity Summary

Table 4.2-2 Area 4 Hydraulic Conductivity Summary

Table 5.1-1 Steady-State Water Level Targets and Calibration Results

Table 5.2-1 Summary of Groundwater Flow Model Sensitivity Analysis Results

Table 5.3-1 Transient Model Set-Up

Table 6-1 Production Scenario Well Rates (in text)

Table 6.1-1 Simulated Drawdown at Area 4 Wells and Potential Off-Site Potable Well Locations

Figures Figure 1-1 Site Location

Figure 2-1 Extents of Previous and Current Groundwater Models

Figure 4-1 Cross-Section A-A’

Figure 4-2 Cross-Section B-B’

Figure 4-3 Cross-Section C-C’

Figure 4.1-1 Boundary Conditions

Figure 4.2-1 Hydraulic Conductivity Distribution Model Layers 1 and 2

Figure 4.2-2 Hydraulic Conductivity Distribution Model Layers 3 and 4

Figure 4.2-3 Hydraulic Conductivity Distribution Model Layers 5 and 6

Figure 4.2-4 Hydraulic Conductivity Distribution Model Layers 7 and 8

Figure 4.2-5 Hydraulic Conductivity Distribution Model Layers 9 and 10

Figure 4.2-6 Hydraulic Conductivity Distribution Model Layers 11 and 12

Final-GWModelReport_10222025 iii

Figure 5.1-1 Steady-State Calibration Scatter Plot – Overall Model

Figure 5.1-2 Steady-State Calibration Scatter Plot – Bedrock Only

Figure 5.1-3 Steady-State Calibration Scatter Plot – Area 4 Only

Figure 5.1-4 Steady-State Simulated Groundwater Elevation and Weighted Residuals Model Layers 1 and 2

Figure 5.1-5 Steady-State Simulated Groundwater Elevation and Weighted Residuals Model Layers 3 and 4

Figure 5.1-6 Steady-State Simulated Groundwater Elevation and Weighted Residuals Model Layers 5 and 6

Figure 5.1-7 Steady-State Simulated Groundwater Elevation and Weighted Residuals Model Layers 7 and 8

Figure 5.1-8 Steady-State Simulated Groundwater Elevation and Weighted Residuals Model Layers 9 and 10

Figure 5.1-9 Steady-State Simulated Groundwater Elevation and Weighted Residuals Model Layers 11 and 12

Figure 5.3-1 Storage Distribution

Figure 5.3-2 Transient Calibration Simulated Drawdown over Time – Set 1

Figure 5.3-3 Transient Calibration Simulated Drawdown over Time – Set 2

Figure 5.3-4 Transient Calibration Simulated Drawdown over Time – Set 3

Figure 5.3-5 Transient Calibration Simulated Drawdown over Time – Set 4

Figure 6.1-1 Scenario 1 – Area 4 Estimated Sustainable Rates: Simulated Drawdown and Steady-State

Area of Flow Contribution in Limestone/Sandstone (Model Layer 10)

Figure 6.1-2 Scenario 1 – Area 4 Estimated Sustainable Rates: Simulated Water Table Drawdown

(Model Layer 1)

Figure 6.1-3 Scenario 1 – Area 4 Estimated Sustainable Rates: Simulated Drawdown and Steady-State

Area of Flow Contribution in Limestone/Sandstone (Model Layer 10) Under Drought

Conditions

Figure 6.1-4 Scenario 1 – Area 4 Estimated Sustainable Rates: Simulated Water Table Drawdown

(Model Layer 1) Under Drought Conditions

Figure 6.1-5 Scenario 1 – Area 4 Estimated Sustainable Rates: Forward Particle Tracking at Top of

Bedrock (Model Layer 9) Under Transient Conditions

Figure 6.2-1 Scenario 2 – Area 4 Simulated Pumping Rates with Existing Well Fields at Current Rates:

Simulated Drawdown and Steady-State Area of Flow Contribution in Limestone/Sandstone

(Model Layer 10)

Figure 6.2-2 Scenario 2 – Area 4 Simulated Pumping Rates with Existing Well Fields at Current Rates:

Simulated Water Table Drawdown (Model Layer 1)

Figure 6.2-3 Scenario 2 – Area 4 Simulated Pumping Rates with Existing Well Fields at Current Rates:

Simulated Drawdown and Steady-State Area of Flow Contribution in Limestone/Sandstone

(Model Layer 10) Under Drought Conditions

Final-GWModelReport_10222025 iv

Figure 6.2-4 Scenario 2 – Area 4 Simulated Pumping Rates with Existing Well Fields at Current Rates:

Simulated Water Table Drawdown (Model Layer 1) Under Drought Conditions

Figure 6.2-5 Scenario 2 – Area 4 Simulated Pumping Rates with Existing Well Fields at Current Rates:

Forward Particle Tracking at Top of Bedrock (Model Layer 9) Under Transient Conditions

Figure 6.3-1 Scenario 3 – Area 4 Simulated Pumping Rates with Existing Well Fields at Permitted Rates:

Simulated Drawdown and Steady-State Area of Flow Contribution in Limestone/Sandstone

(Model Layer 10)

Figure 6.3-2 Scenario 3 – Area 4 Simulated Pumping Rates with Existing Well Fields at Permitted Rates:

Simulated Water Table Drawdown (Model Layer 1)

Figure 6.3-3 Scenario 3 – Area 4 Simulated Pumping Rates with Existing Well Fields at Permitted Rates:

Simulated Drawdown and Steady-State Area of Flow Contribution in Limestone/Sandstone

(Model Layer 10) Under Drought Conditions

Figure 6.3-4 Scenario 3 – Area 4 Simulated Pumping Rates with Existing Well Fields at Permitted Rates:

Simulated Water Table Drawdown (Model Layer 1) Under Drought Conditions

Figure 6.3-5 Scenario 3 – Area 4 Simulated Pumping Rates with Existing Well Fields at Permitted Rates:

Forward Particle Tracking at Top of Bedrock (Model Layer 9) Under Transient Conditions

Appendices

Appendix A Sensitivity Analysis

Final-GWModelReport_10222025 v

Acronyms and Abbreviations % percent

Arcadis Arcadis U.S., Inc.

