Environmental Management
DOI 10.1007/s00267-010-9533-z
The Impact of Future Land Use Scenarios on Runoff Volumes
in the Muskegon River Watershed
Deepak K. Ray • Jonah M. Duckles •
Bryan C. Pijanowski
Received: 31 March 2009 / Accepted: 5 July 2010
Ó Springer Science+Business Media, LLC 2010
Abstract In this article we compared the response of
surface water runoff to a storm event for different rates of
urbanization, reforestation and riparian buffer setbacks
across forty subwatersheds of the Muskegon River
Watershed located in Michigan, USA. We also made these
comparisons for several forecasted and one historical land
use scenarios (over 140 years). Future land use scenarios to
2040 for forest regrowth, urbanization rates and stream
setbacks were developed using the Land Transformation
Model (LTM). Historical land use information, from 1900
at 5-year time step intervals, was created using a Backcast
land use change model configured using artificial neural
network and driven by agriculture and housing census
information. We show that (1) controlling the rate of
development is the most effective policy option to reduce
runoff; (2) establishing setbacks along the mainstem are
not as effective as controlling urban growth; (3) refores-
tation can abate some of the runoff effects from urban
growth but not all; (4) land use patterns of the 1970s
produced the least amount of runoff in most cases in the
Muskegon River Watershed when compared to land use
maps from 1900 to 2040; and, (5) future land use patterns
here not always lead to increased (worse) runoff than the
past. We found that while ten of the subwatersheds con-
tained futures that were worse than any past land use
D. K. Ray  J. M. Duckles  B. C. Pijanowski (&)
Department of Forestry and Natural Resources, Purdue
University, West Lafayette, IN 47906, USA
e-mail: bpijanow@purdue.edu
Present Address:
D. K. Ray
Institute on Environment, University of Minnesota-Twin
Cities, Saint Paul, MN 55108, USA
configuration, twenty-five (62.5%) of the subwatersheds
produced the greatest amount of runoff in 1900, shortly
after the entire watershed was clear-cut. One third (14/40)
of the subwatersheds contained the minimum amount of
runoff in the 1960s and 1970s, a period when forest
amounts were greatest and urban amounts relatively small.
Keywords Land use change modeling  Surface water
runoff  Policy Impacts
Introduction
The distribution of land use influences the hydrology of
watersheds across a variety of spatial and temporal scales
(Leopold 1968; Harbor 1994; Mölders and Rühaack 2002;
Pijanowski and others 2002a; Tang and others 2005a; Tang
and others 2005b). Agricultural and urban dominated
landscapes increase runoff compared to a landscape that is
in its natural state (Leopold 1994; USEPA 2000; Tong and
Chen 2002; NRC 2008). The increased runoff carries with
it nutrients (e.g., nitrogen and phosphorus) from farm fields
(Carpenter and others 1998) and contaminants from urban
activities (USEPA 1983; Wayland and others 2002;
Wayland and others 2003; Fitzpatrick and others 2007).
Increased runoff can also lead to more frequent or intense
floods (Bhaduri and others 2000; Booth and others 2004)
and alter the apportionment of water to wetlands and
groundwater (Mitsch and Gosselink 1993). Nutrient rich
runoff from agriculture is considered to be the primary
cause of eutrophication of surface water bodies and the
expansion of dead zones (Diaz and Rosenberg 2008),
which negatively impacts sensitive biological organisms,
such as fish and aquatic invertebrates (Wang and others
1997; Cifaldi and others 2003; Allan 2004).
123

