March 13, 2014

Michael J imenez
Minerals NEPA Project Manager
Superior National Forest
8901 Grand Avenue Place
Duluth, MN 55808

Doug Bruner
Project Manager
United States Army Corps of Engineers, St. Paul District
190 Fifth St. East
St. Paul, MN 55101-1638

Lisa Fey
EIS Project Manager
Environmental Policy and Review
Division of Ecological Services
500 Lafayette Road
St. Paul, MN 55155

Mr. J imenez, Mr. Bruner and Ms. Fey,

Enclosed please find the comments of Great Lakes Indian Fish and Wildlife Commission
(GLIFWC) staff on the Supplemental Draft Environmental Impact Statement (SDEIS) for the
proposed PolyMet project. GLIFWC is an intertribal agency exercising delegated authority from
11 federally recognized Ojibwe (or Chippewa) tribes in Wisconsin, Michigan and Minnesota.
1

1
GLIFWC member tribes are: in Wisconsin -- the Bad River Band of the Lake Superior Tribe
of Chippewa Indians, Lac du Flambeau Band of Lake Superior Chippewa Indians, Lac Courte
Oreilles Band of Lake Superior Chippewa Indians, St. Croix Chippewa Indians of Wisconsin,
Sokaogon Chippewa Community of the Mole Lake Band, and Red Cliff Band of Lake Superior
Chippewa Indians; in Minnesota -- Fond du Lac Chippewa Tribe, and Mille Lacs Band of
Chippewa Indians; and in Michigan -- Bay Mills Indian Community, Keweenaw Bay Indian
Community, and Lac Vieux Desert Band of Lake Superior Chippewa Indians.


Mr. Doug Bruner and Mr. Bill J ohnson
March 13, 2014
Page 2

Those tribes have reserved hunting, fishing and gathering rights in territories ceded in various
treaties with the United States. GLIFWCs mission is to assist its member tribes in the
conservation and management of natural resources and to protect habitats and ecosystems that
support those resources.

As you know, the proposed PolyMet mine is located within the territory ceded in the
Treaty of 1854. GLIFWC member tribes have expressed concern about the potential impacts of
sulfide mining, whether those impacts occur within the 1854 ceded territory, in the 1842 ceded
territory, which includes portions of Lake Superior, or the 1837 ceded territory. The following
comments are submitted by GLIFWC staff with the explicit understanding that each GLIFWC
member tribe or any other tribe may choose to submit comments from its own perspective.

Staff remains, as they have for many years, primarily concerned about the scientific
validity of the SDEIS with regards to modeling, water quantity, water quality, wetlands, and the
assumptions regarding capture efficiencies and long term viability of the engineered structures.
Staff also notes that comments and Major Differences of Opinion (MDO) developed for the Pre-
draft Supplemental EIS have not been resolved and remain points of disagreement. Specifically,
we are submitting comments on the following topics:

Baseflow in the Partridge River Page 1
Discharge from East Berm of Flotation Tailings Basin Page 3
Revised MODFLOW Modeling of Discharge from East Berm Page 7
SDEIS MODFLOW Modeling of Discarded Basin Design Page 10
Perpetual Water Treatment Page 17
Indirect Wetland Impacts Page 19
Seepage Capture Efficiency Page 21
Ability of Goldsim to Accurately Predict Contaminant Concentrations Page 28
Mercury Page 28
Wild Rice Standard Page 28
Alternatives Page 29
No Action Alternative Page 29
Cumulative Effects Page 29
Impacts from Rail Car Spillage Page 30
Loss of High Biodiversity Significance Value Sites Page 30
Financial Assurance Page 31



Mr. Doug Bruner and Mr. Bill J ohnson
March 13, 2014
Page 3

As always, we are willing and available to participate as the lead agencies review and revise the
EIS document. Please feel free to contact me or Esteban Chiriboga in GLIFWCs Madison
office (608) 263-2873 if you have any questions or need further information.



Sincerely,



J ohn Coleman
GLIFWC Environmental Section Leader


Attachments

cc. Tamara Cameron, Chief, Regulatory Branch, Army Corps
Nancy Schuldt, Fond du Lac Water Projects Coordinator
Ken Westlake, USEPA Region 5
Mike Sedlacek, USEPA Region 5
Neil Kmiecik, GLIFWC Biological Services Director
Ann McCammon Soltis, GLIFWC Intergovernmental Affairs Director


Baseflow in the Partridge River

The importance of baseflow in understanding site hydrogeology is hard to overstate.
Unfortunately the quality of flow data collected at the Polymet site is poor and fraught with
uncertainty. Because there has not been a Polymet stream gage at the site and Northshore pit
dewatering has occurred into the Partridge at varying and uncertain times, all flow data from the
site is suspect. Simple upstream, at-site, and downstream flow measurement would have
provided higher quality data but was never collected by the applicant nor required by the state.

There have been several work-arounds to try to overcome the lack of good quality flow
data for the site. The latest has been the addition of 1 cubic foot per second (cfs) of flow to the
Goldsim modeling to account for Northshore pit dewatering. The mine site water modeling data
package very clearly states (SDEIS reference Polymet 2013i, pg 123 & 133) that the 1 cfs added
to Goldsim modeling was to account for constituents added to the Partridge by pit dewatering
from Northshore; It is not relevant to baseflow calculations nor is it relevant to determination of
aquifer conductivity or groundwater travel times.

In determination of baseflow, all GLIFWC's calculations have excluded Northshore
pumping from the calculation. The Dec. 17th MNDNR memo (Attachment A) also picked a
period when pumping for Northshore pit dewatering was not occurring so as to calculate true
baseflow. The 1 cfs added to Goldsim modeling of the Partridge, mentioned in various DNR
documents, is irrelevant to the calculation of baseflow and does not solve the modeling problems
in XP-SWMM, MODFLOW and by extension Goldsim. Some of the implications of incorrect
baseflow are highlighted on page 114 of the water modeling data package (March 2013), in our
memo of 2012-03-02, and in GLIFWC's baseflow summary of 2014-02-13 (Attachments B, C,
and D respectively).

Because the implications of baseflow are substantial when it comes to a basic
understanding of the mine site hydrogeology, all modeling of flow and by extension contaminant
transport must be re-calibrated to the higher baseflow numbers indicated by GLIFWC's analysis
1

of 2013-07-02 (Attachment E) and DNR's 2013-12-17 analysis (Attachment A). Page 114 of the
mine site Water Modeling Data Package makes it clear that re-calibration of the MODFLOW
model generates new conductivity values that are then fed into Goldsim. It states:
"The revised model calibration resulted in different optimized values for the horizontal
hydraulic conductivity of the surficial aquifer and bedrock, which are used to establish
the distribution of values used for the probabilistic groundwater flow path modeling
(Section 5.2.3.1)."
It is also clear that higher hydraulic conductivities for the aquifers result in faster contaminant
transport to points of evaluation.

Although baseflow assumptions have significant effects on Goldsim modeling, the
implication of re-calibrating the MODFLOW model go beyond the conductivities used in the
Goldsim modeling. Higher baseflows imply higher conductivities that imply faster and greater
groundwater flow rates. This affects:
1) The amount of water expected to flow into the mine pit as it is excavated.
2) The amount of drawdown of Partridge River flow that can be expected due to pit dewatering.
3) The amount of wetland dewatering that can be expected due to pit dewatering.

Given the uncertainty in baseflow numbers due to the poor quality flow data, it is
reasonable to re-calibrate the MODFLOW model to a range of values that included the
previously assumed baseflow and the newer, higher baseflow numbers.
2

Discharge From East Berm of Flotation Tailings Basin:

Significance:
The contaminant transport analysis at the Flotation Tailings Basin (FTB) does not include
any accounting for discharge through the east berm of the basin. There are 3 reasons why
discharge through the east berm will be enough to cause environmental concern:
1) the flow distance between the final FTB pond in cell 1E and the exterior of the east berm is
relatively short compared to flow distances from the pond to the north and west berms (SDEIS
Figure 3.2-29).
2) the east berm is underlain with 25-50 feet of conductive surficial material (SDEIS Figure
4.2.2.-12 and Figure 2 below).
3) the basin pond level is 1720 ft, the land elevation east of the basin is 1660 ft (Lidar data:
http://www.mngeo.state.mn.us/chouse/elevation/lidar.html). The elevation difference between
the pond and the adjacent land surface is substantial; 1720 ft - 1660 ft =60 ft.

Because there has been no prediction of discharge from the east side of the FTB, there
was no flow path established or contaminant transport analyzed in the easterly direction. The
SDEIS is completely devoid of any mention or analysis of flow from the basin toward the east.
Receiving waters for the contaminated discharge would be wetlands adjacent to the basin, Spring
Mine Lake, Spring Mine Creek and wetlands to the north if a proposed stormwater drainage
swale is constructed.