ASTM American Standard Test Method

CSM conceptual site model

Fort Drum United States Army Garrison in Fort Drum, New York ft2 square feet gpm gallons per minute in/yr inches per year

MODALL MODular flow ALLocation

MGD million gallons per day

NOAA National Oceanic and Atmospheric Administration

RSS residual sum-of-squares

2011 groundwater model groundwater flow model developed by the USACE Philadelphia District in 2011

2020 groundwater model groundwater flow model developed by the USACE Philadelphia District in 2020

2024 groundwater model groundwater flow model developed by Arcadis in 2024

USACE United States Army Corps of Engineers, New York District

WSAAF Wheeler-Sack Army Airfield

Final-GWModelReport_10222025 1

1 Introduction Arcadis U.S., Inc. (Arcadis) has prepared this Groundwater Model Update Report to support the design of a new potable water supply source at the United States Army Garrison in Fort Drum, New York (Fort Drum). This work is being conducted on behalf of the United States Army Corps of Engineers, New York District (USACE) under

Contract W912DS20D0002 and the associated USACE Statement of Work dated July 23, 2024. The Statement of

Work and associated draft wellfield design state that a new well field within the area, designated as Area 4, would include five new bedrock groundwater supply wells located northwest of the Wheeler-Sack Army Airfield (WSAAF) and the equipment and infrastructure to connect the wells to the Fort Drum potable water supply (Figure 1-1). The location of the test wells and the observation wells in Area 4 are shown on Figure 1-1.

This report summarizes the updates made to the groundwater flow model over time, the conceptual site model

(CSM), and updates made to the existing groundwater flow model developed in 2024 (Arcadis 2024). This model report also summarizes pumping scenarios developed for five proposed new groundwater supply wells to further evaluate if these wells will be sufficient to address potable water supply needs while not affecting existing water supply wells or receiving impacts from known sources of contamination at Fort Drum. The objective of the modeling study was to evaluate if five new water supply wells installed within Area 4 can produce the minimum desired production rate of 500 gallons per minute (gpm), or 0.72 million gallons per day (MGD).

Final-GWModelReport_10222025 2

2 History of Groundwater Model Development Three groundwater models, identified below, have been developed to assess groundwater conditions at Fort

Drum. The extents of these models are presented on Figure 2-1.

2.1 2011 Groundwater Flow Model

The groundwater flow model for Fort Drum was first developed in 2011 by the USACE (USACE 2011), referred to hereafter as the 2011 groundwater model. The 2011 groundwater model was a watershed level model and

WASH123D was used for the analyses. The objective of the 2011 groundwater model was to evaluate groundwater flow paths based on proposed pumping operations of the well field consisting of Wells 13, 14, 15, 16, 17, and 18, located proximate to the WSAAF, as well as the potential impacts to the Philadelphia Reservoir.

2.2 2020 Groundwater Flow Model Update

The 2011 groundwater model was updated in 2020 by the USACE (USACE 2020), referred to hereafter as the

2020 groundwater model. Between 2011 and 2020, the groundwater model was used to evaluate several production scenarios that were more reflective of anticipated well field operations (USACE 2018). The objective of the 2020 groundwater model was to assess the potential of Areas 3 and 4 to supply clean drinking water for Fort

Drum. Figure 2-1 shows the locations of Areas 3 and 4. Between 2017 and 2018, these areas were identified as areas for further exploration for water supply and field work, including well installation and pumping tests, was conducted. Similar to the 2011 groundwater model, WASH123D was used for the analyses. The 2020 groundwater model expanded the model domain to the north to minimize the possibility of boundary effects when simulating pumping in Area 4. The 3D mesh was adjusted to refine the model in Areas 3 and 4. The geologic framework was updated to account for additional boring data. Various pumping scenarios were evaluated under both steady-state and transient conditions. Based on the results of these analyses, personnel at Fort Drum decided to pursue the installation of production wells in Area 4.

2.3 2024 Groundwater Flow Model Update

The 2020 groundwater model was updated in 2024 by Arcadis (Arcadis 2024), referred to hereafter as the 2024 groundwater model, to further evaluate groundwater production potential in Area 4, based on the results of the

2020 assessment. The 2020 groundwater model was transformed to a MODFOW-USG (Panday et al. 2013) model and further updated using recently collected data. MODFLOW-USG was selected because it uses unstructured grids that can use similar properties as the original 2020 groundwater model, and the MODFLOW family of codes are the standard software in the industry used to simulate groundwater flow.

