Wetlands and GIS
By:
Rebecca Tully
Wetlands are essential for the survival of many organisms. Wetlands offer habitat for organisms that
live in them as well as protection to those that live on the outskirts of
them. In recent years geographic
information system (GIS) has been used in the study of wetlands. GIS is useful in looking at spatial
distribution of wetlands, determining wetland loss or gain over a period of
time, creating visual representations of habitat location for organisms and the
effects of wetland loss on those organisms, as well as many other uses. This bibliography attempts to show the uses
of GIS in relation to the study of wetlands both in the United States and
abroad.
Articles:
Tiner, Ralph W. Estimated Extent of Geographically Isolated Wetlands In
Selected Areas of the United States. Wetlands: Vol. 23, No.
3, pp. 636–652.
U.S. Fish and Wildlife Service Northeast Region 300 Westgate Center
Drive Hadley, Massachusetts, USA 01035 E-mail: ralph_tiner@fws.gov
This paper discusses a study done
by the Fish and Wildlife Service (FWS) that was published as a web-based report
on _Geographically Isolated Wetlands._ Despite isolation these wetlands
perform many of the same important functions that other wetlands perform. The study used Geographic Information Systems
(GIS) to identify potentially isolated wetlands. To facilitate in the identification of these
sites this definition for geographically isolated wetlands was chosen: _a
wetland with no apparent surface-water connection to
perennial rivers and streams, estuaries, or the ocean. They are surrounded by dry land. Three main data sources were used for the
digital data information needed on potential sites to be analyzed using
GIS. The data sources were: 1. National
Wetlands Inventory (NWI) digital data for wetlands 2. USGS digital line graphs
(DLGs) for hydrology 3. USGS
digital rasters (DRGs). The data from these data sources was compiled
and analyzed using GIS ArcView and ArcInfo. A summary
of the GIS procedures used to compile and analyze the data is as follows;
1. Digital data for each quad was
gathered 2. To import selected DLG hydro.sdts files
to an ArcInfo coverage 3. _Select_ perennial
intermittent streams form DLG hydro layer and deepwater habitats form the NWI
data 4. Buffer selected water layers with a 20m buffer 5. _Select_
by theme (NWI & DLG) all NWI deepwater habitats intersecting the 20m buffer
6. _Union_ selected DLG hydro layer with selected NWI deepwater habitat
data 7. Eliminate internal polygon lines in the unioned
data by using the ArcView dissolve geoprocess 8. Take the NWI data and query to select all
features intersecting the unioned 20-m
buffer-selected deepwater habitat data 9. Build out the selection to include any
wetland or deepwater habitat touching one that is within or contiguous to this
buffered data 10. Reverse or switch the selection and the highlighted features
are _isolated wetlands and deepwater habitats._ 11._Select_ isolated wetlands
and deepwater habitats at a distance of 20 m from the unioned
20-m buffer-selected deepwater habitat data; to highlight wetlands located in
the _20–40 m buffer_ for analysis 12. Compare results versus DRG 13. Produce
draft map and data summaries 14. Select out isolated deepwater habitats from
isolated wetlands 15.Generate final map and data summaries. The study found that isolated wetlands
constituted a large portion of wetlands resources in arid and semi-arid to subhumid regions in Karst
topography. 43 of the 72 sites studied
were found to have more than 50% of their total wetlands designated as
isolated. The results of this GIS analysis present one perspective on the
extent of geographically isolated wetlands in the country and represent a
starting point for more detailed assessments.
J.F. Gottgens,
B.P.Swartz, R.W.Kroll, M. Eboch. Long-term GIS-based records of
habitat changes in a Lake Erie coastal marsh. Wetlands Ecology and
Management 6: 5-17, 1998.
The study in this paper attempted
to quantify long-term variability in marsh habitat characteristics and relate
that to gross environmental changes and human impact. The
goal was to
quantify changes in habitat distribution and habitat features under diked and undiked conditions in
an area with different natural protection form open water wave/wind
energy. The Winous
Point marshes were used for this study because they have a well-documented
history and are the largest private wetland in Ohio. Geographic Information Systems (GIS)
technology was applied to a 120-year record of images of this marsh
system. The GIS was used to compile and
analyze data as well as study time trends from the historic images showing
terrain conditions. The analysis was
focused on habitat characteristics related to emergent aquatic vegetation. Spatial analysis for this study was done using
vector-based GIS techniques on two hand-drawn maps and eight aerial
photos. For three regions (West Marsh,
Muddy Creek Bay, South Marsh) gross habitat types were
classified, their distribution mapped, and area calculated. The linear length of the Marsh edge was also
calculated in GIS as a cumulative boundary between two map categories in each
region for each image. The study presented
in this paper suggested that losses in emergent plants corresponded with high
lake-levels, particularly in recent decades.