Several best management practices exist to protect sur-
face water bodies, such as rivers and lakes, from urban and
agricultural runoff (O’Callaghan 1996; Diebel and others
2008). These best management practices include control-
ling the rate of development (cf. Paul and Meyer 2001),
protecting highly erodible soils or areas with significant
hydrologic impact from agricultural use (e.g. Bradford and
others 1987) and the creation of setbacks or buffers (Bren
1998) that prevents urban development or agriculture
occurring within the riparian zones of rivers and streams.
Controlling the rate of urban development however is
challenging. At current rates of urbanization, a doubling in
the amount of urban area will occur in 20 years for most of
the United States (Alig and others 2004). In the Upper
Midwest of the United States urban development has been
rapid over the last 10–20 years (Pijanowski 2006; Wolter
and others 2006), with many areas in this region increasing
their urban footprints at an average of 4.4% per year
(Pijanowski unpub). In Michigan’s Muskegon River
Watershed, the focus study area of this article, Tang and
others (2005a) found that runoff and non-point source
pollution was highly dependent on urbanization rates; a
doubling of urban would increase runoff by 1.5 times over
a baseline land use distribution (assumed as 1998). Thus,
one way to control the future runoff rates would be to slow
urban growth rates.
Increasing forestland can also reduce runoff rates. In the
Muskegon River Watershed, nearly the same amount of
forestland was added to the watershed from 1978 to 1998
as was new urban, partially as a result of the implemen-
tation of USDA’s Conservation Reserve Program (CRP) in
1985 to reduce sediment loading to streams by converting
high erosion cropland to natural vegetation. However,
while watersheds across the United States continue to
urbanize, the relative effect of increasing forestland that
serve as a mitigation strategy towards reducing the runoff
from increased urbanization is unknown.
The creation of setbacks on rivers and streams has also
been argued to protect surface water from non-point source
pollution (Bren 2000) by reducing surface water runoff
rates. Traditionally, riparian setback regulations establish
minimum distances that buildings or certain agricultural
activities can occur from surface waters. Establishing set-
backs or buffers along streams is also known to decrease
bank erosion (Shields and others 1995). However, there is
considerable debate as to the appropriate size and config-
uration that these buffers need to be in a watershed
(Dosskey 2002) to control runoff. It is accepted by many
researchers that buffer sizes and configurations are likely to
be influenced by topography (e.g., slope), soil type and
land use; there is likely to be tremendous variation in slope,
soil type and land use across regional watersheds making
the establishment of standard setback distances unpractical.
123
Environmental Management
For example, in the Muskegon River Watershed, the mid-
portion of the Muskegon River mainstem is highly chan-
nelized but upper and lower portions of the watershed flow
over relatively flat landscapes. It would be important to
know the effect of using a fixed size buffer as a policy
option in reducing runoff volumes.
Walter and others (2009) argue that buffer configura-
tions should be made on the basis of ‘‘the potential for an
area to generate runoff’’. Such an assessment can be made
using simple runoff models (Walter and others 2009) such
as the Soil Conservation Service (SCS) Curve-Number
(CN) equation (USDA 1986). The CN equation estimates
runoff volume from a ratio of effective rainfall and maxi-
mum soil storage capacity. It has also been argued (e.g.,
Tomer and Burkart 2003; Dosskey 2002) that areas that are
hydrologically sensitive, such as locations in watersheds
where runoff would impact streams and rivers directly
through high hydrologic connectivity, should be targeted
for setback regulations.
Given the myriad of management options available to
control runoff, which are most important? Addressing these
factors in land use planning and natural resource manage-
ment has been successfully explored through the use of
coupled land use-hydrologic models (Harbor 1994). When
applied in a policy context, these coupled models can be
used to forecast land use change or best management
options into the future using policy scenarios as one
component of the land change model (e.g., Koomen and
Stillwell 2007) and hydrology examined using any number
of hydrologic models.
Assessing the relative impacts of different policies or the
implementation of best management practices on future
water quality and runoff rates will require a benchmark
with which to compare. These benchmarks could involve
comparing output of coupled models to simulations of
current conditions (e.g., Tang and others 2005a) or optimal
configurations of land use in a watershed (Tang and others
2005b; Chaubey and others 2008). Historical trends offer a
third means of generating baselines for comparing future
projections of runoff due to changes in land use patterns.
Historical comparisons could be very useful as many areas
of the world have undergone significant changes in land
use. Via such comparison, we can answer the question
whether the surface runoff volumes projected for future
land use changes would be any worse than those that have
occurred in the past.
This article presents the results of coupled land use-
runoff simulations for Michigan’s Muskegon River
Watershed designed to address the following research
questions:
1. What is the impact of afforestation patterns on runoff?
Does the presence of new forests cancel the effects of
Environmental Management
further projected urbanization? What is the effect of
different urban development rates on total runoff
amount?
2. What is the affect on runoff rates by establishing
different setback distances (none versus 100 m versus
300 m) on the mainstem of the Muskegon River? Are
fixed setbacks along riparian corridors sufficient or
should setbacks be defined more ecologically using a
hydrologically defined setback distance on runoff
amounts?
3. How do future scenarios of land use-runoff differ from
runoff patterns generated for past changes in land use?
Are future scenarios of runoff better or worse than the
past?
We show that (1) controlling the rate of development is
the most effective policy option to control runoff; (2)
reforestation can abate some of the runoff effects from
urban growth but not all; (3) the size and configuration of
buffers are not as effective as controlling urban growth; (4)
land use patterns of the 1970s produced the least runoff
rates in most cases when land use maps from 1900 to 2040
are compared; and, (5) future projected land use patterns
would not always lead to increased (i.e. worse) runoff rates
than the past.
Background
Study Area
This study was conducted in the Muskegon River Water-
shed, located in the west-central part of the Lower Penin-
sula of Michigan, USA (Fig. 1). The Muskegon River
Watershed (7,057 km2) in 1998 was dominated by forests
in the north, agriculture in the central portion, and urban in
the south. The watershed was heavily logged in the late
1880s through the 1890s, mostly in efforts to rebuild
Chicago after the 1871 fire (Alexander 2006), and agri-
cultural expansion occurred in the 1910–30s and then again
in the 1960s (Ray and Pijanowski 2010). After World War
II, a lot of abandoned farmland was taken over by the state
of Michigan and converted to State Forests. Currently,
approximately 20% of the watershed is in public owner-
ship, either as State Forest (a majority), Federal Forests or
Federal Wildlife Reserve. Between 1978 and 1998, the
amount of urban land use nearly doubled, growing from
4.1% of the watershed to 7.8%. The amount of forests also
increased during this twenty-year period, rising from 46%
of the watershed to over 49% in 1998.
Twelve counties fall within the Muskegon River
Watershed. Land use planning occurs at the township, city
or village level of government (also called Minor Civil
Divisions or MCDs). There are 124 MCDs that fall wholly
or partially within the watershed. Each is required to
develop Land Use Master Plans and create ordinances and
zoning laws that control land use decisions. The city of
Muskegon, the largest city in the Muskegon River Water-
shed with a population of about 40,000 in 2000, is located
along southern shores of Muskegon Lake.
The main river, the Muskegon River, runs the length of
the watershed from its headwaters at two large inland
lakes (Houghton and Higgins Lakes) to its outlet in
Muskegon Lake which is a drowned river mouth. Mus-
kegon Lake connects to Lake Michigan through a narrow
channel. The watershed is composed of 40 subwatersheds.
Major subwatersheds connect inland lakes, wetlands and
tributaries to the mainstem of the Muskegon River mak-
ing this a very hydrologically connected watershed. The
peak flows in this system tend to be in the spring asso-
ciated with either snowmelt or large rain events. The
average annual rainfall from 1899 through 2001 was
83 cm (32.7 in).
The Muskegon River Watershed serves as an important
spawning and nursery site for economically important
Great Lakes fish such Chinook salmon (Oncorhynchus
tshawytscha), steelhead (Oncorhynchus mykiss) and wall-
eye (Sander vitreus). Assessment of the environmental
conditions of the Muskegon River and its watershed has
been made several times over the last 30 years. In the
1980s, a series of reports by the Michigan Department of
Natural Resources (cf. O’Neal 1997) identified runoff,
thermal pollution, toxic substances from industry, and
aquatic invasive species as the main threats to the health of
this river. In the late 1990s, a group of scientists from a
regional university (Grand Valley State University—
GVSU), members of natural resource agencies (e.g.,
Michigan Department of Natural Resources), city govern-
ment officials (e.g., planners, drain commissioners), and
personnel from regional economic groups began meeting to
discuss the health of the Muskegon River Watershed
(GVSU 2002a, b). These efforts culminated in a Sec-
tion 319 grant (cf. EPA 1972) awarded to GVSU from the
USEPA and Michigan Department of Environmental
Quality. This grant funded the development of a Watershed
Management Plan (GVSU 2002a) that identified several
areas of water quality concerns, including thermal pollution
and excessive nutrient loading via runoff from agricultural
fields. Several best management practices were recom-
mended (GVSU 2002c) as part of the 319 study. These
included adopting best management practices for (1) con-
struction sites; (2) reducing chemical inputs and stream-
bank erosion (e.g., reducing fertilizer application); (3)
reducing runoff conveyance and outlets; (4) sedimentation
control (e.g., buffer/filter strips); (5) vegetation establish-
ments (i.e., to reduce runoff and sedimentation); and (6)
123

