EPA-600/D-85-043
                                            June 1985
 TOTAL ALKALINITY OF SURFACE WATERS:   A MAP
        OF THE UPPER MIDWEST REGION
                     by

              James M. Omernik
Corvallis Environmental Research Laboratory
    U.S. Environmental Protection Agency
            200 S.W. 35th Street
          Corvallis, Oregon 97333

                    and

             Glenn E. Griffith
          Northrop Services, Inc.
            200 S.W. 35th Street
          Corvallis, Oregon 97333

-------
Abstract

     This map illustrates the regional  patterns  of  mean  annual  alkalinity  of
surface water in the northern portions  of  Minnesota,  Wisconsin, and Michigan.
As such, it provides a qualitative graphic overview of the  relative potential
sensitivity of surface waters to  acidic input  in the  upper  midwest portions of
the United States.  The map is based on data from approximately 14,000  lakes
and streams and the apparent spatial  associations between these data and macro-
scale watershed characteristics that are thought to affect  alkalinity.
                                DISCLAIMER

  The  information in this document has been funded by the United States
  Environmental Protection Agency.  It has been subjected to Agency
  review  and  approved for publication.

-------
     A major goal specified in the National  Acid Precipitation  Assessment  Plan
(Interagency Task Force on Acid Precipitation,  1982)  is  the  quantification  of
the extent of sensitivity of the nation's lakes and streams  to  acidification.
Most earlier efforts to determine patterns of surface water  sensitivity to
acidic deposition have relied on interpretations of bedrock  distribution and
chemistry (Galloway and Cowling, 1978;  Likens et al., 1979;  Hendrey  et a!.,
1980; National Atmospheric Deposition Program,  1982).  One effort  was  based on
soil sensitivity (McFee, 1980) and another on surficial  geology (Shilts, 1981).
While each of these contributed to the general  knowledge of  the extent of
surface water sensitivity, they are in sharp disagreement for many portions of
the country.  More importantly, there is a lack of spatial correlation between
the patterns drawn by these efforts and the  observed  patterns of surface water
alkalinity.

     Although there is general agreement that surface water  alkalinity is
directly related to mineral availability, it is apparent that maps of  rock  type
or soil type alone are inadequate to express patterns of mineral availability
that are meaningful in terms of surface water sensitivity.  For instance,
results from several recent studies of patterns of surface water sensitivity in
different regions of the United States (Eilers  et al., 1983; Haines  and
Akiela'szek, 1983; Twaroski et al., 1984) indicate that no single factor  (e.g.,
bedrock geology) can explain observed patterns  of surface water alkalinity.
Rather, these studies indicate that one must consider a  variety of driving  or
integrating spatial factors that affect alkalinity such  as land use, physio-
graphy, and soil type (as well as geology),  and that  the relative  importance of
any one, or a particular combination of those factors, may vary within or  among
regions.

     A recent report of the National Academy of Sciences (Environmental  Studies
Board, 1984) defined several important geochemical and hydrological  processes
of watersheds that determine whether waters  will acidify and the rate  at which
acidification would proceed.  These processes are not yet defined  on a regional
scale and, therefore, cannot presently be used in a definition  of  relative
sensitivity of regions to acidic deposition.

     In light of the above, it is clear that caution  must be used  in any effort
to use a single measure such as alkalinity to assess  the sensitivity of  surface
waters to acidic deposition because the actual  response of a given lake  or
stream is determined by numerous biogeochemical and hydrological factors of the
watershed plus chemical processes within water bodies.  Alkalinity is  certainly
the most readily available measure of the acid-neutralizing  capacity of  surface
waters.  Although alkalinity measurements do not completely  incorporate  the
influences of all factors into a definition  of  surface water sensitivity, they
do reflect the interactions of biogeochemical and hydrological  processes that
ultimately influence sensitivity.

     With this rationale, we approached the  problem of depicting the  likely
patterns of surface water sensitivity in the conterminous United States  by
synoptically analyzing spatial patterns of surface water alkalinity  as an
integrator of the various factors which determine sensitivity.   We accomplished
this by:  (1) assembling available alkalinity data on as many representative
surface waters as necessary and/or possible; (2) plotting these data on

                                      1

-------
large-scale maps;  and (3)  analyzing the patterns  of  the  values  of the  plotted
data for spatial  correlations with  other characteristics  such as land  use,
geology, and physiography.

     A national map compiled earlier (Omernik  and Powers,  1983) described the
general patterns  of surface water alkalinity in the  conterminous United
States.  By comparison,  the regional map presented here  is based on two  orders
of magnitude more data (for the Upper Midwest  portion  of  the national  map) and
depicts the spatial patterns of surface water  alkalinity  in greater detail and
at a greater resolution  than was possible in the  national  map.

     The alkalinity ranges of the five map units  were  chosen to reflect
potential sensitivity patterns on a regional scale,  as compared with the
broader ranges used for  the national map.  Although  it is  not possible to
define exact break points  between sensitive, moderately  sensitive, and insensi-
tive waters, it is generally agreed that waters  of total  alkalinity >  200 ueq/1
are relatively insensitive to acidic deposition.

     As was the case with  the national map, our  purpose  is to show what  range
of alkalinity one might  expect to find in most of the  surface waters most of
the time.  Relative to the national map, the regional  map provides more
detailed ancillary information on ranges of conditions,  significant apparent
regional and local relationships between alkalinity  and  macro-scale watershed
characteristics such as  land use and physiography, seasonal variations,  and
other factors.  This information in turn provides a  basis for understanding the
confidence with which predictions and estimations of potential  surface water
sensitivity might be made for the region, or parts of  the region.  Me  emphasize,
however, that the map and the ancillary information  are  not intended for making
precise predictions of sensitivity for individual water  bodies  or specific
locations.  Rather, this map and the other regional  maps  are  intended  to help
fill the urgent need to  understand the relative  potential  sensitivity  of
surface waters in different parts of the country in  order to  provide a national
perspective of the potential problem, provide  rationale  for selecting  geo-
graphic areas for more detailed studies, and allow more  accurate  regional
assessments of acid deposition impacts on aquatic resources.