This GIS-based study showed that diked marshes
duplicated pre-settlement conditions for certain gross habitat characteristics.
J.P. Roise*, T.H. Shear and
J.V. Bianco. Sensitivity
analysis of transportation corridor location in wetland areas: a multiobjective programming and GIS approach. Wetlands
Ecology and Management 12: 519–529, 2004.
The object of this paper was to test a logical approach to
Army Corps of Engineers requirements that when constructing something effort be
made to avoid wetlands and if that is not possible efforts must be made to
minimize the impact on wetlands. They
start by using a multiobjective function to represent
the costs of developing roads in an area containing wetlands. There are two components to this function: 1.
Construction costs that go into building a road 2. Wetland impact costs. Wetland impact costs are based on negative
effects to habitat, hydrology, and water quality. This means a loss of the _wetland functional value._ Next a mathematical
formulation was used to show multiple objective costs that go into building a
road through a wetland. This formulation
was tested by creating a program using Geographical Information Systems (GIS)
software to generate least cost pathways between two points for a given set of
weights. The same program was also used
to report estimated construction costs, loss of wetlands function, and the
acreage of wetlands that will be impacted by the road. The objective of the tests in this paper was
to demonstrate how this approach to road corridor location might be a useful
tool for determining the trade-offs between environmental damage cause by
building the road and the financial costs of road construction. The GIS program can be used to map out
alternative placements for roads; displaying them visually with the associated
costs and environmental impacts.
McCauley, Lisa A. and Jenkins, David G. GIS-based
Estimates of Former and Current Depressional Wetlands in an Agricultural
Landscape. Ecological Applications: Vol.15, No. 4, pp.
1199-1208.
Before the settlement of Europeans in the Midwestern
United States 23% of Illinois was covered in wetlands. Due to the fact that Illinois has a
topography consisting of gently rolling hills many of the lost wetlands were
most likely depressional wetlands.
Depressional wetlands are areas with hydric
soils that are located in an area of lower elevation, surrounded by higher
elevation, so that surface flow is not sufficient to drain the area. They provide _ecosystem services,_ by contributing to species diversity and genetic
diversity in a landscape. There is a
lack of knowledge of former wetlands in the state so restoring wetlands in a
now agricultural area must happen without the knowledge of presettlement
conditions McCauley et al dealt with this deficiency by creating a model of
former depressional wetlands in a heavily drained county of Illinois based on
Geographic Information System (GIS). The
GIS, along with field verification, was used to estimate spatial extent,
density, pattern, and sizes of former and existing depressional wetlands in
Champaign County, Illinois. Three
different models, Digital Raster Graphics (DRG), Digital Elevation Models
(DEM), and Digital Orthophotography Quarter Quadrangles
(DOQ), were made using publicly available data sets to depict the locations of
former depressional wetlands and overlaying them with digital hydric soils. Models
were also made for current depressional wetlands and all four models were
tested against visual observations of the sites. Lastly a combination of the DRG and DEM
models was derived. The DRG, DEM, and
DOQ models were tested by observing sites after precipitation occurred, where
saturated soils and/or standing water without visibly noticeable drainage were
considered positive evidence for a former depressional wetland. The GIS models gave data on the density,
sizes, and spatial configuration of past and current depressional wetlands. The distances between wetlands were analyzed
by nearest neighbor analyses to consider the potential effects of large-scale
drainage on metapopulation dynamics of wetland
dependent species. Nearest neighbor
distances were found using the ARC/INFO POINTDISTANCE command. This gave each polygon in the past and
current depressional wetland models a label point at the center of the polygon
and the distance from the center of each polygon to its nearest neighbor was
found. The polygon centers rather than
boundaries were used because depressional wetlands tend to be seasonal and
their boundaries change depending on hydrology.
Frequency distributions for nearest neighbor distances were also
calculated to compare former and current landscapes. Former and current landscapes were analyzed
by: frequency distributions; glabal Moran_s I, for which a significantly larger value that expected indicates wetlands were spatially clustered;
and Geary_s C, for which significantly smaller than
expected values indicates wetlands were clustered. The DRG-based model was found to be the most
reliable of the three tested with 95% of the predicted wetlands being observed
in field tests. The DRG-based model
identified 1077 wetlands covering 1108 ha.