Polymet Modeling of Flow from the Basin:
Polymet modeling with MODFLOW (RS13 Attachment A-6 2007; RS13B Attachment
A-6 2008; Polymet 2013j Attachment A 2011), for the FTB has prevented any discharge of basin
water to the east by erecting a no-flow boundary at the surface of the berm and at the ground
surface. This no-flow boundary is an artificial construct that has no basis in reality. In reality,
flow to the east will be controlled by the relative head pressures and the conductivity of the
materials in the FTB, beneath the FTB and in the berms.
3


Geology Beneath the East Berm:
Examination of the geologic data for the site indicates that the east berm of the FTB sits
on a bedrock valley filled with surficial material that is 25 to 50 feet deep. The bedrock valley
under the east berm is the historical stream channel for Trimble Creek prior to the creation of the
current tailings basins (Figure 1). The thickness of the surficial material under the east berm is
indicated as 25 to 50 feet in the depth to bedrock map of the SDEIS Figure 4.2.2.-12 (Figure 2)
and in the depth to bedrock map MN Geological Survey M-126.

Figure 1. USGS Topo map showing the historic stream channel of Trimble Creek.

4


Figure 2. SDEIS Figure 4.2.2.-12: Depth to bedrock map.

The distribution of bedrock under the FTB has been represented in 2 ways during
Polymet MODFLOW modeling. Technical document RS13 of Nov. 16, 2007 Attachment A-6
Fig. 4-2 showed bedrock in the 2007 MODFLOW model as extending under the eastern quarter
of the tailings basin. In technical document RS13b of Sept. 8, 2008 Attachment A-6 Fig. 4-7h,
bedrock in the 2008 model did not extend under the basin but rather showed the basin to be
underlain with surficial material. The text of RS13b, section 4.6.1 of Attachment A-6 states:
"The location of the bedrock hills that flank the Tailings Basin to the east and south were
updated. The location of the bedrock hills is used in the model to define the extent of the
low hydraulic conductivity zone that represents the bedrock. Because the footprint of the
5

Tailings Basin Mitigation Design is closer to these hills on the southeast side of the
footprint than was the footprint for the proposed design, it was important to get the
location of these hills as accurate as possible. The location of the bedrock hills was
defined using information from the Minnesota Geological Surveys map M-164. The
resulting zones of hydraulic conductivity can be seen on Figure 4-7."
The extent of the tailings basin footprint represented in RS13b is the same extent as
currently proposed in the SDEIS. However, evaluation of flow from the basin using
MODFLOW and Goldsim appears to have fallen back to the 2007 representation of the basin
footprint and of the underlying bedrock (see GLIFWC comment re: SDEIS modeling and
mitigation basin design).

Conceptual Model of East Berm:
A conceptual diagram of the east berm is provided below. The head difference between
the top of the basin (~1720 ft), the head pressures expected in the surficial deposits below the
center of the basin (1700 ft; RS13b, 2008), and the head pressure at the toe of the basin (1660 ft)
will push water toward the toe of the east berm. The 25-50 feet of surficial deposits in the
bedrock valley under the east berm will conduct water under the east berm and beyond.
Figure 3. Conceptual diagram of the east berm of the FTB.

6

Revised MODFLOW Modeling of Discharge from East Berm:

In order to investigate the approximate magnitude of discharge that would exit the east
berm of the FTB, we conducted modified MODFLOW modeling of basin flows in year 20 of the
project. To simulate the basin but without the no-flow boundary imposed in previous Polymet
modeling, we used the 2008 Polymet MODFLOW model (RS13B Draft-01), with the sole
modification being the placement of model drain cells at the east berm.
The original 2008 model predicted flows of 3340 gpm from the basins, 570 of which was
predicted to flow to the seepage barrier on the south side of the basins (SD026) but no flow to
the east because of the no-flow boundary instituted in that model (RS13B Draft-01). Our
placement of drain cells in the east berm area of the MODFLOW model enabled water to move
east from the berm, rather than reverse flow to the north, west and south as was dictated by the
no-flow boundary. The use of drain cells at the east berm to allow eastward movement of water
is an identical approach as that implemented by Polymet for the south berm of the tailings basin
where the discharge to SD026 is modeled by drain cells.
Depending on the exact placement of the drain cells, the modified modeling resulted in
an estimate of 588 to 847 gpm of flow through the east berm of the basin. This flow is on a scale
similar to the flow predicted for the south berm discharge at SD026 (570 gpm, RS13B Draft-01;
or 540 gpm, Polymet 2013j). That the predicted discharges at the south berm and at the east
berm are similar is logical because both areas are underlain by bedrock valleys filled with high
conductivity surficial deposits. In the context of the predicted total discharge from the FTB at
year 20 (3340 gpm, RS13B; or 3230 gpm, Polymet 2013j) the 588-847 gpm prediction suggests
that approximately 1/5 of the FTB water would exit through the east berm.

Implication of Faulty Modeling of Discharge to the East:
At least three problems arise from the current situation of SDEIS modeling of the FTB
with a no-flow boundary on the east and inaccurate representation of bedrock:
1) There is no contaminant transport modeling or evaluation of the water leaving the east side of
the basin. Without substantial engineering to remove the water, a lake toward the 1680 foot
contour would form (Figure 4) until water spilled toward Spring Mine Lake. The Flotation
7

Tailings Management Plan (Polymet 2013m, page16) discusses the need for a drainage swale to
release stormwater from the topographically closed area to the east of cell 1E. In the SDEIS or
supporting documents, there is no discussion of tailings pond water exiting the basin into this
topographically closed area. There is no accounting for contaminants moving eastward, and there
is no description of their possible impact on receiving ground or surface waters.

Figure 4. Cell 1E discharge area.

2) There are potential receiving surface waters near to the east berm; wetlands at the toe of the
east berm, Spring Mine Lake & Spring Mine Creek to the east, and wetlands and an unnamed
creek to the north of the proposed drainage swale.
3) The Polymet MODFLOW modeling was designed to prevent any water from leaving the east
side of the basin by establishing no-flow boundaries on that side of the model. Because of the no-
flow boundaries, the model output files (Northmet Model Files DVD, BARR J uly 2012) show
extremely unrealistic groundwater heads in the aquifer surrounding the east side of the FTB. For
example, the Polymet MODFLOW 2011 model predicts groundwater head to be over 1800 ft in
elevation where the ground elevation is 1660 ft on the east side of the tailings basin. A model
with such distorted groundwater head predictions is unlikely to produce accurate flow
8

information, rendering the flowpaths to the north, west and south and flow quantities used by
Goldsim in the SDEIS unreliable.
Realistic flow modeling of the proposed FTB must be conducted to determine flow
directions, flow quantities and travel rates for environmental impact prediction. Information on
water flow direction and quantity is also needed so that water management plans can be
formulated.
9

SDEIS MODFLOW Modeling Appears to be of Fatally Flawed and Discarded
Tailings Basin Design
Modeling in the SDEIS appears to be of a Flotation Tailings Basin (FTB) design that was
discarded several years ago and does not model the currently proposed basin design. The 2007
FTB design, that is modeled in Attachment A (2011) of Polymet 2013j, was deemed to be
"fatally flawed" by the MNDNR (Mitigation Table, Arkley email of 2008/12/09) and was
replaced by the "mitigation" design developed in 2008.
GLIFWC staff have posed a series of questions to the lead agencies regarding the
modeling for water quantity and flow direction at the FTB. ERM has provided a series of written
responses to those questions. The 2014-03-10 Response 4 from ERM re: the Plant Site
MODFLOW modeling identified Attachment A of the Water Modeling Data package of March
2013 (SDEIS reference Polymet 2013j) as the documentation of the tailings basin flow modeling
for the SDEIS.
Careful examination of the scant information in the above referenced Attachment A
(2011) indicates that the modeling done in 2011 for that attachment was not of the FTB as
currently proposed. The footprint modeled for attachment A is the footprint of an early FTB
proposal from 2007 (Figure 5) that was supplanted by the FTB design developed during the
"Mitigation Options" process of 2008. The 2008 mitigation FTB design (Figure 6) is the current
design footprint assumed in the text of the SDEIS (SDEIS Fig. 3.2-23). In addition to using a
discarded FTB design footprint, the modeling in Attachment A also used a crude representation
of bedrock that was supplanted by a more refined bedrock representation during the modeling of
the 2008 mitigation design (RS13B Draft-01, 2008).
The diagrams and model files supporting Appendix A (2011) further demonstrate that the
modeled footprint is of the 2007 fatally flawed FTB design (see footprints in layer 1 of 2007
(Figure 7) and 2011(Figure 8) models, attached), instead of the mitigation basin design (see
footprint in layer 1 of 2008 model, (Figure 9)). The rejected basin design had a smaller footprint
and did not extend as far to the south and south-east. Unlike the current design, the rejected
design did not cover the ash disposal site in the south-east end of the FTB. It appears that the
10