MAP DEVELOPMENT

     The methods used to develop this map were similar to those used to  prepare
the alkalinity map of the New England and New  York Region (Omernik and Kinney,
1985), although the map scales were larger and amount  of data used for the
Upper Midwest Region was much greater.  The alkalinity data were  selected and
mapped according to several categories, with separate  designations given for
streams, lakes, and reservoirs, as well as to  sites  associated  with watersheds
of  less than 260 sq km (100 sq mi).  Each data point was scrutinized for
representativeness by keeping the watershed size consistent with  the relative
homogeneity of major watershed features thought  to influence  surface water
alkalinity such as physiography, vegetation type, and  land use.  In areas of
relative heterogeneity,  most of the data were  associated with small watersheds
less than  130 square kilometers.  Representative data  was imperative for
detecting  spatial patterns of alkalinity, possible correlations with patterns
of  other characteristics, and, ultimately, extrapolation of the data.  To

-------
include non-representative data from sites having large watersheds  of widely
differing characteristics, or data downstream from major industrial waste
discharges, would mask these spatial patterns.

     The map of the Upper Midwest Region,  similar to  the national map and the
New England/New York map, was based on patterns  of the actual  alkalinity values
and the apparent spatial  associations of these values with  aerial characteris-
tics that are believed to be driving or integrating factors affecting alkalin-
ity.  Driving factors, as used in this paper, refer to those that are generally
believed to directly affect alkalinity (e.g., geology and soils).   Integrating
factors, on the other hand, are considered those that tend  to  reflect combina-
tions of driving factors; for example, land use  and potential  natural vegeta-
tion reflect regional combinations (or an  integration) of driving factors such
as soils, land surface form, climate, and  geology.  We believe that the
importance of each of these factors, and the hierarchy of importance relative
to the combinations of factors, varies from one  region to another and even
within regions.  Clarifying these regional factors is a major  goal  of our
overall synoptic analyses.

     We acquired alkalinity data from a variety  of sources. As expected,
uniformity of coverage and temporal consistency  was lacking.  In general, the
amount of data we acquired was a function  of its availability  and,  more  impor-
tantly, the apparent complexity of regional patterns  of alkalinity.  In some
areas, after gathering and plotting a certain amount  of data,  the spatial
patterns became clear enough so that the addition of more data merely supported
the patterns already set.  In other areas, more  data, and/or analyses at
increasingly larger scales, were necessary to distinguish the  alkalinity
patterns.

     The most recent data (1979-1983) for  the Upper Midwest Region  were
obtained from STORET, an EPA computer-based water quality data storage and
retrieval system.  Data collected by the EPA, the University of Minnesota-
Duluth, and the Wisconsin Department of Natural  Resources (Glass et al., 1983)
and the Minnesota Pollution Control Agency (Twaroski  et al., 1983;  Thornton et
al., 1982; Heiskary and Thornton, 1983) from 1978 to 1983,  which had not been
entered into STORET at the time of our collection, were also used.  These data
were plotted on 1:250,000 or 1:500,000-scale topographic maps, with each site
represented by a small circle color-coded  to an  alkalinity class value; the
exact value and the water body type were noted beside the circle.   From STORET
data for the earlier period, 1960 to 1975, 108 lake and stream points (mostly
from the early 1970s) were determined to be suitable for our mapping purposes
and were plotted on overlays.  These three sources yielded approximately 1,648
values  (86% from lakes and 14% from streams) in the Upper Midwest  Region.  Of
these values, 56% represented one sample only, 8% were the mean of  two samples,
and 35% were the mean of three or more samples.

     For Michigan and Wisconsin, we relied heavily on earlier  (1960 to 1975)
data, because of a paucity of more recent  data and the relatively complex
alkalinity pattern.  In Michigan's Upper Peninsula, in addition to  306 points
from STORET and Glass et al. (1983), small-scale dot maps from Schneider  (1975)
showing the distribution of lakes by methyl orange alkalinity  class were used
to aid in the determination of the areal extent  of our alkalinity classes.  The

-------
patterns of these historical  data  indicated  the  location  of  clusters of  lakes
with less than 400 ueq/1  alkalinity.   By  superimposing  these patterns on our
1:500,000-scale maps containing  the more  recent  alkalinity data, and by noting
the apparent relationships of lake type to alkalinity,  lake  size to alkalinity,
lake vs. stream alkalinity, and  relevant  macro-scale  watershed characteristics,
we were able to interpolate and  estimate  areas of  specific alkalinity classes.
This process was also enhanced by  transferring and plotting  the data by alka-
linity class on 1:250,000-scale  topographic  maps.

     For Wisconsin, an abundance of historical alkalinity data was available
from that state's Department of  Natural Resources  (and  one of its predecessor
agencies, the Wisconsin Conservation  Department) in publications of "Surface
Water Resources" by county (Various Authors, 1960-1980).  These included maps
that showed the alkalinity classification for each lake and  stream sampled.
For the eighteen-county area delineated in the map inset, alkalinity values
were given for about 10,500 lakes  and 2,500  stream sites. On these large-scale
county maps, lake and stream sites were color-coded according to alkalinity
class, and the resultant patterns  indicated  where  boundaries might be drawn.
The delineation of our alkalinity  classes on the county maps was made after
noting lake types, lake sizes, stream courses, and physical  features from
1:250,000-scale topographic maps as well  as  consideration of other driving and
integrating factors.  In addition, comparisons were made  to  the spatial  patterns
of recent alkalinity data plotted  on  the  1:500,000-scale  Wisconsin map,  which
included 537 lake and stream points from  STORET  and Glass et al. (1983).