The objective of the 2024 groundwater modeling study was to evaluate if five new water supply wells installed within Area 4 can produce the minimum desired production rate of 500 gpm, or 0.72 MGD. The 2024 groundwater model was updated and calibrated based on initial well testing data from 2024 production tests in Area 4 (a 72-hour wellfield pumping test), using both quantitative (e.g., calibration statistics) and qualitative (e.g., groundwater flow directions) methods for both steady-state and transient conditions. The transient model was further validated using drawdown data collected during the 72-hour pumping test for Area 4, the pumping test for Well 17, and the pumping test for Area 3. The resultant calibrated model was used to evaluate four different production scenarios:

Final-GWModelReport_10222025 3

Scenario 1 – Area 4 Estimated Sustainable Rates;

Scenario 2 – Area 4 Simulated Pumping (Wells 3 and 4 moved);

Scenario 3 – Area 4 Simulated Pumping Rates with Existing Well Field at Current Rates; and

Scenario 4 – Area 4 Simulated Pumping Rates with Existing Well Fields at Permitted Rates.

All scenarios were evaluated under steady-state conditions with the same parameters (e.g., recharge and hydraulic conductivity) as the calibrated model. Scenarios 2, 3, and 4 were also evaluated under drought conditions (25 percent [%] reduced recharge) to assess the increased drawdown and the increased area of flow contribution for the pumping wells. Scenarios 3 and 4 were also evaluated under transient conditions to evaluate the potential for contaminant migration from the WSAAF.

Final-GWModelReport_10222025 4

3 Conceptual Site Model The CSM presented herein is a written and/or illustrative representation of the physical processes in operation at

Fort Drum. The goal of a CSM is to provide a description of relevant site features and characteristics that represent the site processes that result in the observed site conditions. The CSM provides a framework of conditions to construct a mathematical model (i.e., a computer-generated representation of the actual flow system). The CSM is an iterative tool that may be further developed and refined as information is obtained throughout the site investigation.

3.1 Geology

Unconsolidated deposits from the Pleistocene era within the model domain represent glacial kame, lacustrine delta, lacustrine shoreline, and ice contact deposition (Reynolds 1986). Unconsolidated deposits within Area 4 are predominantly recent silty clays with scattered clay dominated lenses while, in some areas, a sandy layer between one and ten feet in thickness was observed. These deposits vary between 16 and 60 feet in thickness throughout Area 4 and increase in thickness toward the south and east as the deposits transition to the deltaic deposits of the Pine Plains aquifer underlying the WSAAF. The WSAAF is situated atop the Pleistocene Pine

Plains deltaic deposit (Reynolds 1986), which is generally comprised of silty fine sand, with silt and clay content increasing with depth throughout the area. Geologic cross-sections (Reynolds 1986) indicate bedrock underlying the WSAAF was truncated by a glacial ice lobe, producing a buried valley that increased the thickness of

Pleistocene deposits in the ice lobe’s former channel. Based on geologic cross-sections provided in Hydrogeology of the Fort Drum Area (Reynolds 1986), the thickness of the unconsolidated deposits ranges from approximately

100 to 220 feet. Generally, the thickest of these deposits occur in the vicinity of the WSAAF. The Pine Plains unconsolidated deposits underlying the WSAAF consist of deltaic silty sands, lacustrine bottom clays and silts, and clayey gravel, with a thin layer of basal till at the bedrock contact. These deposits thin considerably towards the north and west of the WSAAF and toward Area 4 as the deltaic complex ends and are replaced by thinner lacustrine and more recent floodplain deposits consisting of sandy silts and clays overlying bedrock.

Bedrock in the region consists of a complex of dolomitic limestones and sandstone overlying a basement metamorphic unit consisting of mylonitic gneiss. Cambrian-aged dolomitic and calcitic sandstones (e.g., Potsdam and Theresa Sandstones and Theresa Dolostones) and Ordovician-aged limestones of the Black River Group

(e.g., Pamelia, Lowville, and Chaumont Limestones) are underlain by the Potsdam Sandstone (Reynolds 1986).

Dolomite concentrations are higher in the lower Black River Group than the upper units, such that the Pamelia is defined as a dolomitic limestone (Johnsen 1971). The Chaumont, however, is a pure limestone within which most of the karstic features in the region are expressed. However, published geologic information for this area shows that the Chaumont is not present within Area 4. The bedrock surface varies in elevation within the study area, between 480 and 570 feet above mean sea level, and increases in elevation toward the east of Area 4 while maintaining a somewhat level trend north to south. The contacts between the three limestones appear relatively flat under the airfield; however, as presented in Hydrogeology of the Fort Drum Area (Reynolds 1986), these contacts dip westerly towards the airfield, in the vicinity of the Black River. Boring data from the test and observation wells in Area 4 similarly indicate that the upper formations generally dip toward the west to the west-northwest, as does the contact between the sedimentary units and the underlying basement rock.

Based on regional and local geologic cross-sections and bedrock mapping (Reynolds 1986; Johnsen 1971), the approximate thickness of Cambrian sandstones and Ordovician limestones near the airfield is 200 and 150 feet, Final-GWModelReport_10222025 5 respectively. However, the limestone units thin north and northwest of the WSAAF toward Area 4 and ultimately pinch out in this direction to the north and west of Fort Drum (Reynolds 1986; Johnsen 1971). The dolomitic limestone units thin northwesterly toward Area 4 to thicknesses of less than 50 feet, and the Potsdam Sandstone unit similarly thins northwesterly, varying in thicknesses of up to 40 feet within Area 4. Based on geologic cross-sections in Hydrogeology of the Fort Drum Area (Reynolds 1986), the Potsdam Sandstone also pinches out locally towards the east of Fort Drum and is not present south and east of the Black River. Basement rock in the study area consists of a generally massive mylonitic gneiss of undefined thickness, with no significant water bearing fractures within the study area.