SDEIS Goldsim (water quality) modeling is based on MODFLOW (water quantity) modeling of
an old FTB design that was deemed fatally flawed and is not modeling the currently proposed
FTB design.
11


Figure 5. 2007 FTB Footprint.
12

Figure 6. 2008 FTB Footprint.
13


Figure 7. Footprints in layer 1 of 2007 MODFLOW model.
14


Figure 8. Footprints in layer 1 of 2011 MODFLOW model.
15


Figure 9. Footprints in layer 1 of 2008 MODFLOW model.
16

Perpetual Water Treatment
The proposed Polymet project would require long term treatment of water at both the
plant and mine sites. This treatment would be needed for centuries but the lead agencies have not
required that the applicant provide an estimate of when treatment would no longer be needed.
Therefore, as articulated in Chapter C, GLIFWC staff maintain that water treatment for the
proposed Polymet mine is perpetual.
GLIFWC staff are gravely concerned that the lead agencies are attempting to minimize
the issue of perpetual/long term treatment by using vague and confusing language in the SDEIS.
In addition, the language the lead agencies have used has changed during the development of the
document even though the model results have not.
The SDEIS states on page 5-7:
Mechanical water treatment is part of the modeled NorthMet Project Proposed Action
for the duration of the simulations (200 years at the Mine Site, and 500 years at the Plant
Site). The duration of the simulations was determined based on capturing the highest
predicted concentrations of the modeled NorthMet Project Proposed Action. It is
uncertain how long the NorthMet Project Proposed Action would require water
treatment, but it is expected to be long term; actual treatment requirements would be
based on measured, rather than modeled, NorthMet Project water quality performance, as
determined through monitoring requirements. (Emphasis added)
In response to comments on the PSDEIS (Comment GLIFWC1) the Co-Lead agency
disposition states:
Modeling predicts that treatment activities will be a minimum 200 years at the
Mine Site and a minimum of 500 years at the Plant Site. While long-term, these time
frames are not necessarily perpetual. The owning company would be held accountable to
maintenance and monitoring required under permit and would not be released until all
conditions are met (Appendix C SDEIS) (Emphasis added)
17

It is impossible to reconcile these 2 statements. We agree that the duration of simulations
were based on capturing the highest predicted concentrations of the modeled action. However,
those concentrations require water treatment to avoid violating water quality standards.
This treatment is at minimum 200 years at the mine site and 500 years at the plant site. As the
lead agencies indicate, these time estimates are only minimums and there is no information that
points to a time when water treatment would not be needed.
Finally, while the maximum contaminant plume is predicted to occur at the 200 and 500
year mark for the mine and plant sites respectively, this does not mean that contaminants
immediately drop to zero. The reduction would be gradual and perhaps last for another few
centuries. In addition the SDEIS states on page 5-56:
The attenuation effect resulting from sorption is significant enough that arsenic, copper,
and nickel are not predicted to travel from source areas to any evaluation locations or the
Partridge River within the 200 year model simulation period (Barr 2013f). Analytical
calculations suggest that the travel times for these solutes would be in the order of
thousands of years."
This statement suggests that water treatment activities would be required far beyond the
200 year time frame at the mine site and would be on the order of thousands of years. Therefore,
the only logical conclusion is that water treatment is perpetual at this project.
It is also important to note that, in the response to GLIFWC comments on the PSDEIS,
the lead agencies acknowledge monitoring and maintenance requirements during the same 200
(mine site) and 500 (plant site) year timeframe.
The SDEIS requires substantially more transparency on one of the most fundamental
issues at stake for this project. The fundamental question is: how long will the company be
required to operate and maintain expensive mechanical treatment to meet water quality
standards? This singular issue has significant repercussions for the public interest determinations
and the scale of required financial assurance.

18

Indirect Wetland Impacts
The methods used in the analysis of indirect wetland impacts in the SDEIS are essentially
the same as the 2009 DEIS. GLIFWC staff reiterate the comments we have provided in the past
that the method is overly simplistic, based on a flawed conceptual understanding of hydrology at
the mine site and inadequate for the NEPA process of a large scale sulfide mine.
The SDEIS has underestimated baseflow at the mine site. The entire conceptual model of
perched wetlands with hydrology that is completely decoupled from groundwater was supported
by the use of unrealistically low baseflow numbers. Now that the applicants XP-SWMM model
has been discredited and that it is obvious that the movement of groundwater at the mine site is 3
times greater than the SDEIS indicates, the assumption that wetlands will not be impacted by
groundwater drawdown should be abandoned. The higher baseflow numbers support the
independent analysis of indirect wetland impacts provided by the tribal cooperating agencies in
Appendix C.
The lead agencies have also based their analysis on the Bog Memo prepared by the Army
Corps of Engineers (Eggers, Steve (2011) MEMORANDUM SUBJ ECT: Distinguishing
Between Bogs That Are Entirely Precipitation Driven Versus Those with Some Degree of
Mineral Inputs from Groundwater and/or Surface Water Runoff). This memo uses plant
community information to determine the degree of hydrologic connectivity between a wetland
and groundwater. The conclusions in the memo are appropriate for a system that is not
experiencing depressurization of the aquifer (drawdown). However, when mine induced
drawdown occurs, new downward pressure gradients are created. Whittington and Price
documented that these downward hydrologic gradients can in fact dewater wetlands that are
entirely surface water dependent under normal conditions )Whittington, PN and J S Price, The
effects of water table drawdown (as a surrogate for climate change) on the hydrology of a fen
peatland, Canada. HYDROLOGICAL PROCESSES, 20(17), 3589-3600. 2006). The bog memo
is not an assessment of the hydrologic conditions of wetlands in a dewatered state but rather an
assessment of surface hydrology under normal conditions. The indirect wetland impact analysis
should be performed using realistic hydrologic assumptions and appropriate mitigation should be
required.
19

Figure 10. Summary of issues related to indirect impacts to wetlands.
20

Seepage Capture Efficiency
As detailed in comments submitted to the lead agencies for the 2009 DEIS and for the
current SDEIS, water quality analyses for the Partridge and Embarrass Rivers are inadequate.
The results, be they deterministic (DEIS) or in the form of probability distributions (SDEIS) are
based on a flawed understanding of hydrology at both mine site and plant site. This flawed
understanding, reflected most prominently in the errors in baseflow calculations, is carried
forward to the MODFLOW hydrologic modeling. At the mine site MODFLOW under-predicts
the amount of water that would flow into the mine pits and thus under-predicts the amount of
water treatment needed for both short and long term closure. At the plant site, the MODFLOW
model is constructed in a way that is not representative of reality and therefore yields results that
are not logical. The lead agencies appear to disregard these problems because there is faith that
the seepage capture and treatment systems will work at over 90% effectiveness for centuries. The
SDEIS claims of long term compliance with applicable water quality standards depend entirely
on this leap of faith. On conference calls scheduled to discuss these issues, the lead agency
consultants have stated that the effectiveness of the capture systems have not been questioned
and the lead agencies have not been able to provide any references that would support their
position. We suggest that there are substantial reasons for skepticism regarding capture
efficiency for the flotation tailings basin, hydrometallurgical tailings basin, and category 1
stockpile seepage capture systems. This skepticism is based on available literature and the
performance of other facilities in the immediate vicinity.
The EPA conducted an analysis of the effectiveness of seepage capture systems (Evaluation of
Subsurface Engineered Barriers at Waster Sites, United States Environmental Protection Agency
(EPA), 1998). This analysis looked at capture systems at 36 facilities and evaluated their
effectiveness based on the performance requirements at each site. It is difficult to extrapolate the
results of this analysis to the Polymet setting because a) the required effectiveness varied from
facility to facility; b) the way in which effectiveness was measured was different (i.e. water
quality improvements downstream versus change in hydrologic head pressure) and c) data
collection varied between facilities. Despite these difficulties, the report indicates that 10% of the
reviewed containment systems failed to meet the desired performance objectives and required
21

corrective action. An additional 19% of the evaluated facilities did not have sufficient data to
conclude whether the containment system was operating successfully or not. Furthermore, there
is no information on the effectiveness of any of these facilities at timeframes remotely
comparable to the needs at Polymet. In the EPA report, long term is considered 30 years whereas
the water capture needs at Polymet are perpetual for the flotation tailings basin, category 1
stockpile and hydrometallurgical tailings basin. Finally, none of the facilities in the study are as
large as the one proposed at Polymet.
At the tailings basin, Polymet has proposed to install a seepage collection system around
the north and west sides of the facility. The scale of this engineering control is extensive. It
would be approximately 5 miles long and would have to be keyed to bedrock that is 25 to 50 feet
below ground surface. The most likely pathway for leakage at this barrier will be in the vicinity
of the key with bedrock (EPA 1998). This feature, and the similar containment system at the
Category 1 waste rock stockpile are assumed to capture 93% of water leaving the facilities for an
indeterminate period of time. As previously stated, there is no scientific justification for this
number. The only examples we are able to identify at this time suggest capture rates that are
lower.
22