     The data for Minnesota used in our mapping  were  primarily of  recent
origin.  Approximately 805 lake  and stream  values  from  STORET and Glass  et al.
(1983) were plotted on 1:250,000-scale topographic maps by alkalinity class
(e.g., 1 = < 50 ueq/1, 2 = 50 to 100  ueq/1,  3 =  100 to  200 ueq/1, etc.).
Data listed in Twaroski et al. (1983), Thornton  et al.  (1982), and Heiskary and
Thornton (1983) were checked for any  lakes  not previously plotted or  listed in
the STORET or Glass et al. (1983)  data sets. An interpretive process similar
to that used for Wisconsin and Michigan,  considering  the  relevant driving  and
integrating factors, was also used for Minnesota.   Since  the relationship
between alkalinity and lake type and  lake size is  quite different  in  northeast
Minnesota than in other parts of the  Upper  Midwest Region, the final  class
boundaries there were based largely on the  spatial pattern of the  alkalinity
data itself, but also with some  considerations of  bedrock type,  land  surface
form, and land use.

     Spatial patterns of the alkalinity  values were the most important  factor
for delineating alkalinity map units  throughout  the Upper Midwest  Region.  Land
use and vegetation information were also  useful, but  in a more  general  way of
delineating areas where low alkalinity waters were unlikely. Low  alkalinity
waters are generally found only in predominantly forested or wetland  areas.
Wherever agriculture (including  grazing)  occurs, even as  a fairly  small  portion
of the land use mosaic, mean annual alkalinity  values are commonly well  over
400 ueq/1.  Geology maps were used to focus  our  attention and data  collection
efforts on areas of suspected sensitive  rock types.  However, final map unit
alignments were nearly always based  on patterns  of alkalinity values  or
physiographic features.  Locally,  physiographic  and hydrologic  features were
extremely helpful for map  unit delineations  (e.g., where strong  associations
were apparent between lake type and alkalinity  or  glacial feature  and low
alkalinity lake type).
                                      4

-------
DATA ASSESSMENT

     The use of historical  alkalinity  data  raises  questions  concerning the
comparability of older and  newer methods  of measuring  surface water alkalinity.
Methodologies vary even among recently collected data.  For  example,  of the
recent data (1978-1983) used for the Upper  Midwest map,  alkalinity values for
approximately 46% of the sites were determined  using the titration method of
Gran (1952), 22% were determined by single  endpoint titration  (potentiometric)
(American Public Health Association, 1980), 11% were determined  by double
endpoint titration (potentiometric) (American Public Health  Association, 1980),
and 21% were of unknown methodology.  While these  recent data were of great
importance, especially in Michigan  and Minnesota,  they only  represent about
one-tenth of all the data points used.  For the entire Upper Midwest  data set,
including the large amount  of historical  data from Wisconsin, roughly 88% of
the alkalinity values were  determined  by  colorimetric  methodologies (usually
methyl orange or methyl purple); only  5%  were determined using the method of
Gran (1952), 3% by single endpoint  titration (potentiometric)  (American Public
Health Association, 1980),  1% by double endpoint titration  (potentiometric)
(American Public Health Association, 1980), and 3% unknown.

     For low alkalinity waters, the most  commonly  used fixed endpoint proce-
dures (either potentiometric or colorimetric) often yield overestimates of
alkalinity (Dillon et al.,  1978; Zimmerman  and  Harvey, 1979-1980; Jeffries and
Zimmerman, 1980; National Research  Council  of Canada,  1981;  Henriksen, 1982;
Kramer and Tessier, 1982; Church, 1983).  Precision may  also be  significantly
less with colorimetric procedures because of uncertainty as  to the exact
endpoint (Kramer and Tessier, 1982; Church, 1983).  In contrast, the  double
endpoint procedure and the  procedure of Gran are unbiased and more precise for
low alkalinity waters (Gran, 1982;  American Public Health Association, 1980;
Church, 1983).

     When making our final  interpretations  of spatial  patterns of the data and
subsequent map unit delineations, we attempted  to  compensate for the  probable
bias introduced by selected analytical procedures.  If actual endpoint pH
values of the titrations had been known,  then quantitative procedures might
have been applied to correct for bias  (National Research Council of Canada,
1981; Henriksen, 1982; Kramer and Tessier,  1982; Church, 1983),  but because of
the lack of such information, adjustments were  not possible. However, a
comparison of values for a  large group of lakes in northern  Wisconsin where we
obtained recent values determined by Gran's titration, as well as earlier
values determined by colorimetric endpoint, indicated  a  pattern  of overestima-
tion by the colorimetric methods of about 30 or 60 ueq/1, depending on the
particular colorimetric method used (Figure 1).  Haines  and  Akielaszek (1983)
found a similar overestimation (32 ueq/1) in their comparisons of values for a
set of New England lakes.  Based on these comparisons, we adjusted the earlier
Wisconsin data by subtracting 30 ueq/1 (if  methyl  orange) or 60  ueq/1 (if
methyl purple) when we color-coded  the county maps. Elsewhere in the Upper
Midwest where representative sites  had borderline  or slightly above borderline
values between alkalinity classes (e.g.,  50, 100,  200, and 400 ueq/1) and where
the alkalinity methods had  been other  than  double  endpoint or Gran's  titration,
we assigned the respective  areas to the lower alkalinity class and drew the map
units accordingly.  However, in many cases  the  compensation  may  not have been

-------
                 700
                 600
             o   500
             in „
             -'
             0,0
                 400
                 300
                 200
                 100
o..

• . .


..ONEIDA CO
N « 123 ' >
..VtLAS CO-
N = 6« i ~
X = IITY +

t*

93
3*
                       100
                            200
                                 300
                                      400
                                            500
                                                 600
                                                      700
                                                           800
                          WISCONSIN CONSERVATION DEPARTMENT
                             Vilas Co. -I960 & 61 Methyl orange
                             Oneida Co. -1966 Methyl purple
Figure 1.  Single sample alkalinity values from two time periods for lakes  in
           Vilas and Oneida counties, Wisconsin.  Source:  Andrews and
           Threinen (1966), Black et al. (1963), Depository of Unpublished  Data.
enough to account for the bias due to methodology as suggested in the sources
cited above.  Hence, the areas in the lower alkalinity classes may actually  be
slightly larger than is shown; e.g., the actual areas shown as map unit #1
(< 50 ueq/1) may include some of the adjacent areas shown as map unit #2  (50 to
100 ueq/1), the areas shown as map unit #2 may include some of the adjacent
areas shown as map unit #3, etc.