The DEM model predictions did not match the DRG model. Only 11.4% of past depressional wetlands
identified by the DEM overlapped with the DRG.
The DEM model was 89% accurate in the field, yielding 3401 former
depressional wetlands covering 1884 ha.
Due to the fact that both models were fairly accurate but different the
authors also developed a model that combined the two. This new model had an accuracy of 91.5% and
estimated 4524 former depressional wetlands in Champaign County. The DOQ model produced several thousand more
wetlands than the other models but had a low accuracy of 44.2%. GIS models provide exact
locations, which allows the details of drainage-driven spatial
distributions to be looked at. Fifty percent of the former nearest neighbor
distances occurred within 259m of each other while today only 7.8% of current
wetlands occur within that same distance.
This study found that GIS-based models can accurately predict former and
current depressional wetlands if DRG and/or DEM models are applied. Combining the two models may yield and
accurate model with the advantages of both being present. The approach used in this paper and it_s results are relevant to spatial ecology, dispersal
kernels, and the effects of wetland loss and fragmentation on freshwater
biodiversity.
Kent, Barbara J., and Mast, Joy N. Wetland Change Analysis of San Dieguito Lagoon, California USA: 1928-1994. Wetlands:
vol. 25, No. 3, pp. 780-787
The study in this paper analyzed changes in a coastal
wetland using remote sensing, image processing, and geographic information
systems (GIS) techniques. Wetlands are
second only to tropical rain forests as the endangered habitats in the
world. In the United States, an
estimated 52.3% of the original wetlands in the lower 48 states have been
turned into non-wetland. In this study
remote sensing, image processing, and GIS were used to determine the change in
the San Dieguito Lagoon, a coastal wetland near Del
Mar, California over three periods from 1928-1994. Regional studies such as this one can reveal
specific reasons for landscape change and can increase the publics
awareness which will in turn help to generate local support for the protection
of these types of environments. Over the
past 100 years human activities have helped to degrade the lagoon. The starting date of 1928 represents a time
when the Del Mar region was rural and undeveloped and was the earliest date
that aerial photographs for the area were developed. 1945 marks the date when San Diego began to
implement planning strategies to attract tourists. The 1975 date coincides with the period when
various wetland and water protection policies were enacted. The most recent date of photographs at the
time was 1994 when the study was done.
The images were scanned at 118 dots per cm, imported into ERDAS Imagine,
and converted to image files for image processing. The digital photos were georeferenced
to the Universal Transverse Mercator (UTM) coordinate system and resampled to a 3-m resolution. This was a landscape-level study so a binary
classification of wetlands versus non-wetland was required. Wetland areas were identified using both
supervised and unsupervised classification techniques. After the classification was completed ground
verification was done in 1996 for the 1994 image and an accuracy assessment was
performed. If one of the points were located
on water of in urban areas then the class was clearly wetland of non-wetland
respectively. A GIS change detection
model was used to compare classified data sets from 1928, 1945, 1975, and 1994,
calculating the changes in overall wetland area. The change-analysis images include only the
area common to all four images, a total of 736 ha. The image processing showed that the majority
of wetlands were lost after 1928 and 1945.
Overall, a large amount of wetland loss compared to a small amount of
wetland gain throughout the entire study area.
An accuracy assessment of the results was done and found that in this
study wetland was never misclassified as non-wetland, but non-wetland was
misclassified as wetland in a few instances.
No matter what year the film was or its type areas of wetland in the
images appeared as areas of wetland in the classification. This study provides a set of baseline
landscape data, combined with status and trends for reforestation of the San Dieguito Lagoon. The
results from the landscape-scale analyses illustrate a spatial pattern of
wetland change since the 1920s.
Semlitsch, Raymond D. and Bodie, Russell J. Are
Small, Isolated Wetlands Expendable?
Conservation Biology, Vol.12, No. 5. (Oct., 1998), pp.
1129-1133.