3.2 Hydrogeology

A drainage divide extends from west to east through the Pine Plains aquifer within the model domain. Surface drainage within Area 4 and northern portions of the Pine Plains aquifer is toward the north into the Indian River watershed and north to the St. Lawerence River drainage. Southern portions of the Pine Plains aquifer along the

Black River discharge to the Black River and into the Lake Ontario watershed.

Groundwater flow occurs within both overburden materials (sand, silty sand, and clay/silt) and bedrock. Although complex flow patterns are known to exist within the overburden, particularly around the OASIS site, the model includes a simplified representation of the overburden geology that simulates average flow conditions and flux to the bedrock. The bedrock aquifer is the principal water-bearing unit at Fort Drum, with the primary water-bearing units being fracture-dominant portions of the upper dolomitic limestones/sandstones of the Pamelia and Theresa formations and the underlying Potsdam Sandstones. The mylonitic gneiss basement rock underlying the region is not a source of water within the model domain. Generally, the bedrock aquifer is thinnest near the northwestern end of Fort Drum and thickens closer to the Black River. The carbonate and sandstone units also thin and pinch out to the northeast toward the Black Creek drainage (Reynolds 1986).

The groundwater system is fracture-flow dominant, with differences in pumping rates in wells within the bedrock aquifer suggesting that the availability of groundwater from bedrock is dependent on the size and number of fractures intersected by wells. Water-bearing fracture zones were present within the uppermost limestone units and the sandstone unit at the top 10 to 20 feet of the bedrock. Within some areas, especially in the vicinity of Test

Wells 3 and 4, these fracture zones were heavily weathered, resulting in some calving and deterioration of the borehole walls during drilling. Other areas of Fort Drum with thicker limestone units and more extensive fracture zones can produce significant water from the dolomitic limestone/sandstone units (such as the existing well field shown on Figure 1-1). Groundwater within the bedrock generally flows from southeast to northwest; however, there are some local differences, particularly in the area where the bedrock is thickest close to the Black River.

3.2.1 Assessment of Hydraulic Conductivity in Area 4

The drawdown data obtained from the 72-hour pumping test conducted at Area 4, described in the Updated Field

Investigation Summary Report (Arcadis 2025), was analyzed using AQTESOLV (Duffield 2007) to determine the transmissivity of the bedrock materials within Area 4. The results of the pumping test analysis and resultant

AQTESOLV estimates for aquifer parameters are summarized in the Updated Field Investigation Summary

Report. AQTESOLV has numerous options for solutions that can be used to estimate transmissivity results from the data. In this case, the Hantush Leaky Confined Solution (Hantush 1960, 1961a/b, 1964) was used. The results of the analysis for the bedrock wells are presented in the Updated Field Investigation Summary Report, along with a discussion of the rationale for the applicability of the selected analytical solution. The pumping test

Final-GWModelReport_10222025 6 analyses showed that hydraulic conductivity within the bedrock in Area 4 varied from approximately 20 to 200 feet per day. It should be noted that those hydraulic conductivities represent the bulk hydraulic conductivity for the entire bedrock and do not reflect localized higher hydraulic conductivity areas within the bedrock that may have a more extensive fracture network.

3.3 Groundwater Recharge

The primary recharge mechanism within Fort Drum is precipitation (natural recharge). Natural recharge is variable from year to year, changing with annual precipitation. Over the last 20 years (2005 to 2024), annual precipitation has ranged from 32.80 inches per year (in/yr) to 63.08 in/yr, with an average of 46.64 in/yr (National Oceanic and

Atmospheric Administration [NOAA] 2025). The annual precipitation for the past 2 years (2023 and 2024) was

48.13 and 54.39 in/yr, respectively, indicating recent annual precipitation was similar to the 20-year average.

Recharge from precipitation typically ranges from 5 to 25% of annual precipitation. The spatial distribution of recharge is variable, and several factors influence the rate and spatial distribution of recharge, including:

Recharge may be significantly reduced in winters, after the ground has frozen and precipitation occurs as snow. Snowmelt and early spring rains may restrict the ability to infiltrate and preferentially drain as surface runoff.

The developed areas of Fort Drum are drained by a storm sewer designed to intercept and remove runoff that might otherwise infiltrate to groundwater.

The airfield sanitary landfill has an installed cap designed to reduce recharge to the underlying groundwater system.

3.4 Conceptual Site Model Summary

Based on a review of published and subsurface geologic and hydrogeologic data, the aquifer system underlying the WSAAF and vicinity consists of an upper unconfined aquifer overlying a bedrock aquifer that is partially confined within the study area. The upper unconfined aquifer is comprised of glacially derived soils and extends throughout the WSAAF and southern portions of the Cantonment. Within the soils of the upper water-bearing unit are units of silty fine sand, silty clay, and a basal boulder till. The silty clay units are assumed to be laterally discontinuous and leaky confining units for the underlying bedrock aquifer. The boulder till is also laterally discontinuous but is highly conductive to groundwater flow where present. The Pine Plains aquifer is discontinuous to the north and west of the WSAAF, where the aquifer rapidly pinches out into lowlands of thin silty sands and clays overlying bedrock.