     Another area of concern is the use of single sample values as representa-
tive data points.  Since the alkalinity of surface waters can fluctuate on a
daily, weekly, monthly, and annual basis, a single sample from a water  body  may
not seem to be representative of that lake or stream's average alkalinity.   We
believe, however, that for the purposes of this map, one-sample data points  are
sufficiently representative to be used in the assessment of spatial patterns
and delineation of the alkalinity class boundaries.  The comparison of  alka-
linity values for northern Wisconsin lakes determined using different labora-
tory methods and taken during different time periods (Figure 1) also serves  as
an illustration of the usefulness of single sample values for our purposes.

SEASONAL VARIATION

     Surface water alkalinity of lakes and streams is subject to seasonal  and
annual fluctuations due partially to climatic, meteorologic, and related
hydrologic events.  Seasonal variations of alkalinity concentrations in the
Upper Midwest Region appear to have distinct patterns, due especially to  runoff
from spring snowmelt.

-------
     Recent data showing complete month by month and seasonal  trends  for
surface water alkalinity in the Upper Midwest Region are scarce  for  lakes  and
for streams with small  watersheds.  However,  analysis of thirty-one  stream
sites across the region with long-term monthly data  showed  an  annual  pattern of
higher alkalinity values in winter;  a rapid drop in  spring,  with the  lowest
alkalinity values occurring in April; and then another rise with a peak  in
August (Figure 2).  Although this pattern is  based on data  from  mostly  higher
alkalinity streams, some with very large watersheds, a similar seasonal  pattern
seemed to occur in most streams throughout the region regardless of  alkalinity
level or watershed
difference between
                   size.  For the forty-one year-long data  sets,  the  average
                   the low value month and the mean  annual  value  of each
sampled stream was 52% (range 14-87%;  median 55;  standard deviation 14.8%).
Watershed and hydrological characteristics vary greatly from stream to  stream,
of course, and have different influences  on seasonal  alkalinity trends.
Thornton et al. (1982) found that streams in Minnesota which are  highly depen-
dent on runoff tend to have greater fluctuations  in  flow and greater  fluctua-
tions in stream chemistry, while streams  with large  surface water storage  areas
will have lesser fluctuations in flow  and small  fluctuations in stream  chemistry,
         3000-
          2500-
          2000-
          1500-
          1000-
          500-
                 JFMAMJJASOND

Figure 2.  Seasonal  variations in alkalinity (ueq/1)  at  thirty-one  stream  sites
           in the Upper Midwest Region (twenty-two  in Minnesota,  four  in
           Michigan, and five in Wisconsin).  Forty-one  year-long data sets  are
           represented.  Source:  STORE! 1979-1981.


     Declines in alkalinity concentrations  during spring snowmelt and  in
conjunction with precipitation events  can be dramatic for streams,  but for
lakes the seasonal  fluctuations appear to be less extreme.   Low  alkalinity
lakes also appear to have lower spring values,  but  the data  are too  limited  to
adequately assess seasonal  fluctuations and trends.   Most of the  low alkalinity
lakes are represented by only one or a few  alkalinity samples pe.r year.  A
comparison of spring, summer, and fall alkalinity values for twelve  northern
Minnesota lakes revealed that, on the  average,  the  spring alkalinities were
                                     7

-------
only 14% lower than the mean values  (Heiskary  and  Thornton,  1983).  A greater
percentage difference might have  resulted  had  values  of  the  lowest month only
(April or May) been compared to mean annual  values that  included  higher winter
values as well.  However,  from our synoptic  analyses  of  alkalinity values from
thousands of lakes in the  Upper Midwest, seasonal  fluctuations  do not appear to
vary greatly outside the range of values in  the  classes  designated on our
alkalinity map.

REGIONAL PATTERNS

     The Upper Midwest Region exhibits  great diversity  and spatial hetero-
geneity in its patterns of surface water alkalinity.  Several factors contrib-
ute to the complexity of these patterns, especially lake types, sizes, and
their hydrologic characteristics; bedrock  geology; land  surface forms; and soil
characteristics.  The formation of these  lakes,  surface  features, and soils
reflect the region's glacial history.   As  a  result of glacial action, this
region contains one of the densest concentration of lakes in the  world.

     While nearly all of the streams and a great majority of the  lakes in the
Upper Midwest Region have  high mean  annual alkalinity values (> 400 ueq/1),
there are several areas where a considerable number of  lakes have low alkalin-
ity values (< 50 ueq/1).  Although relatively  large in  number,  the lakes in the
lowest alkalinity class tend to be small in  size and  comprise only a small
percent of the region's total surface water area (Table  1).  The  lakes of
average alkalinity values  < 200 ueq/1 comprise only 6.6% of  the region's
surface water area.  Clusters of  lakes  with  alkalinities < 50 ueq/1 are found
in portions of Vilas, Oneida, Lincoln,  and Langlade counties of Wisconsin, as
well as Gogebic, Iron, Alger, and Chippewa counties of  Michigan.

     The map depicts generalized  patterns  of surface  water alkalinity and is
intended to show the range of values within which  one might  expect the mean
annual alkalinities for most of the  surface waters for  each  classified area.
            Table 1.  Area (hectares)  in lakes by alkalinity  class.*
Alkalinity
Class
< 50 ueq/1
50 - 100
100 - 200
Wisconsin
6,400
5,595
17,843
Minnesota
0
13,982
68,423
Michigan
1,686
1,426
8,948
Total
8,086 (0.4%)
21,003 (1.1%)
95,214 (5.1%)
  < 200
29,838
82,405
12,060
124,303 (6.6%)
Total lake area in the Upper Midwest Region (Northern
Wisconsin, Northeastern Minnesota, and Northern Michigan)    1,868,799 hectares
* Lakes > 6 hectares (15 acres).  Based on extent of surface water appearing
  on 1:250,000-scale USGS topographic maps.
                                       8

-------
Although some areas are predominantly of low surface water alkalinity, other
areas are characterized by wide ranges in alkalinity and contain  various  lake
sizes and lake types that are in close proximity to one another.  Thus, care
should be taken in interpreting the mapped patterns.  Even in  areas classified
as containing waters of predominantly > 400 ueq/1,  there may  be a few  lakes,
but probably no streams, that are of low alkalinity.