Fig. 1 Study region of
Muskegon River Watershed
showing the land use categories
found (red = urban,
yellow = agriculture, light
green = shrubs, dark
green = forests, blue = water,
light brown = wetlands). The
vector lines demarcate the
individual sub-watersheds and
numbers indicate subwatershed
number (Color figure online)
improving agricultural management (e.g., fencing, manure
control, planting strategies).
Planning and Management
In summer of 2002, the Great Lakes Fishery Trust (Wiley
and others 2008, 2010) identified the Muskegon River as a
new ‘‘River Initiative’’ and in response to this initiative, a
two-day retreat (Wiley and others 2008) was held to
determine how simulation models could be used to address
what-if questions related to land use, water quality and
runoff, and, in particular, which of the several dozen rec-
ommended best management practices could be examined
more thoroughly with coupled land use-hydrologic models.
Participants were asked to discuss the various best man-
agement practices that were identified by the 319 study and
determine scenarios for coupled land use-hydrology mod-
els. Workshop participants were interested in having
models address: (1) the impact of riparian setbacks,
especially size, on hydrologic dynamics of the watershed;
and (2) the impact of controlling urban development on
hydrologic dynamics.
Following the focus group meeting, an assessment of the
size of setbacks was conducted using GIS to determine
reasonable setback sizes and configurations. Analysis of
setback sizes and configurations across the mainstem and
tributaries resulted in the following conclusions: (1)
establishing setbacks (no development) on all rivers and
123
Environmental Management
streams would exclude a large portion of the watershed
from development, which was not realistic and (2) buffers
on the mainstem would exclude less than 10% of the
watershed if they were 300 m or smaller. Thus, we
examined setbacks of 100 m and 300 m and on the river
mainstem only.
Methods
Data Preparation
To prepare data for analysis and modeling, the two land use
vector datasets (1978 and 1998) were rasterized (i.e., a grid
of cells were created from polygons of land use/cover
patches) at 26.5 meters to match the digital elevation model
(DEM) available for the region. Along with the two time
steps of land use for the region, we obtained vector data for
roads, rivers and lakes in order to parameterize drivers for
the LTM (Pijanowski and others 2007). The vector datasets
were obtained from the Michigan Geographic Framework,
version 4 (www.michigan.gov/cgi).
Surface water runoff modeling required a map of
hydrologic soil groupings for the region which was obtained
from the State Soil Geographic (STATSGO) database. We
then subsetted the soil grouping polygons in the study-region
and rasterized and recoded soils based upon hydrologic soil
type. The hydrologic soil groups were assumed to remain
fixed for the time period investigated.
Environmental Management
Modeling Land Use Changes
Future urban growth was projected using the LTM fol-
lowing Pijanowski and others (2002a, b, 2005). The LTM
uses drivers of land use changes as inputs; then using an
artificial neural network the LTM ‘‘learns’’ the relationship
between the drivers of land use change and observed
changes. Once trained, the LTM is used to project the
future urban growth using calculations of future population
growth estimates and per capita urban use requirements.
Stakeholder derived scenarios were articulated within the
LTM by excluding urbanization in a 100 m development
setback region, 300 m development setback region and
1 year ground water travel time (GWTT) setback (cf.
Pijanowski and others 2007) in the watershed area. The
100 m and 300 m development setbacks were made using
a vector line dataset of mainstem river segments; these
setbacks represented 3% and 6% of the watershed area
respectively. The stream setback method was applied based
on a GWTT model described further in Pijanowski and
others (2007) and based on MODFLOW-2000 (Boutt and
others 2001; Wayland and others 2002, 2003). We created
a policy buffer in GIS which prevented future urbanization
from occurring within areas corresponding to GWTT of
1 year or less. This setback covered 22% of the watershed
area. The arrangement of these three setbacks is shown in
Fig. 2. GWTT is an important factor in the flow of
groundwater pollutants into rivers and streams. In a similar
fashion (but obviously using a different set of driver and
regrowth rates) forests were projected for the future. The
details of this procedure are given in Appendix 1. A
summary of the future land use scenarios examined in this
article are contained in Table 1. The historical land use
changes were generated using a modified version of the
LTM called the Backcast model (Ray and Pijanowski
2010). Details of the Backcast modeling are given in
Appendix 1.
Runoff Assessment
In order to evaluate the hydrologic impacts of land use
changes in the Muskegon River Watershed, we chose to
use the curve number approach (Ponce and others 1996;
Harbor 1994; Garen and Moore 2005) to estimate runoff.
The curve number method was developed to aid watershed
managers by calculating peak storm runoff based on the
land use and soil types at a location. It is a frequently used
approach to model surface water runoff. We used this
method to calculate the peak runoff response to a hypo-
thetical storm event, across each land use scenario, treating
each of the 40 sub-watersheds individually. Details of the
specific calculations made are given in Appendix 2. We
simulated the expected runoff in response to a 5 inch storm
event for each of the subwatersheds to investigate how
past, current and future land use patterns affects runoff
during such an identical storm event. A five-inch storm
event was selected since it represented a large enough
storm event that would be sensitive to a variety of land use
patterns across this very heterogeneous watershed and its
impact could have economic effects as well and thus be
important for policy. Based on historical rainfall events at
three weather station locations in the Muskegon River
Watershed such a storm is classified as a one in 30 year
storm event.
In an attempt to determine which future land use sce-
nario was most effective at controlling runoff, we also
examined the relative contribution of each policy option on
runoff amounts compared to a baseline or reference year.
To accomplish this, the proportional changes in runoff for
all years and simulation treatments were standardized
across all subwatersheds using the following equation:
Ryr;i;tmt  R1975;i
Pyr;i;tmt ¼ ; yr [ 1975 ð1Þ
R1975;i
where Pyr;i;tmt is the proportional runoff difference for year
(yr), subwatershed (i), and treatment (tmt); Ryr;i;tmt is the
runoff for that year, subwatershed and treatment, and
R1975;i is the simulated runoff for the reference year of 1975
of each subwatershed. Note that prior to 1975 there is no
treatment as these are historical years and thus we did not
calculate Pyr;i;tmt for these years.
The final coupled model and associated tools are illus-
trated in Fig. 3. The main two models, the land change
models and the surface runoff models use the 5 inch pre-
cipitation event to generate average area weighted surface
runoff per subwatershed. GIS data processing (e.g., soils
and moisture conditions) are also necessary to create the
CN runoff model and the processing steps are shown in
Fig. 3.
Results
Comparison of Land Use Distributions Across
the Simulated Scenarios
A business as usual (Urbbase) urban growth in the Mus-
kegon River Watershed was projected to cover 19.9% of
the watershed with urban land cover by the year 2040,
whereas the slow growth mode (Urbslow) was projected to
have 13.7% total urban cover by 2040 (Fig. 4; Duckles
2008). In contrast, the 1998 map of land use for the
watershed derived from aerial photography contained 7.4%
urban. The six percent difference between urban cover in
these two scenarios results in different allocations of
123
 Environmental Management

Fig. 2 Map showing the

configuration of the three
setback buffers used in our
simulations

Table 1 Experimental design showing matrix of simulation forecast than doubles; as forest is added a large proportion of it is
scenarios (H = simulations were produced for this scenario) with eventually urbanized. This is further highlighted in the
simulation names
Forslow scenario, where forest cover increases from 58%
Transitions Buffers Urban rates of growth to over 61%; in this scenario, urban growth is controlled. In
Fast Slow the fast urban growth scenario with no addition of forest
results in a decrease of forests from 58% in 1998 to around
New forests and 100 m H (for100 m) 51% by the year 2040.
urban growth 300 m H (for300 m) When riparian development and GWTT setbacks were
1 yr GWTT H (for1 yr) applied to the Muskegon River Watershed, the setback
No buffer H (forbase) H (forslow) restrictions pushed the spatial distribution of new urban
Urban growth only 100 m H (urb100 m) cells outside the setback regions in each of the three sce-
300 m H (urb300 m) narios. Not surprisingly, beginning in 2005, the four
1 yr GWTT H (urb1 yr) urbanization scenarios had 93% cell by cell placement
No buffer H (urbbase) H (urbslow) agreement (i.e., 93% of the urban areas were the same
location for each scenario); by 2040 this map agreement
agriculture and forest land across the watershed, and a was reduced to 80% across the four urbanization scenarios.
small difference in amounts of shrub land use category. Afforestation scenarios had the same urban extent as their
The decline in the amount of agriculture over the 10 sce- respective urban scenarios.
narios is significant, with a majority of the forest scenarios Results of the backcast land use modeling (Ray and
resulting in slightly more than 11% agriculture, down from Pijanowski 2010) are shown in Fig. 5. Note that in 1900,
about 16.5% in 2005. The least decrease in agriculture more than half of the watershed is agriculture, mostly as a
occurred in the Urbslow scenario, which is not surprising result of the massive deforestation in the late 1800s. Over
as urban growth is controlled and reforestation is not added time, forests regrow, starting around 1930, making up more
as a transition to the forecast model. The amount of forest than half of the watershed’s land cover by 1960. Urban was
left in 2040 across the 10 scenarios differs as well. Adding not evident until 1940, continuing a steady increase
forests at historical rates of forest regrowth results in a through 1970. The simulation produced shrubland in about
stable amount of forest cover percentage over the 35 year a constant proportion throughout the 1900 to 1970 time
forecast period (2005–2040). This is because urban more frame.

123
Environmental Management

Fig. 3 Illustration showing

how all the model components
are interlinked

Fig. 4 LTM generated future

land use maps for two scenarios:
urbbase and urbslow

Fig. 5 Simulated land uses for

1900–1970 in decadal time
steps. Maps are generated using
the backcast model

Runoff Amounts and Land Use Forecasts
The amount of runoff generated over time and within each
subwatershed is highly variable. Figure 6 shows the runoff
plotted from 1900 through 1975 for the historical land
use and then from 2005 through 2040 for the urbbase
scenario only. Several trends are evident. For example,
subwatershed 1 has very little runoff over this 140 year
timeline although there is a steady increase over time.
Subwatersheds 9 and 31 on the other hand had moderately
high runoff in the past, and then from about 1980 through
2040, runoff volume is reduced. Several subwatersheds
also contained sharp drops in runoff starting around 1940,
which is soon after a lot of agricultural land was abandoned

