GIS Time Series Analysis of the Loxahatchee Slough, Florida
by Susan M. Hohner
ABSTRACT
The Loxahatchee Slough is the largest wilderness island refuge of
relatively undisturbed wetland remaining in Palm Beach County. Aerial
photography and satellite imagery from pre- and post-channelization
dates are classified by vegetation communities. GIS analysis is used to
compare these diverse data sets. Using vegetation communities as an
indicator, results show a trend toward dryer hydroperiod. In 1979, the
Army Corps of Engineers began raising water levels back toward
pre-channelization levels, in a portion of the study area known as the
Historic Region. Results indicate a positive response, with a net
increase of longer hydroperiod vegetation in this region.
DESCRIPTION OF THE STUDY AREA
In 1984, the Palm Beach County Commission adopted the Natural
Areas Park Preserve Program, aimed at protecting the last remaining
wilderness island refuges in Palm Beach County. Identification and
evaluation of potential sites was completed in 1988 by Dr. Grace
Iverson and Dr. Daniel Austin of Florida Atlantic University, in a report
entitled Inventory of Native Ecosystems in Palm Beach County, Phase
III Report, Location and Evaluation of Sites for Possible Preservation as
Wilderness Island Park Preserves. The largest site identified in the
report is the Loxahatchee Slough, spanning approximately 7,720 acres,
including 7,593 acres of undisturbed ecosystem. The Slough is located
in the City of Palm Beach Gardens within Public Land Survey System
(PLSS) Townships 41S and 42S, and Ranges 41E and 42E. This is within
the 105.8 square mile C-18 drainage basin (Cooper,1988).
Historically, the Loxahatchee Slough was the Northernmost part of the
Everglades, with drainage diverging into both the Everglades and the
Loxahatchee River (Parker, 1955 and 1974). Inundation of the
emergent wetland community occurred 50 to 80 percent of the year.
However, with the completion of the C-18 canal in 1958, inundation of
wetland habitat fell below 50 percent, causing a transition to drier
mesic shrub/scrub community (United States Army Corps of Engineers,
1983).
Beginning in April 1979, the hydroperiod was increased for that portion
of the Slough located in the SouthWest region of the study area,
referred to as the "Historic Region" by the U.S. Fish and Wildlife
Service (U.S. Army Corps of Engineers, 1983).
METHODOLOGY
The objective of this study was to detect culturally induced changes in
the hydroperiod of the Loxahatchee Slough. Vegetation types are used
as an indicator of hydroperiod. The earliest remotely sensed data
available for the study area is from 1940, and hydrologic conditions at
the time are considered to be similar to historic hydrology. Between
1940 and 1979 the C-18 canal was built, along with other cultural
impacts which reduced the hydroperiod. In 1979, the Army Corps of
Engineers raised the water levels in the Historic Region. A 1989
Spot image was the most recent data available when this project
began. Prior studies of this type have not attempted to compare aerial
photography with satellite imagery. By transforming the data into
classified choropleth maps with consistent classification categories,
these three data sets can be compared and land cover changes analyzed
in a GIS (Figure 1). Software used for Digital Image Analysis was
IDRISI 4.0 Image Processing software, and ArcInfo 6.1.1 was used for
building the GIS. Hardware included a Sun SPARCstation LX (UNIX,
SunOS 4.1.3c) and a Howtek Scanmaster 3+ scanner.
Figure 1: Methodology Flowchart
The primary data for 1940 consists of two overlapping 1940
United States Department of Agriculture (USDA) 1:40,000 scale black
and white aerial photographs (CJF 4 86, CJF 6 81) taken on February
28 and March 9 (Figure 2). This is the earliest remotely sensed data
available of the study site (Richardson, 1977).
Figure 2: 1940 USDA Aerial Photographs
The primary 1979 data consists of two overlapping 1979 USDA 1:40,000
scale black and white aerial photographs (12099 178-63, 12099 178-65)
taken on December 26 (Figure 3).
Figure 3: 1979 USDA Aerial Photographs
In both cases, for 1940 and 1979, the pairs of photographs were scanned
and imported into GIS as grids. The grids were geo-rectified and each
pair merged into a single grid for that year (Figures 4 and 5).