     Examination of the physical and chemical  characteristics  of  streams  and
lakes suggests three loosely delineated subregions  within the  Upper Midwest
Region.  Distinctions can be made between:  (1)  most of northern  Wisconsin and
east central Minnesota; (2) northeast Minnesota  comprising Cook,  Lake, and
St. Louis counties; and (3) the major portion  of Michigan's Upper Peninsula and
the northern portions of Bayfield, Ashland, and  Iron counties  in  Wisconsin
(Figure 3).  These subregions, while individually containing  variations in
surface water alkalinity, can be broadly differentiated by certain physical
characteristics that influence the alkalinity  patterns.  Factors  that  directly
affect alkalinity such as geology, land surface  form, and soils,  show  subtle
variations in the different subregions.  Integrating factors  such as potential
natural vegetation, or especially land use, which generally correlated with
alkalinity in some other parts of the nation (Omernik and Powers, 1983),  were
of less significance for determining alkalinity  patterns in the Upper  Midwest
Region.
                           Subregion Boundary
               Figure 3.  Subregions  of the Upper Midwest  Region,
Wisconsin-East Central  Minnesota

     In the northern Wisconsin-east central  Minnesota subregion,  the  differ-
ences in lake types [without inlets or outlets  (Type A),  with  inlets  or  outlets
(Type B), spring-fed, acid bog,  alkaline bog],  and the large number of  lakes  of
each type, contribute to the extreme heterogeneity of surface  water alkalinity.
While most of the lakes without  inlets or outlets  are of  low alkalinity, the
streams, spring-fed lakes, and lakes with inlets  or outlets  are generally of

-------
high alkalinity.  The continental  glaciation  left  a  relatively thick layer of
glacial  drift over this  part  of  the  Upper  Midwest, and most of the lakes occur
in pitted outwash and end moraines.   These lakes are usually situated well
above bedrock and are relatively shallow.   For  example, in Oneida County,
Wisconsin, 88% of the lakes  are  less  than  7.6 meters deep (Andrews and
Threinen, 1966).  Apparently,  for  much  of  this  subregion, bedrock has very
little influence on lake chemistry and  lake morphology.

     The major land forms in  this  subregion include  outwash plains, ground
moraines, end moraines,  lacustrine plains, and  drumlins.  In the Minnesota
portion, in particular Carlton,  Crow Wing, and  Itasca counties, Twaroski et al.
(1984) concluded that lake forms seem to exert  a greater influence on lake
chemistry and lake morphology  than does bedrock.   They found that, based on
slope and soils, moraine areas are most likely  to  contain low alkalinity lakes.
These lakes are generally small, with no inlets; are located in interfluvial
areas; are surrounded by steep slopes and  soils that create a quick release of
water to the lake; and receive no  groundwater inflow (Twaroski et al.,  1984).
While they found that lakes  tended to have high alkalinities in outwash plains
in east  central Minnesota, many  of those portions  of Wisconsin within our map
unit #1  (< 50 ueq/1) and map  unit  #2 (50-100  ueq/1)  are found in outwash plains
and in end moraines.  The general  difference  between moraines or outwash plains
appears  to be of less significance than more  specific factors that influence
lake alkalinity in this  subregion.  Of  more importance seem to be groundwater
contact, lake size, watershed size,  stream inflow  and outflow, as well  as the
texture and mineralogy of the soil and  the local relief.  High alkalinity lakes
tend to be large, with significant inlets  and outlets, with large watersheds,
and with substantial groundwater contact.   A  key factor affecting the alkalin-
ity of lakes in this subregion is  hydrology,  or the  route and residence time of
water in the watershed (Eilers et  al.,  1983;  Twaroski et al., 1984).  The
relationship of groundwater to lakes is of major importance and helps to
explain why high alkalinity  lakes  are found in  close proximity to low alkalin-
ity lakes, but it is the least known aspect of  lake  hydrology (Winters, 1977).

     Several areas of surface waters with  mean  annual alkalinity < 50 ueq/1 are
found in this subregion, centered  around Vilas  and Oneida counties, Wisconsin.
Including all of the low alkalinity  areas  in  Wisconsin, as well as the  area in
southern Gogebic and Iron counties,  Michigan, this subregion has roughly 7,250
hectares of lakes (> 6 hectares) that fall within  the map unit #1 boundaries.
These areas predominantly contain  clusters of low  alkalinity lakes without
inlets or outlets (Type A).   Although the  number of  lakes in these areas is
generally large, the total area of surface water is  relatively low since Type A
lakes tend to be small in size.  For example, of the 1,327 lakes in Vilas
County,  Wisconsin, 776 (59%)  are Type A lakes of less than eight hectares in
size, but these account for only 4%  of  the county's  lake area (Table 2).  On
the other hand, only 88 (7%)  of the  lakes  in  the county are Type B lakes or
spring lakes greater than 80 hectares,  yet these types  comprise 63% of  the
county's lake area.  Thus, for Vilas County,  the greatest extent of lake area
is in the higher alkalinity classes, and the  average (uncorrected methyl
orange) alkalinity of 298 ueq/1  (Black  et  al.,  1963) is skewed downward by the
large number of low alkalinity but small Type A lakes.
                                    10