123

Fig. 6 Absolute values
(1900–2040) of runoff grouped
by subwatershed. Only the
urbbase future scenario is
plotted from 2005 to 2040 for
each subwatershed
in the Great Depression and the state of Michigan con-
verted these lands to state forests.
A summary of the proportional changes (Eq. 1) in runoff
volumes by future scenario and year and plotted by sub-
watershed is given in Fig. 7. The Box plots illustrate that
for all scenarios, runoff amounts increase (i.e., are greater
than 0.0, which represents no change compared to 1975)
over time. Outliers (designated with open circles and stars,
these are values that exceed 2 SE of the mean) are most
common later during the forecast period, reaching 2 to 3
times the original runoff amount of 1975. Note that even in
the first year of the simulation, 2005, a few subwatersheds
experience a doubling (near 1.0) of runoff compared to
1975. Forslow and Urbslow produce runoff changes that
are all less than 2.0.
The relative effect of each policy scenario on propor-
tional changes in runoff values was examined across policy
scenarios, as treatment effects (4 buffer x 2 rates x 2
afforestation scenarios), and across years, using a Mixed-
Model ANOVA in SYSTAT 12. The Basic Statistic
command in SYSTAT 12 (2008) was used to determine
whether the distributions of values were normal. Skewness
and kurtosis tests showed, however, that all distributions
were heavily skewed to the left with a long right tail (all
P-values were \ 0.01). As a result, proportional runoff
values were transformed using Box-Cox transformation
with k = 0.2. A Mixed-Model ANOVA was performed in
SYSTAT 12 (Systat Inc 2008) using buffer (no buffer,
100 m buffer, 300 m buffer and 1 year GWTT buffer),
urbanization rate (slow and historical average) and affor-
estation (no new forests versus forests regenerated at rate
123
Environmental Management
observed in last 20 years) as fixed effects. Interactions of
transition type with buffer type and transition type with
urban expansion rate were explored as well. ANOVAs
were executed for each year, 2005 through 2040. A sum-
mary of the F-statistics for the ANOVAs for each year is
given in Table 2. Note that the urban growth rate (i.e. Rate)
effect contains the largest differences in all years; years
2025 and 2030 are significant at the P \ 0.05 level and
years 2035 and 2040 are significant at the P \ 0.01 level.
The presence of additional forest cover was slightly sig-
nificant (P \ 0.10) in years 2025 and beyond. In all years,
buffer configuration did not alter runoff. Interactions were
also not significant. Typically, multiple statistical com-
parisons require a Bonferoni adjustment (Sokal and Rholf
1995) of the individual a (false positive error) rates; such
an adjustment produces a family-wise alpha (a/N) of
0.00625 (0.05/8). If this is used to judge significance, then
only the rate effect for 2035 and 2040 are significant at the
P \ 0.05. However, the increasing values of the F-statistic
over time for both the Rate and Affor effect suggests that as
urbanization increases and the amount of new forests are
added to non-urban areas, the effects of these two policy
options on runoff increases.
The potential effects of buffer size on runoff were
explored further using the following approach. Each buffer
was extracted by subwatershed and the total area of the
buffer expressed as a proportion of the subwatershed was
calculated using ArcGIS; simulations without buffers were
excluded from this analysis. The proportion of subwater-
shed occupied by the buffer was then transformed using a
natural log. A Pearson correlation between natural log of
Environmental Management

for100m for1yr for300m forbase

3 3 3 3

prop_change
prop_change

prop_change
2 2 2 2

prop_change
1 1 1 1

0 0 0 0

-1 -1 -1 -1
0

0
00

01

02

03

04

05

00

01

02

03

04

05

00

01

02

03

04

05

00

01

02

03

04

05
2,

2,

2,

2,

2,

2,

2,

2,

2,

2,

2,

2,

2,

2,

2,

2,

2,

2,

2,

2,

2,

2,

2,

2,
year year year year

forslow urb100m urb1yr urb300m

3 3 3 3
prop_change

prop_change
prop_change

2 2 2

prop_change
2

1 1 1 1

0 0 0 0

-1 -1 -1 -1
0

0
00

01

02

03

04

05

00

01

02

03

04

05

00

01

02

03

04

05

00

01

02

03

04

05
2,

2,

2,

2,

2,

2,

2,

2,

2,

2,

2,

2,

2,

2,

2,

2,

2,

2,

2,

2,

2,

2,

2,

2,
year year year year

urbbase urbslow
3 3
prop_change

prop_change

2 2

1 1

0 0

-1 -1
0

0
00

01

02

03

04

05

00

01

02

03

04

05
2,

2,

2,

2,

2,

2,

2,

2,

2,

2,

2,

2,

year year

Fig. 7 Relative proportional increases of surface runoff by scenario and across years. The Box Plot represents the mean, 1SE, 2SE and 3SE from
mean for proportional increases in the 40 subwatershed

Table 2 F-statistics of Mixed Model ANOVAs for tests of effects of buffer size (none, 100 m, 300 m, groundwater delineated), rate of
urbanization (fast and slow) and whether forests are added to watershed at recent (1978–1998) historical rate and pattern
Fixed-effect df 2005 2010 2015 2020 2025 2030 2035 2040

Buffer 3 0.001 0.005 0.005 0.010 0.012 0.016 0.012 0.013

Rate 1 0.364 1.046 1.046 2.043 3.756** 5.528** 6.718***H 7.313***H
Affor 1 0.324 0.805 0.805 1.362 2.731* 3.015* 3.062* 3.144*
Affor 9 Buffer 3 0.000 0.004 0.002 0.001 0.003 0.003 0.007 0.008
Interaction of aforestation patterns and buffers also examined. ANOVAs were executed for each year. Significance of effect is given with *, **
and *** for P \ 0.10, P \ 0.05 and P \ 0.01, respectively. H Significant at the P \ 0.10 level if a family-wise error rate is used following
Bonferoni (N = 8 ANOVAs)

buffer size and natural log of runoff differential (using Comparing Runoff Variations in the Past with Future
1975 as the base simulation) was calculated for each Scenarios
forecast year (2005 through 2040 in 5 year time steps). We
found that all Pearson R2 values were small, ranging from Table 3 lists the year in which maximum and minimum
0.088 and 0.207. In other words, even in subwatersheds runoff occurred for each of the subwatersheds. Out of the
that had large areas, in terms of their proportion that fell 40 subwatersheds in the Muskegon River Watershed, 25
within a buffer, runoff was still not reduced. were at their maximum runoff condition in 1900. While the

123
 Environmental Management

Table 3 Year in which maximum and minimum runoff occurs across

the 40 subwatersheds
SW Max Min SW Max Min

0 1940 1975 20 1900 2005

1 2040 1930 21 1900 2030
2 1920 1935 22 1900 2035
3 1920 2035 23 1900 1975
4 2040 1935 24 1900 1975
5 1940 2035 25 1900 2020
6 1940 1970 26 1900 1975
7 1925 1970 27 1900 1975
8 1940 1970 28 1900 2005
9 1940 2030 29 1900 1945
10 1900 2005 30 1940 2040
11 1920 1965 31 1900 2005
12 1940 2040 32 1900 1945
13 1900 1975 33 1900 1950 Fig. 8 Per unit area (per square m) of runoff generated for each
14 1900 1975 34 1900 1945 subwatershed plotted as the maximum historical runoff amount
15 1900 1970 35 1900 1975 (x-axis) versus the maximum runoff for all scenarios and future years
(y-axis) (designated as *) and maximum historical runoff (x-axis)
16 1900 2040 36 1900 1945
compared to minimum runoff for all scenarios and future years
17 1900 2015 37 2040 1930 (yaxis) (designated as open circles). Subwatersheds that fall on or
18 1900 2025 38 2040 1930 near the line do not have differences between the past and future.
19 1900 1975 39 1900 1945 Subwatersheds that fall above the diagonal line have futures that are
worse than the maximum runoff amounts of the past (N = 13).
Likewise, subwatersheds that fall below the diagonal line have futures
that are better than the maximum runoff amounts of the past (N = 27)