Figure 11. Summary of issues surrounding the proposed seepage capture systems.
In the Iron Range, GLIFWC staff are aware of 2 examples that are directly analogous to
the proposed Polymet containment system. These are the seepage collection system at SD026 on
the LTV basin itself, and the seepage collection system at the MINTAC tailings basin.
SD026
NorthMet water management plan version 2 states that the south side seepage capture
facility is already operational. The SDEIS further states that the system is operating effectively
and capturing all seepage out of the south end of the facility. This statement is factually
incorrect. MPCA indicates that the seepage capture system at SD026 it is not working properly
and additional work must be performed if it is to achieve the desired water quality improvements
23

in Second Creek (MPCA Personal Communication). There are two immediate comments
regarding SD026. First the SDEIS text must be corrected to accurately describe the lack of
effectiveness of the seepage capture system and second, a quantitative assessment of the
cumulative water quality effects of this wastewater seepage to Second Creek and the Partridge
River should be performed. In addition, the NorthMet water management plan v2 states that the
seepage capture system would be redesigned if necessary. Given that it is necessary, the redesign
of the system should be included in the EIS document.
MINNTAC
The MINNTAC tailings basin is of similar age and design as the LTV tailings basin that Polymet
proposes to use. Both are large, unlined facilities that are designed to allow water seepage to
surface and groundwater in order to maintain structural stability. Both facilities have been
discharging thousands of gallons per minute of high sulfate wastewater into the environment for
decades. MINNTAC, as part of a schedule of compliance, has begun constructing a seepage
capture system that is intended to bring the facility into compliance with applicable water quality
standards. The capture system is similar to the one proposed by Polymet in that it consists of a
trench to capture seepage and a system that would pump tailings water back into the facility. The
MINNTAC system was originally intended to extend to bedrock but that extension was not
possible in some locations because of the presence of large boulders that made construction
difficult. Because the geology of the surficial deposits is similar at the LTV facility, it is likely
that similar difficulties will be encountered by Polymet that would decrease capture efficiency. It
is important to note that seepage capture of greater than 95% is needed at MINNTAC in order to
achieve compliance with applicable water quality standards (Subsurface Evaluation and Seepage
Evaluation Report, MINNTAC Tailings Basin, Mountain Iron Minnesota, US Steel Corp., 2008).
However, this high capture efficiency was not considered feasible and MINNTAC predicted that
their capture efficiencies would not exceed 60% (Subsurface Evaluation and Seepage Evaluation
Report, MINNTAC Tailings Basin, Mountain Iron Minnesota, US Steel Corp., 2008). Actual
performance of the capture system is below 50%. Ultimately, the main purpose of the system is
to comply with water quality standards. The capture system will not be able to achieve that goal.
24

Because MINNTAC is the only facility that is analogous to the LTV basin, there are serious
doubts about the predicted 90% or greater capture efficiency used in the Polymet SDEIS.
Seepage in bedrock is incorrectly characterized. The lead agencies have maintained that
little to no water flows through the bedrock at the site but have not provided sufficient
justification for this assumption. In fact, the SDEIS assumes that the bedrock is a no flow
boundary and therefore assumes that no water moves through bedrock at all. Mapping of known
faulting in the area indicates that there is a strong possibility for water to move quickly through
faults and fractures (Figure 12). Evidence for fault and fracture flow is also found in the water
quality sampling done at the mine site. Water samples in two deep boreholes at the mine site
found Tritium and un-ionized ammonia. The presence of these constituents indicates a
hydrologic connection with surface water. Tritium indicates water found in the deep boreholes
was surface water post 1950, because it is only after nuclear testing that this constituent entered
surface waters. Un-ionized ammonia is produced by blasting activities at taconite facilities. The
Northshore pits, which are the closest sources for this constituent, are located one mile northeast
of the sample boreholes, and are connected to the Polymet mine site through bedrock fractures.
Review of the Northshore pit discharge monitoring data for SD001, in 2006 and 2008, shows the
average concentration of un-ionized ammonia exceeded the 0.04 mg/l NPDES permit limit. This
indicates that groundwater travel time through bedrock faults and fractures will be orders of
magnitude faster than project modeling for Polymet suggests. There is no reason to expect that
fractures and faults do not occur at the plant site. Therefore, tailings water will escape through
bedrock in quantities and speeds that exceed those described in the SDEIS. Finally, faults and
fractures could exacerbate the problem of water bypassing the seepage containment system at the
top of the bedrock.

Summary for Seepage Capture Comments
The prediction of water quality standard compliance for this proposed project hinges on
the perfect operation of the water capture systems. The reliance on this engineered containment
system that uses overly optimistic capture rates and must function in perpetuity is not
scientifically supported and therefore is not appropriate for the SDEIS.
25

The water quality and quantity impacts at both plant site and mine site should be
remodeled by using a range of capture efficiencies. We suggest 60%, 70%, 80% capture rates be
modeled for the tailings basin and category 1 stockpile. Water quality values for each of these
capture rates should be reported. This will allow the public and decision makes to have a realistic
picture of the risk and uncertainty for this project.
Seepage capture at the flotation tailings basin does not account for seepage out of the east
side of the basin. The seepage capture system should be expanded to account for this expected
discharge. A MODFLOW model was developed to assess the amount of seepage that would flow
out of the basin. As detailed in GLIFWC comments, that model is designed in a way that does
not conform to reality and therefore the results are unreliable.

26

Figure 12. Known Faults and Fractures in the area of the Polymet Project. (EMorey, G.B., and Meints, J oyce,
compilers, 2000, Geologic Map of Minnesota, bedrock geology (3rd edition) : Minnesota Geological Survey State
Map Series S-20.)
27

Ability of Goldsim to Accurately Predict Contaminant Concentrations:
We remain concerned about the inability of Goldsim to accurately predict current and
future contaminant concentrations. This is particularly troubling in the lower Partridge River
(e.g. SW005) and in Colby Lake where Goldsim predictions of current conditions appear to be
inaccurate. In recent conversations with the lead agencies and ERM, there has been agreement
that the modeling in the SDEIS does not accurately capture the environmental conditions at
Colby Lake. Additional modeling of this waterbody is needed to assess impacts of the proposed
project and to evaluate the suitability of Colby Lake water for use in augmenting the flow of
other waterbodies. In addition, the discrepancies between modeled and observed data at SW005
should be addressed in detail.
Mercury
The SDEIS does not adequately address mercury concerns as detailed in Appendix C.
The issue of bioaccumulation of methyl mercury in fish, especially in a sulfate rich environment
remains unaddressed for both project specific impacts and cumulative impacts.
Wild Rice Standard
The concerns over the MPCAs interpretations and recommendations regarding the wild
rice sulfate standard have not been resolved. The information provided in Appendix C is still
applicable to the SDEIS.
In addition, staff believe that water quality modeling underestimates the amount of
sulfate at points of compliance. Even with this problem, contaminant modeling suggests that the
sulfate standard will be violated in the Partridge River points of compliance approximately 10%
of the time. While this may meet the lead agencies arbitrary evaluation criteria (standard met
90% of the time) it certainly is not enough to warrant the issuance of an NPDES permit. At the
Embarrass River the standard is already exceeded at the point of compliance because of historic
contamination from the tailings basin and the area 5 pits. It is not clear if the capture system
around the tailings basin will function well enough to allow the standard to be met.

28

Alternatives
Staff continue to believe that the underground mine and west pit backfill alternatives
have not been properly explored given the environmental benefits they could bring to the project.
Our comments stand as detailed in Appendix C.
In addition, there are a number of alternatives that the SDEIS fails to explore. These
include paste backfill, immediate operation of the RO treatment facility at the mine site, etc.
Additional details are found in the comments submitted by the Fond du Lac Band.
No Action Alternative
During the review of the PSDEIS GLIFWC staff commented that continuation of existing
conditions was not an appropriate No Action Alternative (Table 8-1, Item 10, Chapter 8). The
lead agencies responded by defending their work in the PSDEIS and disagreeing with our
position. However, in the SDEIS the lead agencies completely eliminated any analysis for the No
Action alternative. SDEIS Section 5.2.2.2.3, page 5-78 of the SDEIS was rewritten to point out
that:
"It is important to note, however, that this modeled Continuation of Existing Conditions
Scenario is not the same as the No Action Alternative, which is described in Section
5.2.2.4."