Figure 4: 1940 Merged Grid
Figure 5: 1979 Merged Grid
These merged grids were displayed and digitized on screen. Location
of the boundaries to be digitized between land cover types was
determined through photointerpretation. No ground truthing material
was available for circa 1940 or 1979, so no accuracy assessment was
made. Polygon topology was built and each coverage was clipped to
the study area. Land cover type was attributed to each polygon
(Figures 6 and 7).
Figure 6: 1940 Land Cover Polygon Coverage
Figure 7: 1979 Land Cover Polygon Coverage
The primary data for 1989 consists of a 1989 SPOT HRV multispectral
(green, red and near infrared) 20 meter resolution satellite image
scanned on May 13. Processing began with clipping the SPOT satellite
raw image to the study area boundaries, applying image enhancement
techniques to improve visual distinction of land cover, and creating a
supervised classification from training fields selected from the
enhanced image (Figure 8).
Figure 8: 1989 SPOT Satellite Image
During the process of image classification of the 1989 satellite image,
the classification system developed for spectral signatures was
reclassed down to four categories which also can be distinguished on
the USDA aerial photographs for 1940 and 1979, namely, Forested Land
(Low Hammock and Pine Flatwoods), Forested Wetland (Cypress Swamp),
Nonforested Wetland (Open Water, Marsh and Wet Prairie) and Disturbed
(Road and Canal). Ground truthing was done by the cluster sampling
technique and by aerial observations, due to the inaccessibility of
the terrain. Observations were insufficient to develop an error
matrix, but visual comparison did appear to generally confirm its
validity. The supervised classification image was imported as a grid
into a GIS, geo-rectified, and converted to a polygon coverage
(Figure 9).
Figure 9: 1989 Land Cover Polygon Coverage
Total areal extent for each land cover type was used to calculate
percentages for land cover (Figure 10).
Figure 10: Land Cover Percent Graph
In order to use vegetation types as an indicator of hydroperiod, a
hydroperiod value must be associated with each vegetation class. The
hydroperiod values used here are based on Brown (1989) (Figure 11).
Figure 11: Vegetation Type and Related Hydroperiod
In the present study, the Nonforested Wetlands class includes both of
Brown's Wet Prairie and Marsh classes. This gives a hydroperiod range
of 150-365 days. Nonforested Wetland, or Cypress, falls within this
range, and therefore the Forested Wetland and Nonforested Wetland
classes must be combined into a single class, Wetland, with a
hydroperiod of 150 days or more. Brown's Hammock class indicates the
hydroperiod for Low Hammock, which is included here in the Forested Land
class, is 100-150 days. Forested Land also includes Pine Flatwoods,
which has a shorter hydroperiod than Low Hammock. Therefore Forested
Land has a hydroperiod of less than or equal to 150 days. In the
present study, the Disturbed category is rated as functioning at the
lowest level for supporting a wetlands ecosystem, and therefore
equivalent to the shortest hydroperiod. The ordinal ranking of these
three classes, then, from longest to shortest hydroperiod, is: Wetland,
Forested Land and Disturbed. Areal percentages for these revised
categories were calculated (Figure 12).
Figure 12: Hydroperiod Percent Graph
The polygon coverages, for 1940, 1979 and 1989, were overlaid. This
intersected coverage was then divided into the Historic Region (where
water levels were increased as of 1979) and the remainder of the study
area, labelled the Non-Historic Region. This allows for a spatial
analysis of where hydroperiod has increased, decreased or remained
the same for 1940 to 1979, 1979 to 1989, and 1940 to 1989 (Figures 13,
14 and 15).
Figure 13: 1940 - 1979 Hydroperiod Change
Figure 14: 1979 - 1989 Hydroperiod Change
Figure 15: 1940 - 1989 Hydroperiod Change
Examination of these change maps indicates that along some of the tree
islands and portions of the canal, hydroperiod has increased along one
edge while decreasing along the opposite edge. This appears to be an
artifact of mis-registration during the geo-rectification process,
causing the overlays to be slightly offset. This is not unexpected for a
study area with few human artifacts and little ground truthing data for
registration and verification.