runoff decreased throughout the early and mid 20th century

for many of the subwatersheds as forest regrowth occurred,
for most of the subwatersheds, an increasing trend in
overall runoff was observed for future land use scenarios
due to the projected urbanization. The lowest runoff for the
14 subwatersheds occurred between 1965 and 1975.
Interestingly, six subwatersheds are projected to have the
least amount of runoff in years 2035 and 2040. Thus, it is
apparent that the future may not always produce the worst
case scenario; it many subwatersheds the worst case has
occurred already. The years 1965 through 1975 may have
represented a time period during this 140 year timeline
where urban was relatively low and forest cover was high.
We examined how future scenarios compared to the past
in two ways. We plotted the maximum runoff values that
occurred in the past (1900 through 1975) against the
maximum runoffs for any future scenario and any future
year (Fig. 8). This plot was made for each of the 40 sub-
watersheds (starred points). Note that 10 subwatershedss
had historical maximum runoff values that were greater
than any future scenario or year. In other words, the past
was worse than any future scenario for nearly one-quarter
of all subwatersheds. We also plotted (open circles) the
smallest runoff values for any future scenario (and any
future year) in comparison with the maximum runoff for
the past. This shows that 26 subwatersheds (nearly two-
thirds) will have lower runoff amounts in the future com-
pared to the worse case condition of the past.
Discussion
In this article we simulated the surface water runoff for
land use trends occurring over a 140 year period from 1900
to 2040. Future scenarios of land use change were explored
here, including two different rates of urbanization, three
different setback configurations on the Muskegon River
mainstem, and scenarios where forest regrowth was
allowed to occur or forest cover was held constant
(although allowed to urbanize). We also compared future
scenarios with historical trends in order to determine the
relative impacts of future scenarios on runoff volumes. We
examined runoff trends across 40 subwatersheds in the
Muskegon River Watershed region using a fixed five-inch
storm event that would correspond to a 1 in 30 year event.
Fixing the precipitation amount to five-inches allowed the
estimation of the direct effect of land use on runoff over
this very heterogeneous watershed since over this 140 year
period precipitation from a severe storm would differ from
year-to-year not only as a result of natural climate vari-
ability but also due to the effects of climate changes and

123
Environmental Management
land use changes (Feddema and others 2005; Ray and
others 2006, 2009, 2010).
We found that the amount of surface water runoff varied
across the subwatersheds and given plausible futures, in
each of them the slow urban growth scenario, with affor-
estation, resulted in the least amount of runoff. This was
because this scenario had the minimum increases in urban,
while continuing to increase forest area, each resulting in
positive impacts (i.e. reduced runoff rates) on the overall
area-weighted curve number for these subwatersheds.
We found that slowing urban growth was the most
effective means of reducing surface water runoff compared
to any other policy option—such as allowing forests to
grow or adding riparian buffers to the mainstem. This work
supports Walter and others’s (2009) contention that the
amount of key land uses, such as urban or forest, are the
most important policy levers compared to any urban
development setback scenario. We found that the imple-
mentation of stream setback scenarios could result in sur-
face water runoff impacts on subwatersheds where there
are no explicit setback regions. The hydrologically defined
riparian setbacks did not significantly reduce runoff either.
Using land use forecasting models to examine the
impacts of alternate land use scenarios is becoming more
common as a land use planning activity (Landis 1995).
Indeed, with the advent of more powerful computers and
readily available models, using models to explore policy
options is likely to become even more prevalent in the near
future (cf. Bankes 1992). Models are helpful in character-
izing spatial, temporal and ecosystem variation in respon-
ses to human perturbation (Rodrguez and others 2006).
Recently, more attention has been paid to historical land-
scapes (Foster 1992; Kim and others 2002; Baker and
others 2004) such as in this assessment. Past actions put
into place structural and functional aspects of ecosystem
dynamics that need to be considered as we explore sus-
tainable planning solutions and management of landscapes
for the future (Antrop 2005). Using such models, the
effects of future variations in sub-categories of urban land
use and their role on runoff could also be addressed, i.e.
besides setbacks and GWTT, smart growth urban policies
are also alternate policy scenarios. For example, future
studies can evaluate how residential, commercial and
industrial urban landscape is constructed to minimize the
amount of surface runoff on the landscape where runoff is
important.
Engagement of stakeholders in research and modeling
per se is also becoming more common where stakeholders
help to identify scenarios, solutions and provide the context
for ‘‘how we got to where we are’’ (Walker and Salt 2003).
Impacts of research results are shown to be much greater
when stakeholders are engaged early and for a long term
(Johnson and others 2003). Examining an array of possible
scenarios within a matrix, as conducted here, has been
proposed as a means to achieve the most robust decision
making in order to achieve sustainability (Groves and
Lempert 2007). Challenges still exist, as we attempt to
examine not just trends that are apparent on decadal time
steps (e.g., land use change) but those (e.g., unusual storm
events) that present themselves as surprises (Lempert and
others 2002). Other futures, such as those that could be
posed by land use changes as a result of biofuels
(Searchinger and others 2008) may need further attention.
The modeling of the coupled interaction between forest
regrowth, urban expansion and setback scenarios could be
improved further by incorporating more process based
models, especially hydrologic models, that quantify in
greater details water balance components like infiltration,
evapotranspiration, stream flow on a more temporally
refined time scale (examining days or seasonal trends). For
example, Fohrer and others (2005) used a set of hydrologic
models coupled to land use change models to determine
how urbanization impacted stream flow dynamics. They
found that the model predicted annual trends for the
watershed and that only peak flow rates were impacted
significantly by land use changes. Hundecha and Bardossy
(2003) also examined land use change impacts on runoff
patterns in Germany and found that increased urbanization
impacted peak runoffs during the summer but not during
the winter; afforestation on the other hand, reduced peak
and total runoff in both the summer and winter.
The historical assessment provides useful context to
gauge the impacts of future changes. The curve number
analysis of past land covers show that for many of the
subwatersheds the worst runoff condition in the history of
the subwatershed was during the time that the watershed
was largely deforested at the turn of the 20th century
(Whitney 1987). Furthermore, the best hydrologic condi-
tion for many of these subwatersheds occurred during the
1970s and 1980s. The damage done to the runoff hydrology
of the watershed by clear cutting forests, in many cases, far
out-weighs the combined impact of steady abandonment of
agricultural land back to forest and increasing urbanization.
The drastic impacts of deforestation are indicated in the
poor (i.e. large) runoff rates of many of the subwatersheds
for the year 1900. As the century progressed and agricul-
tural lands were abandoned, there was steady improvement
in overall runoff condition for nearly every subwatershed.
The exceptions tend to be heavily urbanized subwatershed
areas, such as subwatershed 4 which has significant urban
and open water, and subwatershed 38 which contains much
of the city of Muskegon. With these exceptions, most of the
watershed areas improved their surface water runoff con-
dition until the late 1970s. The steady subsequent increase
in urbanization was found to increase runoff across many
of the subwatersheds since the late 1970s.
123