Unfortunately the SDEIS has no serious analysis of a No Action Alternative. Section
5.2.2.4 is less than 1 page long and gives a very general and hypothetical discussion. It in no
way represents a serious analysis of a No Action Alternative. The SDEIS needs to have
modeling of a No Action Alternative, as we describe in SDEIS Appendix C, Hydrology Section,
topic 3 so that the impacts of the proposed action can be compared to a scenario where the
project does not happen.

Cumulative Effects
The concerns regarding the cumulative effects analysis have not been resolved. The
information provided in Appendix C is still applicable to the SDEIS.
29


Impacts from Rail Car Spillage
The concerns regarding the hydrologic impacts of sulfide ore dust spillage along the rail
corridor have not been resolved. The information provided in Appendix C is still applicable to
the SDEIS.
Loss of High Biodiversity Significance Values Sites
The concerns regarding the loss of high biodiversity sites such as the 100 mile swamp,
Lynx and Moose habitat and remaining wildlife corridors have not been resolved. The
information provided in Appendix C is still applicable to the SDEIS.
30

Financial Assurance
The Supplemental Draft Environmental Impact Statement (SDEIS) NorthMet Mining
Project and Land Exchange failed to adequately address closure and maintenance costs and
length of time for post-closure treatment in the context of financial assurance requirements.
The failure of the Army Corps of Engineers and Forest Service to adequately address these areas
in the SDEIS, and instead propose they be addressed at a later time when the Minnesota
Department of Natural Resources undertakes the review of mining permits, is an ill-conceived
attempt to either abdicate their federal trust responsibility or delegate it to the state of Minnesota.
The Fond du Lac Band of Lake Superior Chippewa Indians is a member of the Great
Lakes Indian Fish and Wildlife Commission. The Fond du Lac Band retains off-reservation
rights to hunt, fish and gather under the 1854 Treaty including lands and waters that are adjacent
to the proposed NorthMet mine site. All federal agencies, including the Army Corps of
Engineers and Forest Service, have a federal trust responsibility to protect the habitats that
sustain harvests by treaty signatory tribes when completing Environmental Impact Statements.
The Army Corps of Engineers (ACOE) and Forest Service failed to adequately address
the mine closure and maintenance costs, length of time for post-closure treatment, and financial
assurance requirements in the SDEIS. The ACOE and Forest Services position in the SDEIS is
that these items can addressed at a later time by the Minnesota Department of Natural Resources
in the review of future mining permits. This action is an ill-conceived attempt to abdicate their
federal trust responsibility to protect the habitats that support treaty harvests. Despite their
attempts, the ACOE and Forest Service cannot delegate their federal trust responsibility to
protect habitats that sustain treaty harvests to state of Minnesota when it undertakes the process
of permitting the mine.
The SDEIS fails to adequately address the costs for closure and long-term treatment in the context
of financial assurance requirements.
The superficial estimate of financial assurance provides inadequate detail as to how any
of the cost estimates were developed. The DEIS provided a discussion about the options for
31

financial assurance instruments however any substantial discussion of costs and assumptions on
the metrics were not provided and instead postponed until the permitting phase of this Project.
This approach fundamentally contradicts federal environmental policy and must be revised, with
significant additional study, to appropriately evaluate closure, mitigation, reclamation, and
perpetual treatment cost estimates prior to being published in the final EIS. GLIFWC has
identified specific items that need to be addressed in the Final Environmental Impact Statement
on the following pages.
Executive Summary
The Executive Summary fails to provide: 1) an estimated cost for reclamation, 2) an
estimated cost for post-closure maintenance and water treatment, 3) any realistic estimate as to
the length of time that post-closure maintenance and water treatment would be required, or 4)
information as to how financial assurance instruments would be structured to ensure the costs of
post-closure maintenance and water treatment are paid for an uncertain amount of time and for
which models indicate would be longer than 200 years at the mine site and 500 years at the plant
site.
Within the 54 pages of Executive Summary only a single paragraph addresses the issue of
financial assurance as noted below:
State law requires that PolyMet provide financial assurance before a Permit to Mine
can be granted. Financial assurance instruments, such as bonds or trust funds managed
by the state, would pay the estimated cost of reclamation, should the mine be required to
close for any reason at any time or the company is not able to complete its obligations
under the Permit to Mine
1
.
The SDEIS Executive Summary failed to provide either an estimated cost of reclamation
or an estimated cost for post-closure maintenance and water treatment. Since these costs form the
1
Supplemental Draft Environmental Impact Statement (SDEIS) Executive Summary NorthMet Mining Project and
Land Exchange, page ES-54
32


basis for financial assurance requirements and identify key environmental costs of the project
their absence is problematic and a serious omission.
GLIFWC staff have repeatedly requested that the Co-Lead Agencies address the length of
time that post-closure maintenance and water treatment would be required
2
, however edits
prepared by the Co-Lead Agencies for the SDEIS failed to identify a defined time period and
noted only that modelling simulations resulted in (200 years at the Mine Site and 500 years at
the Plant Site)
3
and it is uncertain how long the NorthMet Project Proposed Action would
require water treatment, but it is expected to be long term. The Executive Summary also failed
to explain how financial assurance instruments can be established to cover the cost of
reclamation and post-closure maintenance and water treatment costs if it is uncertain how long
the NorthMet Project Proposed Action would require water treatment
4
. The Executive
summary also failed to communicate water treatment would be longer than 200 years at the Mine
Site and 500 years at the Plant Site.
3.2.2.4.1 Cost Coverage and Estimation
The SDEIS provides a listing of items for which costs must be included in the financial
assurance instrument (i.e. demolition of all structures and remediation of sites [fencing the
perimeters, sloping and seeding the overburden, constructing outlet structures, removing
culverts, etc]) yet fails to provide any estimated costs or the basis for these costs. This section
also notes that Reclamation and post-reclamation costs are required yet fails to provide any
estimated costs or the basis for their estimation (i.e. quantities, unit costs, inflation estimates).
The SDEIS notes, PolyMet would ensure that the financial assurance amount is
established as a function of at least three main variables: 1) extent of surface disturbance and
potential releases from waste storage facilities, 2) reclamation and long-term care standards
(including mechanical water treatment), and 3) reasonable assessment of the costs to execute the
2
Supplemental Draft Environmental Impact Statement (SDEIS) NorthMet Mining Project and Land Exchange,
Table NorthMet Mining Project and Land Exchange PSDEIS (ver.2) - Tribal Comments and Co-Lead Agencies'
Dispositions 8/19/2013
3
Supplemental Draft Environmental Impact Statement (SDEIS) Executive Summary NorthMet Mining Project and
Land Exchange, page 3-59
4
Supplemental Draft Environmental Impact Statement (SDEIS) Executive Summary NorthMet Mining Project and
Land Exchange, page 3-59
33


Contingency Reclamation Plan. The SDEIS provides no discussion as to how these variables are
likely to impact overall costs of the financial assurance instrument and how large the variance of
cost estimates are likely to be.
The SDEIS notes, In addition to the cost of physical closure and reclamation activities as
shown in Table 3.2-15, annual post-closure monitoring and maintenance is estimated to be in the
range of $3.5m - $6m per year. The cost estimates would be finalized by the MDNR during the
permitting processes.

Table 3.2-15 Preliminary Cost Estimate for Closure Year of Closure (end of
year)
Annual Post-
closure Monitoring
and Maintenance
Year 1 Year 11 Year 20
Estimated Range $50m - $90m $160m - $200m $120m - $170m $3.5m - $6m
Source: Foth 2013.
The costs provided in Table 3.2-15 provide no basis for their estimation or other assumptions. The
SDIES failed to provide detailed costs for the physical closure and reclamation of the mine site that
will need to be covered by Financial Assurance Instruments a detailed discussion as to how much
money will be needed from financial assurance instruments and when.

The basis for physical closure and reclamation costs need to be based on the private sector
costs and include realistic profit margins when performing cleanup tasks. Cost to be covered by
Financial Assurance need to include detailed information and cover the following areas: 1) interim
operations and maintenance for agencies when a company declares bankruptcy and leaves the site, 2)
water management and treatment, 3) removal of hazardous wastes and substances, 4) demolition,
removal and disposal of facilities and equipment, 5) earthwork (sloping, backfill, grading), 6) re-
vegetation, 7) long-term operations and maintenance, 8) Monitoring costs, 9) detailed inflation
estimates, 9) provide a cash flow analysis, and 10) detail assumptions in the determination of risk and
uncertainty.

The final EIS needs to include the lifecycle of the pollution control structures built, estimates
for their original construction costs, and projections for replacement costs for timeframes exceeding
34

200 years at the Mine Site and 500 years at the Plant Site. In addition to providing detailed cost
estimation, the final EIS needs to clearly identify and communicate assumptions regarding
inflation rates, rates of return, contingencies, and labor rates. Closure and maintenance costs will
need to be covered years into the future, so a net present value must be included in the final EIS.