This mis-registration error has resulted in artificially inflating the
percents for both Increased Hydroperiod Vegetation Land Cover
Change and Decreased Hydroperiod Vegetation Land Cover Change, by
equal amounts. In order to compensate for this error, the common
increase has been discounted, by calculating the difference between
the percents. While some of the discounted percents may reflect
actual change, rather than being an artifact of mis-registration, how
much is not ascertainable here. This calculation of the difference also
serves to summarize the results of this study. These results indicate
that for the study area as a whole, hydroperiod has dropped since
channelization, but it appears that the rate at which hydroperiod is
dropping for the Loxahatchee Slough as a whole, has slowed since
water levels were increased for the Historic Region in 1979 (Figure 16).
Figure 16: Net Hydroperiod Change for Study Area
And, more significantly, an increase of hydroperiod has been detected
for the Historic Region after water levels were increased in 1979,
(Figure 17) where the area has shown "a positive response of fauna to
raising water levels" (U.S. Army Corps of Engineers, 1982).
Figure 17: Net Hydroperiod Change for Historic Region
CONCLUSIONS
The objective of this study was to examine the culturally induced
changes in hydroperiod of the Loxahatchee Slough. While statistical
confidence in the quantitative results derived in the present study is
unavailable, this study has produced qualitative results which are
indicative of the changing conditions of the Slough. Visual inspection,
alone, of the individual aerial photographs and satellite image do not
reveal the trend toward increased hydroperiod in the Historic Region
since 1979. However, this has been detected through time series
analysis.
REFERENCES
Brown, M.T., 1989. Forested wetlands in urbanizing landscapes. In Proceedings of the Symposium: The Forested Wetlands of the Southern United States. Gen. Tech. Rep. SE-50, eds. D.D. Hook and R. Lea. Asheville, NC: USDA, Forest Service, SE Forest Experiment Station.
Cooper, R.M., and J. Lane, 1988. An Atlas of Eastern Palm Beach County Surface Water Management Basins. Technical Memorandum DRE-244. West Palm Beach, FL: Water Resources Division, Resource Planning Department, South Florida Water Management District.
Hohner, S.M., 1994. Vegetation Time Series Analysis of the Loxahatchee Slough, Palm Beach County, FL: A GIS incorporating Satellite Imagery with Black and White Aerial Photography, Masters Thesis. Florida Atlantic University, Boca Raton, FL.
Iverson, G.B., and D.F. Austin, 1988. Inventory of Native Ecosystems in Palm Beach County, Phase III Report; Location and Evaluation of Sites for Possible Preservation as Wilderness Island Park Preserves. Unpublished report to the Palm Beach County Commission.
Iverson, G.B., and D.F. Austin, 1989. Inventory of Native Ecosystems in Palm Beach County, Amendment Covering Evaluation of Sites in Range 41, Townships 41, 42, and 43. Unpublished report to the Palm Beach County Commission.
Parker, G.G., 1974. Hydrology of the pre-drainage system of the everglades in southern Florida. In Environments of South Florida: Present and Past. ed. P.J. Gleason. Miami, FL: Miami Geological Society. pp. 18-27.
Parker, G.G., G.E. Gerguson, S.K. Love, et.al., 1955. Water Resources of Southeastern Florida. Geological Survey Water-Supply Paper 1255. Washington, DC: United States Government Printing Office.
Richardson, D.R., 1977. Vegetation of the Atlantic coastal ridge of Palm Beach County, Florida. Florida Scientist 40(4):281-330.
United States Army Corps of Engineers, Jacksonville District, 1982. Final Report, Environmental Investigation of the Canal 18 Basin and Loxahatchee Slough,Florida. Jacksonville, FL: Department of the Army, Corps of Engineers.
United States Army Corps of Engineers, Jacksonville District, 1983. Draft Feasibility Report and Environmental Assessment for Canal 18 Basin Loxahatchee Slough, Central & Southern Florida. Jacksonville, FL: Department of the Army, Corps of Engineers.
Susan M. Hohner
Staff Geographer, Database & GIS Support Unit
Hydrologic Data Management Division, WRE
South Florida Water Management District
3301 Gun Club Road
West Palm Beach, FL 33416-4680
Phone: (561) 682-6801
Fax: (561) 682-6442
Email: susan.hohner@sfwmd.gov