The bedrock aquifer is comprised of both carbonate and siliceous sedimentary bedrock that thicken toward the

Black River and thin northwesterly toward Area 4 within the model domain. The bedrock aquifer is the principal water-bearing unit at Fort Drum, with the primary water-bearing units being fracture-dominant portions of the upper dolomitic limestones/sandstones of the Pamelia and Theresa formations and the underlying Potsdam

Sandstones. The system is fracture-flow dominant with differences in pumping rates in wells within the bedrock aquifer suggesting that the availability of groundwater from bedrock is dependent on the size and number of fractures intersected by the wells.

Final-GWModelReport_10222025 7

4 Groundwater Flow Model Update The boundary conditions and hydraulic properties were updated or revised to reflect data collected since the model was last updated in 2024. The model extents and resolution are the same as the 2024 groundwater model

(12 layers and 425,520 nodes covering an area of approximately 40 square miles). In general, the 12 model layers represent the units as summarized in Table 4-1 below.

Table 4-1. Model Layer Description

Model Layer Overall Description Area 4 Description

1,2 Medium and Deltaic Sand Medium and Deltaic Sand

3,4 Silty Deltaic Sand Silty Deltaic Sand

5,6 Clay/Silt Clay/Silt or Limestone/Sandstone

7,8 Boulder Till Clay/Silt or Limestone/Sandstone

9,10 Limestone Limestone/Sandstone

11,12 Sandstone or Gneiss Gneiss (no flow layers in Area 4)

There are local exceptions to these units that are reflected in the hydraulic conductivity distribution presented in

Section 4.2. Changes were made to the model structure in Area 4 to reflect the bedrock conditions in that area. All other areas are the same as the 2024 groundwater model.

Cross-sections showing the model structure within Area 4 are provided on Figures 4-1 to 4-3. The cross-sections are similar to those presented in the Updated Field Investigation Summary Report (Arcadis 2025). However, since cross-sections within the groundwater flow model are along a straight line and geology cannot be projected from locations that are not on the cross-section line, there are some differences between the model cross-sections and the Updated Field Investigation Summary Report cross-sections.

4.1 Boundary Conditions

Boundary conditions used in the model are summarized on Figure 4.1-1 and described in more detail below.

4.1.1 Constant Head Cells

The constant head cell locations are the same as the 2020 and 2024 groundwater models. The elevations of the constant head cells were adjusted as part of this model update. The constant head cell elevations within the northern portion of the model proximate to Area 4 were slightly adjusted from the 2024 groundwater model to account for the new groundwater elevations within Area 4 wells as the general increase in groundwater elevations by approximately 2.5 feet in Area 4 compared with the 2024 groundwater model. All other areas of the model have the same constant head elevations as the 2024 groundwater model.

4.1.2 No Flow Boundaries

The no flow boundaries are the same as the 2024 groundwater model. To be conservative, no flow boundaries were assigned to model layers 11 and 12 in Area 4, which represent the gneiss in that portion of the model. As

Final-GWModelReport_10222025 8 discussed in Sections 3.1 and 3.2, based on field observations, the gneiss is not a source of groundwater within

Area 4.

4.1.3 Recharge

Recharge rates and distribution were the same as the 2024 groundwater model. Recharge rates representing infiltration by precipitation to the subsurface were distributed spatially within the model domain according to land use. Values ranged from approximately 17 in/yr, approximately 40% of annual precipitation [NOAA 2025], to approximately 1 in/yr over the airfield sanitary landfill, where a cap has been installed (approximately 2% of annual precipitation).

Similar to the 2024 groundwater model, drainage features, such as small or typically ephemeral creeks, were not explicitly simulated. Rather, the land surface elevation (top of model layer 1) was used such that if the simulated water table elevation increases above the top elevation of model layer 1 during the simulations, groundwater is removed from the model along drainage areas. This option is activated within the recharge package in

MODFLOW-USG.

4.1.4 Production Wells

The existing well field production wells were simulated in model Scenario 2 (Section 6.2) at the current average pumping rates, representing January 2025 to July 2025. Note that times when Well 16 was down for maintenance

(January through March 2025) were not included in the averages:

Well 14 – 59.3 gpm;

Well 15 – 282 gpm;

Well 16 – 127.7 gpm;

Well 17 – 120 gpm; and

Well 18 – 39.3 gpm.

The steady-state and transient model calibration does not include production well pumping as there is significant uncertainty regarding pumping rates on any given day. The production wellfield area monitoring well groundwater elevations are very sensitive to production well pumping (see Section 5.2 for more information regarding overall model sensitivity); however, groundwater elevations in Area 4 are relatively insensitive to pumping at the existing well field production wells.

4.2 Hydraulic Conductivity

The hydraulic conductivity distribution in the model is largely the same as the 2024 groundwater model. The hydraulic conductivity in model layers 9 and 10 were updated in Area 4 to reflect the results of the pumping test for the groundwater monitoring locations and test wells installed in 2025. All other areas are the same as the 2024 groundwater model. The hydraulic conductivity is presented in Figures 4.2-1 to 4.2-6. Consistent with regional geologic data and the Airfield area of the groundwater model, the bedrock was assumed to have a 3:1 (Ky : Kx) anisotropy ratio within Area 4. Tables 4.2-1 and 4.2-2 present ranges of hydraulic conductivities within each model layer for the overall model and Area 4, respectively.