3.2.2.4.2 Financial Assurance Instruments

The SDEIS provides a listing of contingencies that may have to be covered by financial
instruments including: 1) physical difficulties in implementing reclamation plans, 2) escalating
standards of closure, reclamation, and long-term monitoring, 3) unanticipated liabilities, 4)
unplanned cessation of mining, 5) failure of the mining company, and 6) failure or limitations on the
ability of third parties to pay reclamation costs. Unfortunately the SDEIS provides no discussion as
to any of the costs of the contingencies that are identified. The SDEIS also fails to discuss how
financial instruments would be structured to meet those contingencies or the assumptions made by
PolyMet to ensure an adequate stream of revenue is available to meet closure and maintenance costs

What fundamental economic assumptions are being made when PolyMet proposes to use
surety bonds, irrevocable letters of credit, cash and cash equivalents, trust funds, insurance
policies, or a combination of these Financial Assurance Instruments? The SDEIS failed to clearly
state how the State of Minnesota will determine the maximum bond requirements, how it
estimated direct reclamation costs, how it determined its estimates for inflation (i.e. periodic
bond recalculation or calculate an Inflation factor using a common index, such as the
Construction Cost Indexes (CCI) from the Engineering News Record), and how it will
determine indirect reclamation costs and how it will calculate the total bond amount. The Final
EIS needs to provide information contained in the Reclamation Bond Summary Sheet that is
attached.

3.2.2.4.3 Cessation of Financial Assurance
The SDEIS notes, PolyMet may cancel financial assurance only upon approval by the
MDNR after it is replaced by an alternative mechanism or after being released (in whole or in
35

part) from financial assurance. The SDEIS fails to discuss any federal oversight of this process
and how the federal government will meet its trust responsibility in protecting habitats that
support off-reservation treaty harvests.
4.2.1.4.2 Legacy Contamination

The SDEIS discusses Cliffs Erie site, identifies 62 Areas of Concern (AOCs), and
discusses PolyMets role in site remediation. The SDEIS failed to provide any information as to
cost estimates for addressing the legal requirements for mitigating the AOCs as identified. This
information is needed to ascertain if the proposed project would further contaminant AOCs and
increase clean-up/remediation costs.


36









37

Attachment A
Date: March 8, 2013
NorthMet Project
Water Modeling Data Package
Volume 1 - Mine Site
Version: 12 Page 114


[()]