-------
Table 2.  Characteristics of lakes in Vilas  County,  Wisconsin.   Alkalinity
          values were determined by single endpoint  titration  (colorimetric --
          methyl orange) and, although reported in mg/1,  have  been  converted to
          ueq/1 in the table.  Source:  Black et al. (1963).
Size Class
Hectares
0-7
8-23
24-39
40-80
81-201
202-404
405-808
809
Total
Lake Type A
(Without inlets or outlets)
*
776
145
35
32
13
6
1
-
1008
Mean
Size
(ha)
2.0
13.6
31.1
55.2
117.2
285.9
-
-
10.1
Total
Hectares
1554 (4%J
1969 (5%)
1089 (3%)
1767 (5%)
1523 (4%)
1716 (5%)
582 (1%)
-
10,200 (27%)
Samplod lakoi.
Mean
Alk
174
136
172
158
386
332
40
-
170
*
493
145
35
31
12
6
1
-
723
Lake Type B
(With inlets or outlets)
#
31
24
19
27
38
26
9
4
178
Mean
Size
(ha.)
2.3
15.1
30.6
58.3
127 8
286.9
504.4
1299.6
138.5
Total
Hectares
72
362 (1%)
581 (2%)
1574 (4%)
4858 (13%)
7461 (20%)
4540 (12%)
5198 (14%)
24,646(66%)
Sampled lakes
Mean
Alk.
449
546
640
678
756
762
830
700
664
*
23
22
19
27
38
26
9
4
168
Spring Lakes
*
97
13
12
8
9
2
-
-
141
Mean
Size
(ha.)
1.6
15.6
32.3
60.9
108.5
210.6
-
-
18.7
Total
Hectares
158 -
203 -
388 (1%)
487 (1%)
977 (3%)
421 (1%)
-
-
2.634 (7%)
Sampled lakes
Mean
Alk.
693
626
988
772
902
720
-
-
755
*
27
12
8
8
9
2
-
-
66
     Streams comprise a very small  fraction  of the surface water  area  in  much
of this subregion, and stream alkalinity is, on the average,  much higher  than
that found in lakes.  In Vilas County, Wisconsin,  streams  comprise only 1.4% of
that county's total surface water area, and  most streams  are  small,  reflecting
the general headwater location of the county.  The mean methyl  orange  alkalin-
ity (unconnected) for the 151 streams in Vilas County was  about 794 ueq/1,  with
a range of 130 to 1840 ueq/1 (Black et al.,  1963).  Only  14 streams  (9%)  were
determined to have alkalinities < 400 ueq/1, and only 5 streams (3%) had
alkalinity values < 200 ueq/1.  For much of  this subregion, stream alkalinity
often appears to be twice as high as the mean alkalinity  values of nearby
lakes.  In general, streams have higher alkalinity than lakes as  a result of
having larger drainage areas and greater contact with buffering materials.

     The largest area of lakes with alkalinities < 50 ueq/1 is  located in
Oneida County, Wisconsin, and contains 109 named Type A lakes ranging  from 0.8
hectares to 425 hectares in size.  The mean  size of these  109 lakes  is 27
hectares, ar>d the median size is 12 hectares.  Most of this area  occurs in  a
pitted glacial outwash plain, with  sandy acidic soils of  low  fertility that
typify much of Oneida County.  Frequency distributions for recent and  histor-
ical data sets of thirty-seven representative lakes in this map unit #1 area
give a general idea of:  (1) the proportion  of lakes within the lowest map unit
areas of this subregion that have mean annual alkalinity  values within the <  50
ueq/1 range; (2) the proportion that have higher values;  and  (3)  some  notion of
                                     11

-------
central tendency.  These frequency distributions also illustrate the  differ-
ences in alkalinity values due to the different methodologies, and support  our
adjustment of 60 ueq/1 for the data determined by the colorimetric methyl
purple methodology (Figure 4).  In this comparison, 58 ueq/1 was the  difference
between the mean of the methyl purple alkalinity data and the mean of the data
determined by Gran's titration.
    100
    so
    60
    40
    20
N = 37

mean = 63

median = 36

range = -9 - 368

stand, dev. = 75
100-
80-
cfl
O
£ 6°
0 40-
s«
20-








N - 37
mean =121
median = 100
range =* 40 - 420
stand, dev. = 83



     <0   50  100  150  200  400
            Alkalinity   Meq/l
              Gran titration
                              0    50  100  150  200  400  >400
                                    Alkalinity    neq/\
                                 Colorimetric — methyl purple
Figure 4.  Frequency distributions of thirty-seven  representative  lakes  in  an
           Oneida County, Wisconsin, map unit #1 area.  Source:  Glass et al.
           (1983), Andrews and Threinen  (1966).


Northeast Minnesota

     The northeast Minnesota subregion includes Cook, Lake,  and  northeastern
St. Louis counties, and exhibits considerably more  homogeneity in  surface water
alkalinity than the other subregions.  Lakes of various types and  sizes, as
well as streams, tend to have more similar alkalinity values.  The processes of
glacial erosion and glacial deposition produced an  extremely high  density of
lakes and extent of area in lakes in this subregion.

     A wide variety of bedrock types is  found in this subregion, and because
the soil and glacial till are relatively shallow, bedrock  appears  to have a
greater influence on lake morphology and lake chemistry.   In general,  the
gabbros, granites, and iron formations contain lakes with  relatively lower  mean
alkalinity values, mostly less than 200  ueq/1 (Twaroski et al.,  1984).   These
formations are non-calcareous and resistant to weathering, with  thin overlying
soils that provide little buffering capacity to the lakes.  The  slates  and
greenstone formations contain lakes of higher alkalinity,  mostly 400 ueq/1  or
greater, and, although generally non-calcareous, have some veins of calcite
that may greatly influence  lake sensitivity  (Heiskary and  Thornton, 1983).
Some calcareous soils are found in the Ely greenstone northeast  of Ely,  as  well
as  in the large area west of St. Louis County that  corresponds to  deposits  from
glacial Lake Agassiz.  In addition, the  slates and  greenstone have faults and
fractures which may supply  groundwater to these areas of higher  alkalinity
(Twaroski et al., 1984).  For most of this subregion, however, groundwater
                                     12

-------
resources are very limited due to the shallow drift and Precambrian crystalline
bedrock types and do not affect alkalinity as in Wisconsin or central
Minnesota.