Despite showing that establishing no-development set-
backs along the buffers of the mainstem have less impact
on controlling surface runoff, riparian buffers do have
many other positive benefits to maintaining water quality
of rivers and streams. Riparian buffers composed of veg-
etation will trap sediment and other pollutants (cf. Correll
1997; Lowrance 1998; Lowrance and others 2000) that are
generated from the land surface and flows towards a
stream. Trees in riparian zones also help to reduce stream
temperature (e.g. Macdonald and others 2003). Additional
benefits of riparian buffers composed of vegetation include
the support of movement corridors for birds (Machtans and
others 1996).
The coupled model that we present here is one of several
that are being developed by a team of researchers studying
the dynamics of the Muskegon River Watershed. More
complex hydrologic and hydraulic models are being cou-
pled to the Land Transformation Model and the Backcast
Model (Wiley and others 2008, 2010). These models
incorporate complex processes of overland flow, evapo-
transpiration, recharge, stream flow, etc. Impacts to the
biological community are also being studied (Stevenson
and others 2008). We are currently engaging stakeholders
in various ways across the Muskegon River Watershed in
order to enhance their understanding of how the watershed
works. The development of Research Bulletins (www.
mwrp.net), a web site that is managed by researchers and
watershed organizations, helps to promote the results of our
research (Wiley and others 2008).
Acknowledgments We would like to acknowledge funding from
the NSF Water Cycle Program, (Grant #WCR 0233648), the NASA
Land Cover/Use Change and Hydrology Program, NSF III-XT Pro-
gram (Grant #IIS 0705836), the Great Lakes Fishery Trust and the
Wege Foundation. Dave Hyndman and Anthony Kendall provided the
output from the groundwater travel time model. Kimberly Robinson
read an earlier version of the manuscript and provided useful input.
Appendix 1: Modeling Land Use Changes
Urban Growth into the Future
Five types of urban land classes (residential, commercial
services and institutional, industrial, transportation,
extractive and recreation and cemeteries) in the underlying
land use vector datasets were collapsed into a single urban
land use category for modeling urban changes. Urban
growth was projected using six spatially explicit drivers
processed using ArcGIS 9.3 and then trained using the
LTM following Pijanowski and others (2002a, b, 2005,
2007). The inputs were: distance to county roads, distance
to lakes, distance to rivers, distance to existing urban,
distance to limited access highways (interstates), and
123
Environmental Management
distance to unlimited access to highways (state roads). A
separate GIS layer of non-transitioning locations of areas in
state and federal forests, all urban categories in 1978, and
areas that are in open water, were excluded from the neural
network training exercise. Urban expansion rates were
based upon research performed by the Michigan Land
Resource Project, which modeled the entire state of
Michigan at coarse spatial resolution for the years 2020 and
2040 developing growth rate factors of slow, moderate and
fast growth based upon observed past changes (Levy
2001).
Urban growth rate represents the ratio of percentage
growth of urban area to that of population. The urban area
percentage change U, is calculated as:
u1  u0
U¼ ð2Þ
u0
where u1 is urban in 1998 and u0 is the total urban area in
1978. Population percentage change is calculated similarly
as:
p1  p0
P¼ ð3Þ
p0
where p1 is population in 1998 and p0 is total population in
1978. Thus, the urban growth rate index is the ratio of the
two percentages U and P:
U
rate ¼ ð4Þ
P
Urban growth rate based upon observed past growth
used a factor of 4.2 for the base (i.e., business as usual)
scenario and 2.1 for the slow urban growth scenario
(Table 1).
Forest Regrowth Modeling for the Future
Forest regrowth was modeled separately and projected
based upon training using 12 spatially-explicit drivers for
forest regeneration in the LTM: distance to forests in 1978;
density of agriculture at 10 km, 25 km and 100 km; den-
sity of forest in 1978 at 10 km, 25 km and 100 km; dis-
tance from urban in 1978; distance from agriculture;
distance from road; size of non-forest patch; and distance
from state parks. In the projections, new forests were
allowed to grow where the 1998 land cover was either
agricultural land or shrubs. The resultant map represented
the propensity for a given cell to undergo forest regrowth.
Using the output of the forest LTM we transition cells into
the forest land use class at the rate of forest regrowth that
was observed between 1978 and 1998 in the land use
datasets. The forest regrowth scenario captured the steady
abandonment of agricultural lands and their return to forest.
The rate was forward projected to determine new forest
cells at 5 year intervals. Since wetlands and urban areas are
Environmental Management

unlikely to change, forest transitions were allowed to only 1000

occur in areas that were agriculture or shrub in 1998.
Historical Land Use Maps Generated Using
the Backcast Model
The Backcast model (Ray and Pijanowski 2010) uses the
land use changes between 1978 and 1998 to model urban
‘‘take away’’ and agricultural and forest transitions over
time. Briefly, a back propagation artificial neural network
was used to learn about locations of urban, agricultural and
forest land use/cover change; the artificial neural network
learns about spatial patterns of land use change in rela-
tionship to a variety of spatial features such as distance to
roads, natural amenities (e.g., lakes) and urban infrastruc-
ture. Data from the 2000 Census for year built of the
housing statistics was used to fix the quantity in the model
for the amount of urban for each MCD and the National
Agricultural Summary Statistics amount of land in farms
statistic was used to fix the quantity in the model for
amount of agriculture historically from 1900 to 1978 for
each county. Values for the amount of urban and agricul-
ture between census periods were estimated using a spline.
Land use maps generated between 1900 and 1975 were
used at 5 year time intervals consistent with the forward
LTM projections. The Backcast Model was validated
against [12,000 points of historical land use visually
interpreted from aerial photographs from 1939 to 1976 to
evaluate its accuracy on a decadal time scale (Ray and
Pijanowski 2010). The forecast output of the Land Trans-
formation Model was evaluated using the techniques out-
lined in Pijanowski (2006) and Pontius and others (2008).
Appendix 2: Modeling Surface Water Runoff
The soil conservation service runoff equation is expressed
as:
S ¼  10 ð8Þ
CN
and CN is the curve number. The curve number in turn is
determined from four watershed characteristics: (1)
hydrologic soil group; (2) land use class; (3) hydrologic
surface condition of native pastures and; (4) antecedent
moisture condition. For the Muskegon River Watershed,
the above parameters were correspondingly derived from:
(1) STATSGO Soils from the Natural Resources Conser-
vation Service (2008); (2) 1998 land use/cover and pro-
jected (past as well as future) land cover; (3) assumed to be
‘fair’ (i.e., assumed to be class ‘C’ habitat); (4) assumed as
antecedent moisture condition 2 (AMC2).
Hydrologic soil groups were obtained from the
STATSGO database. STATSGO classifies all soils into
four hydrologic soil groupings (HSG). HSGs are classified
based upon their infiltration rates into A, B, C and D
groups. For this watershed area, we observed all four soil
groups as present in the soil survey map, with 58% of the
watershed in class A, 15% in class B, 13% in class C and
12% in class D. The vector polygons for soil regions were
rasterized as the resolution of the land cover timesteps as
an integer raster grid with hydrologic soil groups mapped
as follows: A = 1, B = 10, C = 100, D = 1000 creating a
soil classification grid. The land cover projection for a
given scenario was multiplied by the soil classification grid
to generate a map with unique combinations of both
hydrologic soil type and land cover.
This combined map was reclassified based on land
cover/soil combinations shown in Table 4 to create a raster
grid of curve number values for each raster location. Table
4 is based on Table 2-2a in Technical Release 55 (USDA
1986) that presents simplified procedures for estimating
runoff and peak discharges in small watersheds, with the
exception of curve number for the urban land cover type.
We fixed urban curve number for all soil types at that of
soil type D, (pers. comm. B. Engel, who has studied soil
compaction in urban areas and estimates them to be of type

ðP  I a Þ2
Q ¼ ð5Þ
ðP  I a Þ þ S Table 4 Curve number (CN) values for seven land use classes, across
four hydrologic soil types for the MRW, based on TR-55
where Q is the runoff in inches, P is the rainfall in inches,
Ia is the initial abstraction, capturing all loses before runoff Land use Soil group
begins and S is the maximum retention after runoff begins. A B C D
Substituting:
Urban 87 87 87 87
Ia ¼ 0:2S ð6Þ Agriculture 67 78 85 89
gives: Shrub 55 56 70 77
Forest 30 55 70 77
ðP  0:2SÞ2 Water 0 0 0 0
Q ¼ ð7Þ
ðP þ 0:8SÞ Wetlands 0 0 0 0
where: Barren 77 86 91 94