Equation 5-22
For example, if the aquifer length is 1000 meters, then the desired dispersion length from
Equation 5-22 is 11.8 meters. Then, because the dispersion length is equal to one-half the cell
length, the cell lengths should be approximately 23.6 meters. Given the aquifer length and the
optimal cell length, the aquifer will be represented using 42 cells to obtain the desired degree of
dispersion.
5.2.3.7 Groundwater Inflow to Mine Pits
For the DEIS modeling, a MODFLOW model of the Mine Site was used to calculate
groundwater inflow rates to the pits during operations and the expected head distribution and
groundwater flow directions during reclamation and long-term closure (Reference (32)). The
groundwater flow rates to the pits were used to develop the water balances for the pits, which
directly affect the water quality within the pits. The distribution of heads in closure was used to
establish the groundwater flow paths that were used in the MT3D models to evaluate dissolved
solute transport and potential groundwater impacts associated with the Project (Reference (11)).
For current modeling, a similar approach is used; however, several modifications were made to
the previous MODFLOW model to incorporate new information. A brief discussion of the
changes to the model is included here, with additional details regarding model setup presented in
Attachment C. The DEIS MODFLOW model was calibrated to a baseflow estimate of 1.43 cfs at
monitoring station SW004. Revisions to the XP-SWMM model since the DEIS modeling
(Section 5.2.4.3) resulted in different baseflow estimates for the Partridge River. The
MODFLOW model was re-calibrated using target baseflow values of 0.41, 0.51, and 0.92 cfs at
SW002, SW003, and SW004, respectively. In addition, groundwater elevations measured at
Mine Site monitoring wells MW-1 through MW-18 were included as targets in the updated
calibration. The automated-inverse modeling code PEST (Reference (67) and Reference (68))
was used to complete the model calibration. Details of the model calibration are presented in
Attachment C. The revised model calibration resulted in different optimized values for the
horizontal hydraulic conductivity of the surficial aquifer and bedrock, which are used to establish
the distribution of values used for the probabilistic groundwater flow path modeling
(Section 5.2.3.1).
To calculate groundwater inflow rates to the pits during operations, MODFLOW simulations
were developed using methods similar to those used for the DEIS modeling (Reference (32)).
The footprints and vertical extent of the mine features was modified from the DEIS model to
reflect the current Mine Plan. Details regarding the simulation setup and results are included in
Attachment C. The estimates of groundwater inflow rates to the pits were used for the overall
water balance of the pits in the probabilistic model (Section 5.2.2.6.3).
Attachment B
550 Babcock Dr., Rm. B102
Madison, WI 53706
608-263-2873 Fax 608-262-2500 1 Baseflow_calibration_v2012-03-02.wpd
GREAT LAKES INDIAN FISH AND WILDLIFE COMMISSION
P. O. Box 9 ! Odanah, WI 54861 ! 715/682-6619 ! FAX 715/682-9294
! MEMBER TRIBES !
MICHIGAN WISCONSIN MINNESOTA
Bay Mills Community Bad River Band Red Cliff Band Fond du Lac Band
Keweenaw Bay Community Lac Courte Oreilles Band St. Croix Chippewa Mille Lacs Band
Lac Vieux Desert Band Lac du Flambeau Band Sokaogon Chippewa
Via Electronic Mail / Original by Mail
March 2, 2012
Memorandum
To: Thomas Hingsberger USACE
Erik Carlson Minnesota DNR
From: J ohn Coleman, Environmental Section Leader
Re: Polymet model calibration to Partridge River low flows
The hydrologic models for the Polymet mine site have been calibrated to targets that
under-represent true baseflow. Models should be calibrated to a strong set of observational data.
Construction of the sites basic hydrologic model to unrealistically low baseflows has
ramifications for all the flow and contaminant modeling at the site.
Under-representation of Partridge River baseflow.
Review of the winter baseflow measurements and comparison to predictions made by
XP-SWMM indicate that XP-SWMM substantially underpredicts baseflow (Barr J une 9, 2011,
Comparison of MDNR winter flow gauging to Partridge River XP-SWMM model). This has
ramifications throughout the parameter sets being used in models characterizing hydrology at the
Polymet mine site.
In the above referenced memo, Barr points out that the average measured baseflow at
Dunka Rd. was 5.0 cfs while the XP-SWMM predicted baseflow is 0.4 cfs. Even when discharge
from Northshore Mining was taken into account, the average baseflow measured at Dunka is 4.3
cfs while XP-SWMM predicts 0.42 cfs.
In its memo, Barr correctly points out that: "At all locations along the main stem of the
Partridge River, the XP-SWMM-estimated baseflow is less than the MDNR-measured baseflow.
The XP-SWMM model provides a conservative estimate of Partridge River baseflow for the
purposes of modeling water quality impacts (e.g., less dilution of loads from the Mine Site)."
What is not acknowledged in the Barr memo is that calibration of hydrologic models to an
underestimate of baseflow produces models that characterize the groundwater hydrologic system
as moving an unrealistically small quantity of water.
Attachment C
550 Babcock Dr., Rm. B102
Madison, WI 53706
608-263-2873 Fax 608-262-2500 2 Baseflow_calibration_v2012-03-02.wpd
Additional flow measures over the last 9 months on the Partridge River at the Dunka
Road (site SW-003) further support the position that baseflow predicted by XP-SWMM under-
represents true baseflow. The least flow measured at the Dunka Road site was 3.8 cfs. While
there have so far been only 7 measurements taken at that site, the flow measured and the stage
recorded by the gauge do not appear to support XP-SWMMs low baseflow predictions for the
upper Partridge River.
Mis-calibration of groundwater flow models.
The calibration of the Modflow model to a Partridge River baseflow of 0.76 cfs predicted
by XP-SWMM results in a model that moves very little water through the groundwater system.
This can result in low predicted rates of inflow to the mine pit and slow movement of
contaminants from sources (stockpiles or reflooded pits) to points of evaluation. More generally,
an incorrect baseflow calibration target results in excessively low estimates of recharge and
likely incorrect estimates of horizontal and vertical conductivity. These hydrologic parameters
are interrelated and getting one wrong, as appears to be the case with baseflow, will almost
certainly result in the other parameters being incorrectly estimated. Although there has been little
sensitivity analysis conducted in the Polymet modeling efforts, flow models tend to be sensitive
to these interrelated parameters.
Based on Modflow model calibration to a baseflow of 0.76 cfs and recharge values set at
0.3 and 1.5 in/yr (see page 61 of Water Modeling Data Package Vol 1-Mine Site v9
DEC2011.pdf and page 11 of RS22, Appendix B), some horizontal and vertical conductivities
(K) were calculated by Barr using PEST (see Table 1 of Attachment B of Water Modeling Data
Package Vol 1-Mine Site v9 DEC2011.pdf). These K values are likely to be inaccurate since
they are calculated with a model that is calibrated to a baseflow that appears to be almost an
order of magnitude too low. It is unlikely that any accurate predictions of water movement,
transport of contaminant mass, or contaminant levels can be made when the characterization of
the hydrologic system is so out-of-kilter.
Unusually low recharge and vertical K:
The low values used for recharge (0.3 and 1.5 in/yr) and the low wetland and till vertical
K (0.0000033 ft/day [1.16X10
-9
cm/s]) used in the Modflow model are a reflection of a model
constructed and calibrated to move an unrealistically small amount of water through the
hydrologic system. For context, note that engineered clay liners in landfills typically aim for
1.0X10
-7
cm/s hydraulic conductivity. I was unable to find any reference in the literature to
wetland soil vertical conductivity as low as is used in the Modflow model. The lower end of the
spectrum I found for wetland soil vertical conductivity was 1X10
-6
cm/s.
Our long standing concern that the mine site hydrologic models incorporate incorrect
assumptions about recharge are supported by Fred Marinelli's comment on line 39 and elsewhere
of: "Agency Responses MS and PS WP and Waste Characterization Data package V7
2-7-12.xls". His comment states that "A net infiltration (recharge) range of 0.3 to 1.5 in/yr
represents 1.1 to 5.4 percent of mean annual precipitation (MAP). This range for local net
infiltration is unrealistically low for this area of the US." These low recharge values and the low
550 Babcock Dr., Rm. B102
Madison, WI 53706
608-263-2873 Fax 608-262-2500 3 Baseflow_calibration_v2012-03-02.wpd
vertical K values are related to calibration of the Modflow model to low baseflow. Until
Modflow, and by extension the other related models XP-SWIMM and GoldSim, are calibrated to
data from the site (e.g. observed baseflow and an adequate number of observed heads) and
incorporate reasonable recharge rates, the results from the models are unlikely to accurately
simulate current or future conditions.
Recalibration of models needed:
The Modflow model, in particular, needs to be calibrated with targets based on observed
baseflow and observed well water heads. Calibration to projections by XP-SWMM, that appear
to be incorrect, means that the fundamental characterization of the site hydrology is likely to be
faulty. In the document referenced above (Agency Responses ...) Barr Engineering states that
many hydrologic model parameters were discussed as part of the IAP process and will not be
considered further at this time. While some parameters were discussed in the groundwater IAP
process, the discussion was almost exclusively concerning water quality parameters, not flow
model parameters such as recharge, baseflow and Kv and Kh. The focus on water quality
parameters to the near exclusion of hydrologic flow parameters is reflected in the Groundwater
IAP summary memo of J une 2011. Groundwater flow modeling underpins contaminant
transport modeling and is interrelated to surface flow models. Without adequate vetting of flow
model parameters and predictions, it is impossible to have confidence in predictions of
contaminant movement and water quality.
Now that the hydrologic models have been more fully articulated by Barr and additional
data are available, the models must be calibrated to observed baseflow and well water levels.
This should include the new water level data from the newly installed mine site wells. PEST can
then be used to more reasonably estimate values for recharge and conductivity. The observed
baseflow and the PEST estimated recharge and conductivity values should then be used in the
XP-SWMM and GoldSim modeling efforts. Modeling efforts that are based on faulty initial
assumptions and not on field observations will not be able to reasonably predict impacts. The
current Polymet modeling effort needs to be well founded on a strong base of observations of the
physical conditions at the site.
Thank you for considering this issue. Please contact me at 608-263-2873 if you have
questions.
cc: Mike Olson, Minnesota DNR
Fred Marinelli, Interralogic
Mike Sedlacek, USEPA
J ames Grimes, USEPA
Marty Rye, USFS
Nancy Schuldt, Fond du Lac Environmental Program
Neil Kmiecik, GLIFWC Biological Services Director
Ann McCammon Soltis, GLIFWC Policy Analyst
2014-02-13
Baseflow Issue at PolyMet
Stream or river baseflow is a key variable in modeling because it is an indicator of the fundamental
characteristics of the groundwater hydrology of a site. It is useful in heling to define the amount and
seed at which ground water moves through the system.
!aseflow is usually calculated by measuring rates of stream flow at stream gauges to define low flow
conditions for a site. "uring low flow eriods the water found in a stream or river is often assumed to be
from groundwater. #or the $oly%et ro&ect' the alicant did not install a stream gauge at the site.
Instead they used data collected in the 1()0*s from a stream gauge located 1+ miles downstream. ,hey
then used a model -.$-S/%%0 to e1traolate that information ustream to the area where the
roosed its would be located -mine site0. ,he result of that e1traolation was a redicted baseflow
rate of 0.2 cubic feet er second -cfs0 in the $artridge 3iver at the "unka 3oad. 4s art of the rocess of
.$-S/%% rediction of baseflow' eriods were chosen when 5orthshore %ine was not discharging it
water into the uer artridge. ,ribal 6ooerating agencies have argued since 200) that the baseflow
rate redicted by .$-S/%% is unreasonably low and imlies recharge to the groundwater system from
reciitation that is not consistent with ublished literature.
Since the initial calculation of baseflow with .$-S/%%' which was used in the first "7IS' the %5"53
has conducted some limited measurements of flow in the $artridge 3iver during winter. /inter flow is
often used as a indicator of baseflow because during winter' it is too cold for rain and other surface
water to enter a stream and thus it is assumed that the flow in the stream is a reflection of groundwater
discharging to the stream bed. ,hose measurements suggested that baseflow was significantly higher
than the 0.2 cfs redicted by .$-S/%%. 89I#/6 staff have' on multile occasions' rovided calculations
of baseflow using alternative methods. ,hose analyses suggest that baseflow in the $artridge 3iver at
the "unka 3oad was in the range of aro1imately 1.1 to 1.) cfs.
In 2011 ,ech 3esources arranged for a stream gauge to be installed in the uer $artridge 3iver which
has allowed the %5"53 and 89I#/6 to conduct calculations of baseflow using data obtained in the
mine site area. ,he %5"53 calculations' released by %5"53 hydrologists in "ecember of 2013' confirm
89I#/6*s osition that baseflow was underestimated by the .$-S/%% model and is in fact closer to 1.2
cfs rather than the reviously redicted 0.2 cfs.
4 confounding variable in the calculation of baseflow is the fact that the 5orthshore %ine discharges it
dewatering water into the $artridge 3iver. ,his means that if flow measurements are taken during times
where 5orthshore is discharging their it water' some of the water measured in the $artridge 3iver may
be mistaken for baseflow. ,his is why both 89I#/6 and %55"53:s "ecember 2013 calculations were
careful to only use $artridge 3iver flow data collected when 5orthshore was not discharging. It is from
these times of no 5orthshore discharge' that the 1.2 cfs of baseflow was calculated.
Attachment D
!aseflow is also highly related to the rate of recharge. 3echarge is the rocess by which rainwater
ercolates into the ground and into the groundwater a;uifers. !ecause baseflow has been
underestimated' the recharge values used in the modeling are also unrealistically low. In fact' the
recharge numbers calculated by the alicant and used in the S"7IS are not suorted by ublished
literature or data collected in the region.
<n =anuary 2)th' 2014 the %5"53 released a statement claiming that $olymet modeling has accounted
for the difference between the 0.2 cfs .$-S/%% rediction and the more recent 1.2 cfs estimate of
baseflow by adding 1 cfs to the 8oldsim model. ,his e1lanation is incorrect because it confuses flow
-which may be a mi1 of water from surface water sources and groundwater sources0 with baseflow
-which is water from groundwater sources only0. ,he %5"53 statement further confuses water
;uantity models -.$-S/%% and %<"#9</0 with the water ;uality model -8oldsim0. >es' a 1 cfs was
added to the water ;uality model -8oldsim0 to account for 5orthshore it dewatering discharges but
this does not relate to baseflow rates redicted and used in .$-S/%% and %<"#9</.
,he difference between flow and baseflow can be demonstrated by the calculation of flow statistics for
the stream gauge on the $artridge. ,he ?(0 for the full data record from the gauge is aro1imately 2.2
cfs. In other words' (0@ of the time river flow was greater than 2.2 cfs at the gauge site. ,his is an
indication of low flow in the $artridge and includes flow from all sources including groundwater and
5orthshore it dewatering. <n the other hand' the ?(0 for eriods when 5orthshore was not
discharging it water into the $artridge is aro1imately 1.2 cfs at the gauge site. ,his is an indication of
flow derived from groundwater discharge to the artridge and is therefore considered an indicator of
baseflow.
<n a conference call conducted on #ebruary 12' 2014 the %5"53 confirmed that the 1 cfs was added to
account for surface water uming from 5orthshore in the 8oldsim model -water ;uality0 and not for
any other urose.
Why Is this Important?
5either the direct winter field observations made by %5"53 -minimum of 3.4 cfs0 nor the values
calculated by 89I#/6 and %5"53 from the stream gauge data -aro1imately 1.2 cfs0' suort the
baseflow redicted by .$-S/%% at S/003 of 0.2 cfs -/ater %odeling "ata ackage Aol.1-%ine Site'
ver12' .130 and $S"7IS ,able 2.2.2-120. .$-S/%%:s low estimates of baseflow have been used in
calibration of the %<"#9</ model and thus influence many asects of the site characteriBation and
imact rediction' including it inflow' dewatering imacts to the $artridge 3iver and wetlands' water
treatment needs' groundwater flow rates' contaminant transort times and concentrations' and
contaminant dilution in the $artridge watershed.
4 higher baseflow rate -1.2 cfs rather than 0.2 cfs0 changes the concetual understanding of how water
asses through the groundwater a;uifer. ,hat change in understanding imacts redictions of how a
mine would affect the a;uifer. ,he conclusions that aear in the S"7IS of no significant imact to
rivers' lakes and wetlands in the mine site are based on the concet that groundwater flows very slowly
through the a;uifer. ,he alicant has assumed' based on the 0.2 cfs baseflow' that wetlands and the
$artridge 3iver are mostly isolated from the groundwater system and that little water will flow into the
oen it from the groundwater system. Cigher baseflows in the $artridge 3iver' as demonstrated by
89I#/6:s and %5"53:s analyses' strongly suggest that the wetlands and river are more connected to
the groundwater a;uifer' that mine it inflow will be greaterD and that groundwater travels through the
a;uifer at a faster rate.
!aseflow is used to formulate the model -%<"#9</0 for calculating the amount of water that would
flow into the oen its during mining. ,herefore' the alicant has underestimated the amount of water
they would need to um out of the its during mining and the amount of water they would need to
treat rior to discharge. It is reasonable to assume that costs of treating this increased ;uantity of water'
both short and long term may have also been underestimated.
!ecause the new baseflow numbers indicate that water moves through the ground faster than the S"7IS
assumes' the lume of contaminants from the reclaimed mine its will likely reach oints of evaluation
faster and and in greater volume. Some have suggested that more water in the system would lead to
more dilution of the $oly%et contaminant lume and thus roduce less water ;uality imact. ,hat is a
naive' and ossibly incorrect' view of a comle1 system. Cigher baseflow relates to higher conductivity
of the a;uifer which has many imacts on flow and contaminant redictions. /ith increased flow
through the a;uifer' contaminant levels at comliance oints could decrease' increase or stay the
same. <nly remodeling of the roosed ro&ect using realistic hydrologic inuts for baseflow and
recharge can rovide an ade;uate answer to this ;uestion.
Subject:
Partridge River baseflow, draft analysis of new data suggest XP-SWMM estimate inaccurate
From: "john.coleman" <jcoleman@glifwc.org>
Date: 7/2/2013 11:56 AM
Attachments: Baseflow_calibration_v2012-03-02.pdf (32.2 KB), 2012-06-12_baseflow info re NorthMet EIS Mine
Site Hydrology Teleconference.eml (2.8 KB), 2012-06-18_watershed ratio predicts baseflow of 1.2cfs
at SW-004 Re Model Calibration, NorthMet EIS.eml (3.1 KB), 2008-09-28_further comments on
RS22 AppenB Draft-03.htm (4.5 KB)
To: thomas hingsberger <thomas.j.hingsberger@usace.army.mil>, Ross Vellacott <Ross.Vellacott@erm.com>,
"Shirley Frank (USFS)" <safrank@fs.fed.us>, "Bill Johnson ( [...]
CC: "Sedlacek.Michael@epamail.epa.gov" <Sedlacek.Michael@epamail.epa.gov>,
"Grimes.James@epamail.epa.gov" <Grimes.James@epamail.epa.gov>
To: Polymet EIS Co-leads 2013-07-02
From: John Coleman, GLIFWC
Re: Partridge River baseflow, draft analysis of new data suggest XP-SWMM estimate inaccurate
We remain concerned that the basic hydrology of the mine site is mis-characterized as being very non-conductive.
The baseflow in the Partridge is a fundamental parameter to which many flow and contaminant transport models are
calibrated. Unfortunate the baseflow at the site used in impact prediction is an estimate make by XP-SWMM.
XP-SWMM appears to do a poor job of predicting baseflow at the mine site, possibly because it is based on a data set
collected 17 miles downstream.
As we note in our recently submitted PSDEIS comments, the MDNR winter flow measurements in the PSDEIS
(Table 4.2.2-9) indicate substantially higher baseflow in the Partridge than predicted by XP-SWMM. This is true even
when the flow data is corrected for any possible Northshore (NS) discharge to the Partridge by subtracting the
farthest upstream measurement from measurements taken farther downstream.
Even more compelling than the winter MDNR flow measurements is the flow data that has been recorded at the
Dunka Road gage over the last 2 years. I have again calculated some statistics on the flow measurements taken at the
Partridge River & Dunka Road, also known as monitoring site SW003. (http://www.dnr.state.mn.us/waters
/csg/site_report.html?mode=get_site_report&site=03155002)
Earlier comments on this topic are attached and previous analysis was submitted to the lead agencies by email on
2012-06-12, 2012-06-18, and on 2008-09-28 (attached).
The stage and flow values measured by stream gage are available at 15 minute intervals. Based on 66,581 stage
records collected between May 2011 and April 2013 and the DNR rating curve, I found:

Q90 at SW003 = 2.32 cfs (90% of the time flow was greater than 2.32 cfs) Q90 is sometimes used as an indicator of
baseflow
Using 586 daily average flows from 2011-05-26 to 2012-12-31 calculated by the DNR and accounting for winter ice
conditions, I found:
Q90 at SW003 = 1.9 cfs
Given that Northshore Peter Mitchel (PM) pit intermittently discharges to the Partridge River, I also analyzed 3
months in 2011 (Jul,Aug,Sep) and 3 months in 2012 (Feb,Mar,Apr) when Northshore (NS) discharged zero (0)
gallons into the Partridge River.
Based on average daily flows calculated by the DNR:
In the 3 months of no NS pit discharge in 2011 Q90 at SW003 = 1.8 cfs
In the 3 months of no NS pit discharge in 2012 Q90 at SW003 = 1.1 cfs
1 of 2 Partridge River baselow, draft analysis of n... 2/24/2014 3:42 PM
Attachment E
Given that both these 3-month periods are typically low flow times, it seems that a baseflow estimate for site SW003
of 1 - 2 cfs would be reasonable.
While analysis based on only 6 months of flow data is not ideal, it should be noted that the XP-SWMM model is
calibrated to only 2 months when Northshore did not discharge to the Partridge in 1985 (PSDEIS page 4.2.2-44, 1st
paragraph).
Neither the direct field observations (minimum of 3.4 cfs) nor the values calculated from the DNR rating curve,
support the baseflow predicted by XP-SWMM at SW003 of 0.51 cfs (Water Modeling Data package Vol.1-Mine
Site, ver12, p.130 and PSDEIS Table 4.2.2-8). XP-SWMM's low estimates of baseflow are used in calibration of the
MODFLOW model and thus influence many aspects of the site characterization and impact prediction, including pit
inflow, dewatering impacts to the Partridge River, water treatment needs, groundwater flow rates, contaminant
transport times and concentrations, and contaminant dilution in the Partridge watershed.
Although it is now an unfortunate time in the NEPA process to try to adequately characterize basic site hydrology, if
appears that predictions of effects of the project may be far from accurate. It is not easy to say how the
mis-characterization of river baseflow would affect compliance predictions because, although more baseflow might
mean more dilution of contaminants, it could also mean transport of greater quantities of pollutants to the river and
more drawdown of the Partridge River. We have repeatedly asked that the data at the Dunka Road gage be formally
analyzed for baseflow as a check of the accuracy of the XP-SWMM modeling. If that analysis indicates that the
XP-SWMM predictions under-represents baseflow, as our draft analysis suggests, that result should be incorporated
into all project model calibration and prediction.
Thank you in considering this issue when revising the SDEIS.
--
John Coleman, Madison Office of the Great Lakes Indian Fish & Wildlife Commission
U.W.-Madison Land Information and Computer Graphics Facility
550 Babcock Drive, Room B102
Madison, WI 53706
608-263-2873 or 265-5639
jcoleman@glifwc.org
2 of 2 Partridge River baselow, draft analysis of n... 2/24/2014 3:42 PM