     A considerable extent of map unit #2 alkalinity (50-100 ueq/1) occurs in
the extreme northeastern part of Minnesota.  We estimate that approximately
14,000 hectares in lakes fall within the three areas designated as map unit #2
(of lakes greater than 6 hectares).  These three areas are based on data from
forty lakes, and were delineated primarily by noting the spatial pattern of
these values as well as surrounding lake alkalinity values.  Approximately 70%
of these forty lakes have alkalinity values less than 100 ueq/1 (Figure 5).
100-


 80-


 60-
                        tft
                        a>
                                        N = 40

                                        mean =93

                                        medians 94
                        
-------
bedrock geoprovince.   Small  amounts  of  limestone,  veins  of calcite, or ancient
glacial lake deposits  may  increase the  buffering of  otherwise sensitive
bedrock.  We found that for  most  of  the Upper  Midwest  Region, bedrock by itself
was not a good indicator of  surface  water  alkalinity.

Upper Michigan

     The third subregion --  most  of  the Upper  Peninsula  of Michigan, and the
extreme northwest portion  of Wisconsin  in  Bayfield,  Ashland, and  Iron counties
-- is physically more  heterogeneous,  with  fewer  lakes  than the other subregions,
and wide variations in surface water alkalinity.   Most of the lakes occur in
glacial outwash plains and moraines,  although  a  few  are  found on  bedrock
outcrops and on ancient lacustrine deposits.

     The eastern part  of the peninsula  varies  from smooth plains  to irregular
plains with hills.  This area is  characterized by  sandy  and silty surficial
features from the predominant ancient glacial  lake bed deposits,  with a mixture
of alkalinity values.   In  the southeast part of  the  peninsula, some lakes are
located on bedrock, but the  presence of limestone  results in high surface water
alkalinities.  The areas of  lowest surface water alkalinity appear to be
located in areas of glacial  till, and the  chemical and physical nature of this
till, along with the amount  of groundwater contact,  drainage basin size, and
lake type, greatly influence the  degree of buffering capacity.  The streams and
lakes in these lower alkalinity areas are  also located in the uppermost reaches
of tne watersheds, where topographic watersheds  are  even roughly  definable.
[Surface and subsurface watersheds frequently  are  difficult or impossible to
define, particularly in areas of  continental glacial topography (Hughes and
Omernik, 1981)].  Many of  the low alkalinity lakes are without outlets.  For
the two areas within map unit #1  (<  50  ueq/1)  in the eastern portion of the
Upper Peninsula there  are  recent  alkalinity values for four lakes and two
streams (Table 3).

     In the western half of  the Upper Peninsula, most  of the lakes occur in
glacial till, and, as  in Wisconsin,  alkalinity values  appear especially related
to lake and watershed  size,  lake  type,  and the amount  of groundwater contact.


Table 3.  Alkalinity (ueq/1) values  in  map unit  #1 in  the Upper Michigan
          Subregion.  Source:  Glass et al. (1983).  Methodology:  Gran titra-
          tion.


                Mean:                 -2.7  ueq/1

                Median:              -8.0

                Range:               -36 to 34

                Standard Deviation:   29

                List of Values:      -36,  -27, -15,  -1,  29, 34


                                     14

-------
A few lakes are located on bedrock in Baraga and Marquette counties,  and  these
tend to be moderately low in alkalinity (100-200 ueq/1).

Summa ry

     Our map of total alkalinity of surface waters  in  the Upper  Midwest Region
of the United States illustrates the general patterns  of  the  relative potential
sensitivity of surface waters to acidic input in that  region.  The map was
developed through analysis of the spatial  patterns  of  alkalinity values from
over 14,000 representative lakes and streams, as well  as  through determination
of apparent spatial  associations between these data and  various  watershed
characteristics believed to be causal.  We found alkalinity patterns  in the
region to be extremely varied and complex.  Many lakes in the  region  have
relatively low alkalinities, but they tend to be small in size,  comprise  a
small percent of surface water area, and occur in relatively  small  clusters.
Most streams exhibit high alkalinities.  In northeastern  Minnesota,  however,
there are areas of relatively low alkalinity in which  all  lakes  regardless of
size, and even streams, have similar values.
                                    15

-------
                                Literature  Cited

American Public Health Association.   1980.   Standard methods for the exami-
     nation of water and wastewater.   Fifteenth Edition.  Washington, D.C.

Andrews, L. M., and C. W. Threinen.   1966.   Surface water resources of Oneida
     County.  Wisconsin Conservation  Department.  Madison, Wisconsin.

Black, J. J., L. M. Andrews,  and C. W.  Threinen.  1963.  Surface water
     resources of Vilas County.   Wisconsin  Conservation Department.  Madison,
     Wisconsin.

Church, M. R.  1983.  The acidic deposition  phenomenon and its effects:  criti-
     cal assessment review papers,  Volume II.  Effects Sciences, Chapter 4 --
     Effects on Aquatic Chemistry.

Depository of Unpublished Data,  CISTI,  National Research Council of Canada,
     Ottawa, Canada K1A OS2.   (Data for Eilers et al., 1983).

Dillon, P. J., D. S. Jeffries, W. Snyder, R. Reid, N. D. Yan, D. Evans, J.
     Moss, and W.  A. Scheider.   1978.   Acid precipitation in south-central
     Ontario:  recent observations.   J. Fish. Res. Board Can. 35:809-815.

Eilers, J. M., G. E. Glass, K. E. Webster,  and J. A. Rogalla.  1983.  Hydro-
     logic control of lake susceptibility to acidification.  Can. J. Fish.
     Aquat. Sci. 40:1896-1904.

Environmental Studies Board,  National  Research Council.  1984.  Acid deposi-
     tion:  processes of lake acidification.  National Academy Press.
     Washington, D.C.

Galloway, J. N., and E. B. Cowling.   1978.   The effects of precipitation on
     aquatic and terrestrial  ecosystems:  a  proposed precipitation chemistry
     network.  J. Air Pollut. Control  Assoc. 28(3):229-235.

Glass, G. E., L. Heinis, J. Regalia,  J. Use, L. E. Anderson, J. Eilers, B.
     Liukkonen, P. Johnston,  J.  Sorensen, S. Tongen, G. Gustafson, F. Simmons,
     0. Loucks, and G. Rapp.   1983.   Documentation and quality assurance
     summary for alkalinity,  hydrogen ion (pH), and specific conductivity
     measurements of water quality  from selected areas of Minnesota, Wisconsin,
     Michigan, and Ontario.  ERL-DUL-523.   Environmental Research Laboratory --
     Duluth, U.S. Environmental  Protection  Agency, Duluth, Minnesota.