123

D for runoff modeling). For each of the 40 subwatersheds
in the Muskegon River Watershed an area weighted aver-
age curve number value was computed. The area weighted
curve number for each of the 40 subwatersheds was cal-
culated across 100 policy maps, resulting in 400 curve
number summarizations for the forward projections and
600 curve number summarizations for the 15 backcast
maps across 40 subwatersheds. Area weighted curve
numbers were input into Eq. 7 with a rainfall amount of
five inches to simulate the runoff for such a storm event.
References
Alexander J (2006) The Muskegon: the majesty and tragedy of
Michigan’s rarest river. Michigan State University Press
Alig RJ, Kline JD, Klein M (2004) Urbanization on the US landscape:
looking ahead in the 21st century. Landscape and Urban
Planning 69:219–234
Allan JD (2004) Landscape and riverscapes: the influence of land use
on river ecosystems. Annual Reviews of Ecology, Evolution and
Systematics 35:257–284
Antrop M (2005) Why landscapes of the past are important for the
future. Landscape and Urban Planning 70:21–34
Bankes S (1992) Exploratory modeling for policy analysis. Opera-
tions Research 41(3):435–449
Baker JP, Hulse D, Gregory S, White D, Sickle J, Berger P, Dole D,
Schumaker N (2004) Alternative futures for the Willamette
River Basin, Oregon. Ecological Applications 14(2):313–324
Bhaduri B, Harbor J, Engel B, Grove M (2000) Assessing watershed-
scale, long-term hydrologic impacts of land-use change using a
GIS-NPS Model 2000. Environmental Management 26(6):643–
658
Booth D, Karr J, Schauman S, Konrad C, Morley S, Larson M, Burges
S (2004) Revising urban streams: land use, hydrology, biology
and human behavior. Journal of the American Water Resources
Association 40(5):1351
Boutt DF, Hyndman DW, Pijanowski BC, Long DT (2001) Identi-
fying potential land use-derived solute sources to stream
baseflow using ground water models and gis. Ground Water
39(1):24–34
Bren LJ (1998) The geometry of a constant buffer-loading design
method for humid watersheds. Forest Ecology and Management
110(1–3):113–125
Bren LJ (2000) A case study in the use of threshold measure of
hydrologic loading in the design of stream buffer strips. Forest
Ecology and Management 132:243–257
Carpenter S, Caraco N, Correll D, Howarth R, Sharpley A, Smith V
(1998) Nonpoint source pollution of surface waters with
phosphorus and nitrogen. Ecological Applications 8(3):559–568
Chaubey I, Maringanti C, Schaffer B, Popp J (2008) Targeting versus
optimization: critical evaluation of BMP implementation for
watershed management. In: World Environmental and Water
Resources Congress 2008
Cifaldi R, Allan JD, Duh J-D, Brown DB (2003) Spatial patterns in
land cover in exurbanizing watersheds of southeastern Michigan.
Landscape and Urban Planning 66:107–123
Correll DL (1997) Buffer zones and water quality protection: general
principles. In: Haycock N, Burt T, Goulding K, Pinary G (eds)
Buffer zones: their processes, potential in water protection.
Quest Environmental, Harpenden, UK
123
Environmental Management
Diaz RJ, Rosenberg R (2008) Spreading dead zones and consequences
for marine ecosystems. Science 321:926–929
Diebel MW, Maxted JT, Nowak PJ, Vander Zanden MJ (2008)
Landscape planning for agricultural nonpoint source pollution
reduction I: a geographical allocation framework. Environmental
Management 42(5):789–802
Dosskey M (2002) Toward quantifying water pollution abatement
response to installing buffers on crop land. Environmental
Management 28(5):577–598
Duckles JM (2008) Impacts of land use change on runoff in the
Muskegon River watershed of Michigan, past, present and
future: a stakeholder driven, scenario-based approach. MS
Thesis, Department of Forestry and Natural Resources, Purdue
University, West Lafayette, Indiana, p 109
EPA (1972) Clean Water Act, Section 319
Feddema JJ, Oleson KW, Bonan GB, Mearns LO, Buja LE, Meehl
GA, Washington WM (2005) The importance of land-cover
change in simulating future climates. Science 310:1674–1678
Fitzpatrick M, Long D, Pijanowski B (2007) Characterizing biogeo-
chemical fingerprints of land use on surface water quality across
a regional watershed in the Upper Midwest, USA. Applied
Geochemistry 22(8):1825–1840
Fohrer N, Haverkamp S, Frede H (2005) Assessment of the effects of
land use patterns on hydrologic landscape functions: develop-
ment of sustainable land use concepts for low mountain range
areas. Hydrologic Processes 19:659–672
Foster D (1992) Land-use history (1730–1990) and vegetation
dynamics in central New England, USA. Journal of Ecology
80(4):753–771
Garen DC, Moore DS (2005) Curve number hydrology in water
quality modeling: uses, abuses and future directions. Journal of
the American Water Resources Association 41(2):377–388
Groves DG, Lempert RJ (2007) A new analytic method for finding
policy-relevant scenarios. Global Environmental Change
17:73–85
GVSU (2002a) Muskegon River Watershed project: watershed plan,
vol I. Annis Water Search Institute, Muskegon, Michigan, p 191
GVSU (2002b) Muskegon River Watershed project: management
plan appendices, vol II. Annis Water Search Institute, Muskegon,
Michigan, p 31
GVSU (2002c) Muskegon River Watershed project: education plan
III. Annis Water Search Institute, Muskegon, Michigan, p 42
Harbor JM (1994) A practical method for estimating the impact of
land-use change on surface runoff, groundwater. Journal of the
American Planning Association 60(1):95
Hundecha Y, Bardossy A (2003) Modeling of the effect of land use
change on the runoff generation of a river basin through
parameter regionalization of a watershed model. Journal of
Hydrology 292(1–4):281–295
Johnson NL, Lilja N, Ashby JA (2003) Measuring the impact of user
participation in agricultural and natural resource management
research. Agricultural Systems 78:287–306
Koomen E, Stillwell J (2007) Modelling land-use change: theories
and methods. GeoJournal Library 90:1–22
Kim Y, Engel B, Lim K, Larson V, Duncan B (2002) Runoff impacts
of land-use change in Indian River Lagoon Watershed. Journal
of Hydrologic Engineering 7(3):245–251
Landis J (1995) Imagining land use futures: applying the California
urban futures model. Journal of the American Planning Asso-
ciation 61:438–457
Lempert R, Popper S, Bankes S (2002) Confronting surprise. Social
Science Computer Review 20(4):420–440
Leopold L (1968) Hydrology for urban land planning—a guidebook
on the hydrologic effects of urban land use. USGS Circular 554
Leopold L (1994) A view of the river. Harvard University Press,
Cambridge, MA, p 298
Environmental Management
Levy D (2001) Michigan land resource project. Technical report.
Public Sector Consultants, Inc
Lowrance R, Altier L, Williams R, Inamdar S, Sheridan J, Bosch D,
Hubbard R, Thomas D (2000) REMM: the riparian ecosystem
management model. Journal of Soil and Water Conservation
55(1):27–34
Lowrance R (1998) Riparian forest ecosystems as filters for non-point
source pollution. In: Pace M, Groffmann P (eds) Chapter 5 in
success, limitations, and frontiers in ecosystem science. Springer
Publishers, New York, p 499
Macdonald JS, MacIssac E, Herunter H (2003) The effect of variable
retention riparian buffer zones on water temperatures in small
headwater streams in sub-boreal forest ecosystems of British
Columbia. Canadian Journal of Forest Research 33:1371–1382
Machtans C, Villard M, Hannon S (1996) Use of riparian buffer strips
as movement corridors by forest birds. Conservation Biology
10(5):1366–1379
Mitsch WJ, Gosselink JG (1993) Wetlands. Van Nostrand Reinhold,
New York
Mölders N, Rühaack W (2002) On the impact of explicitly predicted
runoff on the simulated atmospheric response to small-scale
land-use changes—an integrated modeling approach. Atmo-
spheric Research 63:3–38
National Research Council (2008) Urban stormwater management in
the United States. The National Academies Press, Washington,
DC, 529 pp
NRCS (2008) US General Soil Map (STATSGO2) Description.
http://www.soils.usda.gov/survey/geography/statsgo/description.
html
O’Neal R (1997) Muskegon River Watershed assessment. Fisheries
Division Special Report No. 19. State of Michigan Department
of Natural Resources
O’Callaghan JO (1996) Land use: the interaction of economics,
ecology and hydrology. Chapman and Hall Publishers, London
Paul M, Meyer J (2001) Streams in the urban landscape. Annual
Review of Ecology and Systematics 32:333–365
Pijanowski B, Ray D, Kendall A, Duckles J, Hyndman D (2007)
Using backcast land use change and groundwater travel-time
models to generate land use legacy maps for watershed
management. Ecology and Society 12(2):25
Pijanowski BC (2006) Afforestation patterns in the Upper Midwest,
USA. In: Proceedings of the IUFRO landscape ecology confer-
ence, Locorotondo, Bari, Italy, 26–29 Sept 2006
Pijanowski BC, Pithadia S, Shellito BA, Alexandridis K (2005)
Calibrating a neural network-based urban change model for two
metropolitan areas of the Upper Midwest of the United States.
International Journal of Geographical Information Science
19(2):197–215
Pijanowski BC, Shellito B, Pithadia S (2002a) Using artificial neural
networks, geographic information systems and remote sensing to
model urban sprawl in coastal watersheds along eastern Lake
Michigan. Lakes and Reservoirs 7:271–285
Pijanowski BC, Brown DG, Shellito BA, Manik GA (2002b) Using
neural networks and GIS to forecast land use changes: a land
transformation model. Computers, Environment and Urban
Systems 26(6):553–575
Ponce VM, Hawkins RH, ASCE M (1996) Runoff curve number: Has
it reached maturity? Journal of Hydrologic Engineering 1:11–19
Pontius R G Jr, Boersma W, Castella J-C, Clarke K, de Nijs T, Dietzel
C, Duan Z, Fotsing E, Goldstein N, Kok K, Koomen E, Lippitt
CD, McConnell W, Sood AM, Pijanowski BC, Pithadia S,
Sweeney S, Trung TN, Veldkamp AT, Verburg PH (2008)
Comparing the input, output, and validation maps for several
models of land change. Annals of Regional Science 42(1):11–47
Ray DK, Welch RM, Lawton RO, Nair US (2006) Dry season clouds
and rainfall in northern central America: implications for the
Mesoamerican biological corridor. Global and Planetary Change
54:150–162. doi:10.1016/j.gloplacha.2005.09.004
Ray DK, Pielke RA Sr, Nair US, Welch RM, Lawton RO (2009)
Importance of land use versus atmospheric information verified
from cloud simulations from a frontier region in Costa Rica.
Journal of Geophysical Research. doi:10.1029/2007JD009565
Ray DK, Pijanowski BC (2010) A backcast land use change model to
generate past land use maps for the Muskegon River Watershed
of Michigan, USA. Journal of Land Use Science 5:1, 1–29. doi:
10.1080/17474230903150799
Ray DK, Pielke R A Sr, Nair US, Niyogi D (2010) Roles of
atmospheric and land surface data in dynamic regional down-
scaling. Journal of Geophysical Research 115:D05102. doi:
10.1029/2009JD012218
Rodrguez JP, Beard JTD, Bennett EM, Cumming GS, Cork S, Agard
J, Dobson AP, Peterson GD (2006) Trade-offs across space,
time, and ecosystem services. Ecology and Society 11(1), art28
Searchinger T, Heimlich R, Houghton R, Dong F, Elobeid A, Fabiosa
J, Tokgoz S, Hayes D, Yu T (2008) Use of US croplands for
biofuels increases greenhouse gases through emissions from
land-use change. Science 319:1238–1240
Shields FD, Bowie AJ, Cooper CM (1995) Control of streambank
erosion due to bed degradation with vegetation and structure.
Water Research Bulletin 31:475–489
Sokal RR, Rohlf FJ (1995) Biometry, 3rd edn. WH Freeman and
Company, 887 p
Stevenson RJ, Wiley M, Gage S, Lougheed V, Riseng C, Bonnell P,
Burton T, Hough R, Hyndman D, Koches J, Long D, Pijanowski
B, Qi J, Steiman A, Uzarski D (2008) Watershed science:
essential, complex, multidisciplinary and collaborative. In:
Wetland and water resource modeling and assessment: a
watershed perspective. Taylor and Francis Publishers, Baca
Raton, Florida
Systat Inc (2008) Systat 12, Introduction of linear mixed models,
chap. 5. Systat Software Inc., San Jose, CA
Tang Z, Engel B, Lim K, Pijanowski B, Harbor J (2005a) Minimizing
the impact of urbanization on long-term runoff. Journal of the
Water Resources Association 41(6):1347–1359
Tang Z, Engel BA, Pijanowski BC, Lim KJ (2005b) Forecasting land
use change and its environmental impact at a watershed scale.
Journal of Environmental Management 76:35–45
Tomer M, Burkart M (2003) Long-term effects of nitrogen fertilizer
user on ground water nitrate in two small watersheds. Journal of
Environmental Quality 32:1931–1938
Tong S, Chen W (2002) Modeling the relationship between land use
and surface water quality. Journal of Environmental Manage-
ment 66(4):377–393
United States Department of Agriculture (1986) Tr-55: urban
hydrology for small watersheds. Technical report. Natural
Resource Conservation Service Conservation Engineering
Division
USEPA (1983) Final peport of the nationwide urban runoff program.
US Environmental Protection Agency, Water Planning Division,
Washington, DC
USEPA (2000) Nutrient criteria technical guidance manual: lakes and
reservoirs, 1st edn. EPA 8222-B00-001. Washington, DC
Walker B, Salt D (2006) Resilience thinking. Island Press, Washing-
ton, DC
Walter MT, Archibald J, Buchanan B, Dahlke H, Easton Z,
Marjerison R, Sharma A, Shaw S (2009) New paradigm for
sizing riparian buffers to reduce risks of polluted storm water:
practical synthesis. Journal of Irrigation and Drainage Engineer-
ing 135(2):200–2009
Wang L, Lyons J, Kanehl P, Gatti R (1997) Influences of watershed
land use on habitat quality and biotic integrity in Wisconsin
streams. Fisheries 22(6):6–12
123