Final-GWModelReport_10222025 9

5 Groundwater Flow Model Calibration Calibration of a groundwater flow model refers to the iterative process of systematically adjusting the model boundary conditions and input parameters within a justifiable or published range of values to obtain as close a match as possible between observed and simulated water levels. The generally accepted process for the comparison of simulated and observed water levels is summarized in American Standard Test Method (ASTM) D-

5490-93 (ASTM 1994) and utilizes the concept of residuals, which is the difference between the simulated and observed water level.

The residuals are evaluated using standard statistical measures, residual mean, residual sum-of-squares (RSS), variance, and normalized root mean squared error. The standard deviation (the square root of the variance) is the median error and needs to be small compared to the range in observed data. In Applied Groundwater Modeling, Simulation of Flow and Advective Transport (Anderson et al. 2015), it is indicated that there are no established industry guidelines regarding the acceptable magnitude of residual statistics, including the residual mean, the standard deviation, or the normalized root mean squared error, other than it is desirable to minimize these values.

Moreover, the acceptability of the calibration is somewhat dependent on the complexity of the model, quantity and quality of available data, and modeling objectives. The model’s ability to represent the observed characteristics of the system being simulated (e.g., observed flow patterns, distribution of heads, flow volumes and water balance, adherence to the CSM, and other subjective criteria) provide another measure of calibration.

5.1 Steady-State Calibration

The groundwater model was calibrated under steady-state conditions using 286 water level targets from the following:

27 water levels collected from Area 4 monitoring/test wells during April/May 2025;

93 water levels from monitoring locations collected as part of the per- and polyfluoroalkyl substances remedial investigation in November 2022; and

Water levels from an additional 166 locations (August or September 2023), representing heads from the

OASIS site and throughout Fort Drum, collected as part of the per- and polyfluoroalkyl substances remedial investigation.

The 259 water level targets outside of Area 4 were included in the steady-state calibration presented in the

Groundwater Model Update Report (Arcadis 2024). The 286 water level targets are compiled in Table 5.1-1. The

93 water levels collected in November 2022 were given a weight of 0.5 to account for the much earlier time period from the other water levels. The term residual within this section refers to weighted residuals and all calibration statistics were calculated using weighted residuals.

The steady-state calibration was evaluated for the overall model, bedrock only, and Area 4 only. Results of the overall model steady-state calibration are presented graphically on Figure 5.1-1 in a scatter plot of simulated versus observed head values. As shown by the RSS, the majority of the data points generally fall near a 1:1 ratio line, which indicates that simulated values approximate the field-measured water levels, with outliers on both ends of the available data. Groundwater elevations are generally both over-predicted in the mid-range of observed heads and under-predicted where under-predicted values tend to fall more within higher observed head ranges.

The residual mean, residual standard deviation, RSS, and absolute residual mean (the average of the absolute value of all the residuals) were calculated to be -0.78 feet, 1.56 feet, 1,240.6 square feet (ft2), and 1.56 feet, Final-GWModelReport_10222025 10 respectively. The scaled residual standard deviation, which is the residual standard deviation divided by the range of observed heads and assesses overall model fit adjusted for scaling effects, is 1.15%. The residual mean is close to zero, relative to the large range in water levels, and the absolute residual mean and the standard deviation have similar values, also indicating a well-calibrated model.

Results of the bedrock portion of the model steady-state calibration are presented graphically on Figure 5.1-2 in a scatter plot of simulated versus observed head values. As shown by the RSS, the majority of the data points generally fall near a 1:1 ratio line, which indicates that simulated values approximate the field-measured water levels, with outliers on both ends of the available data. Groundwater elevations are generally both over-predicted in the mid-range of observed heads and under-predicted where under-predicted values tend to fall more within higher observed head ranges. The residual mean, residual standard deviation, RSS, and absolute residual mean were calculated to be 0.32 feet, 2.16 feet, 314.5 ft2, and 2.16 feet, respectively. The scaled residual standard deviation is

1.80%. The residual mean is close to zero, relative to the large range in water levels, and the absolute residual mean and the standard deviation have similar values, also indicating a well-calibrated model within the bedrock.

Results of the Area 4 portion of the model steady-state calibration are presented graphically on Figure 5.1-3 in a scatter plot of simulated versus observed head values. As shown by the RSS, the majority of the data points generally fall near a 1:1 ratio line, which indicates that simulated values approximate the field-measured water levels, with outliers on both ends of the available data. Groundwater elevations are generally both over-predicted in the mid-range of observed heads and under-predicted where under-predicted values tend to fall more within higher observed head ranges. The residual mean, residual standard deviation, RSS, and absolute residual mean were calculated to be -0.74 feet, 2.03 feet, 153.9 ft2, and 2.03 feet, respectively. The scaled residual standard deviation is

2.60%. The residual mean is close to zero, relative to the large range in water levels, and the absolute residual mean and the standard deviation have similar values, also indicating a well calibrated model within Area 4.

The simulated groundwater elevations and residuals for each model layer are presented on Figures 5.1-4 through

5.1-9. The flow patterns are similar to previous versions of the groundwater model, including the 2024 groundwater model.