Gran, G.  1952.  Determination of the equivalence point in potentiometric
     titrations.  Part II.  Analyst 77:661-671.

                                     17

-------
Haines, T. A.,  and J.  J.  Akielaszek.   1983.  A  regional survey of chemistry of
     headwater  lakes and  streams  in New  England:  vulnerability to acidifica-
     tion.  FWS/OBS-80/40.15.   Fish and  Wildlife Service, National Power
     Development Group.   U.S.  Department  of the Interior.  Kearneysville, West
     Virginia.

Heiskary, S. A., and J.  D.  Thornton.   1983.  Acid rain sensitivity:  a study of
     contributing factors in  remote northeastern Minnesota lakes.  Minnesota
     Pollution  Control Agency.

Hendrey, G. R., J. N.  Galloway, S. A.  Norton, C. L.  Schofield, P. W. Shaffer,
     and D. A.   Burns.  1980.   Geological  and hydrochemical sensitivity of the
     eastern United States  to  acid precipitation.  EPA-600/3-80-024.  Corvallis
     Environmental Research Laboratory.   U.S. Environmental Protection Agency,
     Corvallis, Oregon.

Henriksen, A.  1982.   Alkalinity  and Acid Precipitation Research.  Vatten.
     38:83-85.

Hughes, R. M.,  and J.  M.  Omernik. 1981.   Use and misuse of the terms watershed
     and stream order.  In  Proceedings of the Warmwater Streams Symposium, pp.
     320-326.  American  Fisheries Society, Southern  Division.

Interagency Task Force on Acid Precipitation.   1982.  National acid precip-
     itation assessment  plan.   Washington, D.C.

Jeffries, D. S., and A.  P.  Zimmerman.  1980.  Comments on the analysis and
     sampling of low conductivity natural  waters for alkalinity.  Can. J. Fish.
     Aquat. Sci. 37:901-902.

Kramer, J., and A. Tessier.  1982.  Acidification of aquatic systems:  a
     critique of chemical approaches.  Environ. Sci.  Technol. 16:606A-615A.

Likens, G. E.,  R. G. Wright,  J.  N. Galloway, and T.  J. Butler.  1979.  Acid
     rain.  Sci. Am. 241(4):43-51.

McFee, W. W.  1980.  Sensitivity  of  soil  regions to  acid precipitation.
     EPA-600/3-80-013.  Corvallis Environmental Research Laboratory,  U.S.
     Environmental Protection  Agency,  Corvallis, Oregon.

National Atmospheric Deposition Program.   1982. Distribution of  surface waters
     sensitive  to acidic precipitation:   a state level atlas.  NADP Tech. Rept.
     No.  IV, S. A. Norton,  editor.   Department  of Geological Sciences, Univ. of
     Maine, Orono, Maine.

National Research Council of Canada.   1981.  Acidification in the Canadian
     aquatic environment.  NRCC Publication  No. 18475.

Omernik, J. M., and A. J. Kinney.  1985.   Total alkalinity of surface waters:
     a map of the New  England  and New  York region.   Corvallis Environmental
     Research Laboratory, U.S. Environmental Protection Agency, Corvallis,
     Oregon.

                                    18

-------
Omernik, J. M., and C.  F.  Powers.   1983.   Total  alkalinity of surface waters --
     a national map. Ann.  Assoc.  Am.  Geog.  73(1):133-136.

Schneider, J. C.  1975.   Typology  and  fisheries  potential of Michigan lakes.
     Michigan Academician.   Papers  of  the  Michigan Academy of Science, Arts,
     and Letters 8(l):59-84.

Shilts, W. W.  1981. Sensitivity  of bedrock  to  acid precipitation:  modifi-
     cation by glacial  process.  Geol.  Surv.  Can. Paper. 81-14.

Thornton, J. D., S. A.  Heiskary, R. D.  Payer,  and J. Matta.  1982.  Acid
     precipitation in Minnesota.   Report to  the  Legislative Commission on
     Minnesota Resources.   Minnesota Pollution Control Agency, Minnesota
     Department of Natural  Resources,  and  Minnesota Department of Health.

Twaroski, C. J., J. D.  Thornton, S. A.  Heiskary, and D. Anderson.  1983.
     Information utilized  in  a  preliminary mapping of areas sensitive to acid
     deposition in Minnesota  (draft).   Minnesota Pollution Control Agency,
     Minnesota Land Management  Information Center.

Twaroski, C. J., J. D.  Thornton, S. A.  Heiskary, and D. Anderson.  1984.
     Aquatic, terrestrial,  and  peatland ecosystems in Minnesota considered
     sensitive or potentially sensitive to acid  deposition.  Minnesota Pollu-
     tion Control Agency,  Minnesota Land Management Information Center.

Various Authors.  1960-1980.   Surface  water  resources.  (Series by county.)
     Wisconsin Department  of  Natural Resources  (formerly Wisconsin Conservation
     Department).  Madison, Wisconsin.

Winters, T. C.  1977.  Classification  of the  hydrologic settings of lakes in
     the north central  United States.   Water  Resour. Res. 13(4):753-767.

Zimmerman, A. P., and H.  H. Harvey.  1979-1980.  Sensitivity to acidification
     of waters of Ontario  and neighboring  states.  Final Report for Ontario
     Hydro.  Univ. of Toronto.
                                    19

-------
I!U  Hi
?S3S££5£
                                                                                                    H
                                                                                                    O
                                                                                                    H
                                                                                                 C *
                                                                                                 5 >
                                                                                                 S £
                                                                                              » < CD
                                                                                              i   co
                                                                                              o   ~
                                                                                              •   3D
                                                                                              a   CD
                                                                                              m   <0

                                                                                              s   §
o
Tl

CO
c
33
-n
>
O
rn
                                                                                                    m
                                                                                                    33
                                                                                                    co

-------