Wayland KG, Hyndman DW, Boutt D, Pijanowski BC, Long DT
(2002) Modelling the impact of historical land uses on surface-
water quality using groundwater flow and solute-transport
models. Lakes and Reservoirs 7:189–199
Wayland KG, Long DT, Hyndman DW, Pijanowski BC, Woodhams
SM, Haack SK (2003) Identifying relationships between base-
flow geochemistry and land use with synoptic sampling and R-
mode factor analysis. Journal of Environmental Quality
32(1):180–190
Whitney GG (1987) An ecological history of the great lakes forest of
Michigan. The Journal of Ecology 75(3):667–684
Wiley M, Pijanowski B, Jan Stevenson R, Seelbach P, Richards P,
Riseng C, Hynman D, Koches J (2008) Integrated modeling of
123
Environmental Management
the Muskegon River: tools for ecological risk assessment in a
Great Lakes Watershed. In: Ji W (ed) Wetland and water
resource modeling and assessment: a watershed perspective,
chap 20. Taylor & Francis, London
Wiley M, Hyndman D, Pijanowski BC, Kendall A, Riseng C,
Rutherford E, Cheng S, Carlson M, Tyler J, Stevenson R, Steen
P, Richards P, Seelbach P, Koches J (2010) A multi-modeling
approach to evaluate impacts of global change on river
ecosystems. Hydrobiologia. doi:10.1007/s10750-010-0239-2
Wolter PT, Johnston CA, Niemi GJ (2006) Land use and cover
change in the U.S. Great Lakes basin 1992–2001. Journal of
Great Lakes Research 32:607–628