5.2 Sensitivity Analysis

Sensitivity analyses were performed on all the parameters adjusted during the calibration process: horizontal and vertical hydraulic conductivity, groundwater recharge, and existing wellfield pumping. The sensitivity analyses focused on quantitative calibration metrics, RSS, and residual mean. These metrics were assessed to determine if the model calibration could be improved (if the RSS can be lowered or residual mean can be closer to zero) and whether the changes in a model parameter have a large or small effect on the model calibration. These metrics were evaluated for the overall model, bedrock only, and Area 4 only. The results of the sensitivity analyses are discussed in detail in Appendix A of this report. All the parameters used in the calibrated model were sensitive either in the overall model, bedrock only, or Area 4 only. Some parameters were more sensitive than others; for example, the overall model was most sensitive to recharge in the airfield area. Area 4 was also sensitive to recharge variation within the area and to changes in hydraulic conductivity within bedrock, but not within the overburden (which is mostly clay/silt in Area 4). The results of the sensitivity analysis for these parameters suggest that they are likely the best fit for model calibration and provide greater confidence in the predicted well yield from Area 4. When parameters are generally insensitive or unconstrained by the calibration, they do not seem to have significant impact on the overall scenario results presented in Section 6. The results of the sensitivity analyses are summarized in Table 5.2-1.

Final-GWModelReport_10222025 11

5.3 Transient Calibration

5.3.1 Transient Model Set-Up

Validation of the calibrated groundwater model was accomplished by incorporating aquifer storage into the model and simulating the groundwater drawdown at 21 monitoring locations collected during the 72-hour pumping test conducted in May 2025 for the five test wells installed in Area 4. Drawdown was calculated as the steady-state calibration groundwater elevation (non-pumping conditions) minus the transient groundwater elevation. Starting heads for the simulation used the steady-state model described in Section 5.1. Table 5.3-1 presents the transient model set-up. The pumping rate for each of the five test wells were varied over time. Stress periods 1 through 12 represent the active pumping period during the 72-hour pumping test. All other parameters and boundary conditions remained the same as the steady-state calibrated model.

Storage was assigned in the model based on aquifer conditions. Aquifer storage and specific yield were varied for each model layer (Figure 5.3-1).

5.3.2 Transient Model Results

The groundwater flow model was transiently calibrated using a set of 288,504 drawdown measurements collected from 21 monitoring locations during the 72-hour pumping test (Table 5.3-1). The simulated drawdown over time for each location is presented on Figures 5.3-2 through 5.3-5. Fifteen monitoring locations (61% of the monitoring locations) very closely matched both the peak observed groundwater drawdown and the observed recovery (A4-

OBS-10D, A4-OBS-03D, A4-OBS-7D, A4-OBS-05D, A4-OBS-04D, A4-OBS-09D, A4-OBS-14, A4-OBS-15, A4-

OBS-13, A4-OBS-08D, A4-OBS-12, A4-OBS-11, A4-OBS-16, Well-3R, and Well-4R). One monitoring location had simulated the peak drawdown higher than the observed drawdown (Well-1). Five monitoring locations did not simulate as much drawdown as was observed in field measurements (A4-OBS-06D, A4-OBS-02D, A4-MW-1, Well-2, and Well-5). This could be due to preferential flow pathways or fracture-dominant flow that is not simulated in the equivalent porous media-based groundwater flow model. Additionally, groundwater recovery (i.e., drawdown decreasing back to zero and return of the aquifer to near pre-pumping conditions) was simulated at a similar rate as observed in the field data. The residual mean, residual standard deviation, RSS, and absolute residual mean for the transient drawdown targets were calculated to be 0.19 feet, 2.20 feet, 1,412,460 ft2, and

1.17 feet, respectively. Of the 288,504 drawdown targets, 86% are within the 10% range of observed drawdown

(2.58 feet).

To ensure the updated model adequately represents the observed drawdown within the existing production wellfield and Area 3, the 2024 groundwater model was previously validated to those pumping tests previously completed by the USACE for Well 17 (USACE 2011) and within Area 3 (USACE 2018). As there were no changes made to those areas of the groundwater model during the update process, these validations were not replicated for this effort.

Final-GWModelReport_10222025 12

6 Production Scenarios The calibrated model was used to assess conceptual multi-well fields in Area 4. Steady-state pumping scenarios were used to evaluate drawdown and the flow area for each production well. Transient pumping scenarios were run to assess variations in travel times from the WSAAF (an area of known groundwater contamination).

The production scenarios assumed the same boundary conditions as the steady-state calibrated model

(described in Section 4). Drawdown from pumping is shown for each scenario. Drawdown was calculated as the steady-state calibration groundwater elevation (non-pumping conditions) minus the groundwater elevation for each scenario. Consistent with known site conditions, 10% porosity was assumed for bedrock material and 30% porosity was assumed for overburden materials. These scenarios are representative of average flow conditions based on an equivalent porous media approach and do not account for preferential flow pathways.

The percentage (and area) of contributing flow for each production well was evaluated under steady-state conditions using the MODular flow ALLocation (MODALL) program (Potter et al. 2008). This program uses the same

MODFLOW-calculated cell-by-cell flow output as MODPATH (Pollock 1989), but rather than tracking individual particle paths, it directly computes the percentage (and area) of contributing flow for each production well.

Forward particle tracking from the WSAAF (an area of known groundwater contamination) was simulated under transient conditions for all scenarios using mod-PATH3DU (Muffels et al. 2020), a MODFLOW-USG specific pathline analysis program developed by S.S. Papadopulos & Associates, Inc., and the University of Waterloo.