New regulations that were drafted by the U.S. Army Corps of
Engineers that reduce protection for “headwater” or “isolated” wetlands have
sparked a controversy. This paper argues
that small wetlands are very valuable for maintaining biodiversity in a number
of plants, invertebrate, and vertebrate taxa and that
the disappearance of small wetlands will cause a huge reduction in the
ecological connectance among remaining species
populations. Rather than approaches only
concerned with how the total area would be affected by the loss of huge versus
small wetlands determining the frequency distribution of wetland sizes better
addresses the biological importance of wetlands. They also argue in this paper that the number
of individual wetlands is more important than the area that they take up
because is addresses the abundance and distribution of individual wetland
populations, which is the most basic unit of community species diversity and
genetic diversity. In this study
geographic information system data from the 780-km2 Savannah River
Site in South Carolina were used to describe the size frequency and spatial
distribution of all depression wetlands known as Carolina bays at this
site. 46.4% of the bays were 1.2 ha or
less and 87.3% were 4.0 ha or less. Some
available data from Florida suggests that large wetlands may be less diverse
than smaller wetlands. Larger wetlands
are usually more permanent and thus tend to contain predatory fish and a
greater variety or abundance of invertebrate predators that can exclude amphibian
larvae. The consequences
of losing small isolated wetlands lies in potential changes to the
population dynamics of the remaining wetlands.
The two primary effects to consider are; a reduction in the number or
density of individuals dispersing and an increase in dispersal distances among
wetlands. Raymond and Russell looked at
how the loss of individual wetlands affects the straight-line distance to the
nearest wetland. They determined that
there would be a 41.3% increase in distance between nearest bays with the loss
of all wetlands smaller than 1.2 ha and a 136.1% increase in distance with the
loss of wetlands smaller than 4.0 ha.
According to this study small, isolated wetlands are not expendable if
the goal is to maintain the present levels of species biodiversity. The authors of this article strongly advocate
that wetland legislation focus not only on size but also on local and regional
wetland distribution.
Williams, Donald C. and Lyon, John G. Historical Aerial Photographs
and a Geographic Information System (GIS) to Determine effects of Long-Term
Water Level Fluctuations on Wetlands Along the St. Marys
River, Michigan, USA. Aquatic Botany 58 (1997) 363-378.
This paper uses Geographic Information System (GIS) to study
the influences of long-term water level fluctuations on the wetland areas of
the St. Marys River.
The river is a connecting channel between Lake Superior and Huron that
borders on the US and Canada. The water
levels in the River are largely controlled by the levels of Lake Huron such
that wetland areas along the St. Marys undergo
fluctuations corresponding to those of Great Lakes wetlands. The analysis was conducted on a US Fish and
Wildlife Service National Wetland Inventory digital data set on the wetlands of
the St. Marys River.
GIS software was used in the analysis.
The wetland types found in the study areas are characterized as
unconsolidated bottom, emergent wetland, unconsolidated shore, scrub-shrub
wetland and forested wetland using the USFWS classification system. Wetland classes and areas in the St. Marys River were determined by interpretation of aerial
photos from the summers of 1939, 1953, 1964, 1978, 1982, 1984, and 1985. The USFWS used a GIS system to create GIS
files of the wetland polygons in ELAS format, which could be read by ERDAS
software. The National Wetland Inventory
gave maps and WAMS-derived tabular summaries.
Further GIS analysis using the digital files offered several
advantages. The multiple wetland classes
identified by the National Wetland Inventory could readily be aggregated to
simplify the analyses and to employ wetland classes used in comparable wetland
studies on the Great Lakes. National
Wetland Inventory summaries were completed on individual quadrangles and GIS
was used to combine the areas of each quadrangle in to one digital file to
reduce the number of analyses. The
digital files for each of the study years were combined into one mosaic using
the ERDAS GIS software program subset.
GIS was also used to determine the exact common area for each year so
that year-to-year changes in wetland class and area could be determined with
greater accuracy. Using the GIS they
were also able to see spatial changes in wetlands. Between 1964 (the year with the largest area)
and 1978 (the year with the least area) there was a 426 ha difference in
emergent wetland. This difference
represents a 32% change. There was a 514
ha change in the area of unconsolidated bottom between 1978, the year with the
largest area, and 1964, the year with the smallest
area. This was about a 9% change in
area. There was a change of 347 ha
between the maximum and minimum of unconsolidated shore area. There was a decreasing trend in area of
unconsolidated shore until it nearly disappeared in 1984 and 1985. The changes in scrub-shrub wetland were
clearly influenced by water level. The
area of these wetlands varied by 63 ha, nearly 16% of their maximum area. Forest wetland area experience a 37%
change. The transition matrix made from
GIS was of particular help in this study because it showed interclass transfers
from wetland state to wetland state and was helpful in identifying successional effects and distinguishing them from water
level effects. This study indicated that
the use of GIS and historical aerial photographs can be used together to
effectively study water level changes and how they effect
wetland class areas.
Cohen, Marcelo C.L. and Lara, Ruben J. Temporal Changes of Mangrove
Vegetation Boundaries in Amazonia: Application of GIS and Remote Sensing Techniques.
Wetlands Ecology and Management 11: 223-231, 2003.
This paper analyses a series of satellite and aerial images
covering a 25-year period of time from 1972 to 1997. It focuses on identifying and quantifying
areas with vegetation coverage losses of gains in mangroves and elevated
mudflats. Coastal wetlands are thought
to be highly susceptible to sea-level rise.
Due to the fact that there is not that much evidence relating to
sea-level rise and loss of coastal wetlands the authors of this article
recommend intensive monitoring of coastal areas identified as vulnerable to
these effects. The study area here was
the Braganca peninsula. The area is characterized by a mangrove
peninsula that is crossed by tidal channels linking wetlands with the estuary. Visual observations, photographs, and GPS
measurements during the 1999 dry season were used to determine plant species
and charcterise main geobotanical
units in the peninsula. These images as
well as other images obtained were used to create a three-band composition
image using the ERMAPER 5.5 image processing system. It was found that elevated flats of the inner
part of the peninsula were flooded much less frequently that the mangrove
areas. The image time series analysis
indicated net losses of mangrove coverage along a ~166 km stretch of coastline,
including the Braganca coastal plain and adjacent
areas. During the period from 1972-1997
losses were registered along ~42% of the coastline length. There were no changes along 39% of the
coastline and gains occurred along 19% of it.
If the rate of loss in this area persists the mangroves will almost
completely disappear in about 750 years.
The authors concluded that the current rate of vegetation gain/loss in
the study area seem to be related to rates of sea-level rise. Therefore, it seems that sea-level rise may
be a driving force for mangrove death on the Braganca
peninsula.
Burke, Vincent J. and Gibbons, J.W. Terrestrial Buffer Zones and
Wetland Conservation: A Case Study of Freshwater Turtles in a Carolina Bay.
Conservation Biology, Pages 1365-1369, Vol. 9, No. 6, December 1995
In the past decade several conservation statutes have been
enacted due to reports that many wetlands in the United States had been
converted to residential property and agricultural land. Since the enactment of these statutes few
studies have been done to measure their effectiveness. The study conducted in this paper used
geographic information system (GIS) to evaluate whether or not current laws
adequately preserve the wildlife in wetlands.
These evaluations were done by comparing a freshwater turtle community’s
habitat requirements to the varying levels of wetland protection. The GIS analysis generated two models
comparing turtle upland habitat use to habitats contained with the federally
delineated wetland boundary and a 30.5m buffer around it. The study was conducted in Ellenton Bay on
the US Department of Energy’s Savanah River site in
west-central South Carolina. In 1992 and
1993 from April through July 73 gravid mud turtles, 14 Florida cooters, and 6 slider turtles were captured during nesting
forays. After capture the gravid turtles
were equipped with a radio transmitter and put back in Ellenton Bay. When nesting was finished each nest site was
excavated to verify the presence of eggs, restored to original appearance, and
marked with a flag. In 1992 from
September to December 24 mud turtles were equipped with radio transmitters and
monitored to locate their hibernation burrow locations. All the nests and hibernation sites were
precisely located in the field and registered to an aerial photograph. They were then digitized and overlaid onto a
rectified scanned image of the aerial photograph using ARC/INFO GIS software. Delineated wetland/upland boundaries were
recorded in the field using GPS and the line was then overlaid onto the GIS
image. A 30.5m buffer zone was also
generated and overlaid on the GIS image.
Two biologically-based buffer zones were created to reflect the amount
of upland habitat the turtles actually used.
The results showed that only 44% of nests and hibernation burrows were
insulated by the buffer zone and 100% of the sites were beyond the delineated
line boundary. The distributions
strongly suggest that current federal protection is not adequate for freshwater
turtles. For full protection a 275m
buffer beyond the federal delineation line is needed. If the most distal nests and hibernation
sites were eliminated the buffer would need to be smaller, but would still have
to be 73m beyond the line. This study
begins to show the futility of preserving one component (i.e., aquatic habitat)
of an integrated landscape.