This particle tracking is not intended to quantify concentrations that may be observed in the production wells in the future but is only intended to be a qualitative assessment of potential risk posed by the presence of known areas of contamination to the Area 4 wells. While the production scenario simulations below suggest there may be flow contribution from areas of known contamination in groundwater, including the WSAAF, the model cannot evaluate the timing, fate, or transport of compounds present in the model domain, or how dilution from other areas of flow contribution nearer the well may result in dilution of any contaminants. Therefore, any assessment of contaminant flow or potential impacts to the Area 4 wells beyond an assessment of contributing flow areas is beyond the scope of this assessment.

The pumping wells for each Scenario are shown in Table 6-1 below. Each scenario and the results are described in more detail in Sections 6.1 to 6.3.

Table 6-1. Production Scenario Well Rates

Production Scenario

Extraction Rate (gpm)

Well 1 Well 2 Well 3R Well 4R Well 5 Well 14 Well 15 Well 16 Well 17 Well 18

Total

Rate

Scenario 1 – Area 4

Estimated Sustainable Rates 100 115 50 100 135 0 0 0 0 0 500

Scenario 2 – Area 4

Simulated Pumping Rates with Existing Well Fields at

Current (2025) Rates

100 115 50 100 135 59.3 282 127.7 120 39.3 1,128.4

Scenario 3 – Area 4

Simulated Pumping Rates with Existing Well Fields at

Design Rates

100 115 50 100 135 95 300 400 600 95 1,990

Final-GWModelReport_10222025 13

6.1 Scenario 1 – Area 4 Estimated Sustainable Rates

As discussed in the Updated Field Investigation Summary Report (Arcadis 2025), the estimated sustainable rates for the Area 4 production wells yield a combined total of 500 gpm (approximately 0.72 MDG) over the 3-day pumping test:

Well 1 – 100 gpm;

Well 2 – 115 gpm;

Well 3R – 50 gpm;

Well 4R – 100 gpm;

Well 5 – 135 gpm; and

Total – 500 gpm.

The area of contributing flow for each well, drawdown, and impact to nearby water supplies was evaluated. The results of the steady-state analysis in model layer 10 (limestone/sandstone) are shown on Figure 6.1-1. Greater than 1 foot of drawdown was observed within the limestone/sandstone when the five wells were pumping; this drawdown extended to the north and south of Area 4. The primary area contributing to the yield for Wells 1 and 2 originates from beneath the WSAAF. The flow for Wells 3R, 4R, and 5 primarily originates to the east. A small portion of the flow for Wells 3R, 4R, and 5 also originates near the existing production well field when it is not operating. Potential off-site potable wells located to the northeast could have up to 2.2 feet of drawdown in the limestone/sandstone. Water levels at monitoring wells A4-MW-10D and A4-MW-11 should be periodically collected to assess this risk.

The water table drawdown (model layer 1) from the Area 4 test wells is shown on Figure 6.1-2. There is one isolated area where greater than 1 foot of drawdown is shown within Area 4. The simulation results also suggest that the Philadelphia Reservoir is unaffected by pumping from the five test wells in Area 4.

Table 6.1-1 provides a summary of the simulated drawdown in the Area 4 test wells, the Philadelphia Reservoir, and the potential off-site potable well locations, compared to the safe drawdown limits as established by Fort

Drum as their presumed limit for the purposes of permitting the new wells. Drawdown was calculated using both feet below top of casing and feet below ambient (non-pumping) groundwater conditions. The five test wells in

Area 4 do not exceed the safe drawdown limits. The safe drawdown limit of zero (e.g., no impact) was not exceeded for the Philadelphia Reservoir. As noted above, the safe drawdown of zero (e.g., no impact) was exceeded for the potential off-site potable well locations.

6.1.1 Steady-State – Drought Conditions

Drought conditions (reduction in recharge from precipitation) could impact the available drawdown and the area of flow contribution for the pumping wells. To simulate drought conditions, the recharge applied to the model was reduced by 25% below the calibration conditions (see Figure 4.1-1). Drawdown was then calculated as the steady-state calibration groundwater elevation (non-pumping conditions) minus the Scenario 1 groundwater elevation under drought conditions. Therefore, drawdown represents the cumulative impact of the reduced recharge and pumping. This reduction in recharge caused water levels within the bedrock to drop by 3 to 5 feet in the area of Wells 14 to 18 and Area 3 (Figure 6.1-3). In areas of groundwater discharge (Area 4), water levels within the bedrock dropped by less than 1 foot. Due to the drop in water levels, the area of flow contribution for the Area 4 wells expanded slightly to the northeast and south (Figure 6.1-3).

Final-GWModelReport_10222025 14

Potential off-site potable wells located to the northeast could have up to 2.5 feet of drawdown in the limestone/sandstone.

The water table drawdown (model layer 1) from the Area 4 test wells under drought conditions is shown on Figure

6.1-4. There are additional areas of drawdown due to drought conditions, particularly in the airfield area, where there are areas with greater than 10 feet of water table drawdown. There are areas within Area 4 where more than 1 foot of drawdown is shown in the wetlands and intermittent/ephemeral streams due to drought conditions.

The simulation results also suggest that the Philadelphia Reservoir is unaffected by pumping from the five test wells in Area 4. Less than 1 foot of drawdown is expected in the area of the Philadelphia Reservoir due to drought conditions.

Table 6.1-1 provides a summary of the simulated drawdown in the Area 4 test wells, the Philadelphia Reservoir, and the potential off-site…