MAPPING SPATIAL/TEMPORAL DISTRIBUTIONS OF GREEN MACROALGAE IN A
PACIFIC NORTHWEST COASTAL ESTUARY VIA SMALL FORMAT
COLOR INFRARED AERIAL PHOTOGRAPHY*
D.R. Young1, S.P. Cline1, D.T. Specht1, PJ, Clinton2, B.D, Robbins1, and J.O. Lamberson1
'Western Ecology Division, NHEERL, U.S. EPA, 2OAO Corp.
2111 SE Marine Science Drive, Newport, OR 97365-5260, U.S.A.
ABSTRACT
A small format 35 mm hand-held camera with color infrared slide film was
used to map blooms of benthic green macroalgae upon mudflats of Yaquina
Bay estuary on the central Oregon coast, U.S.A. Oblique photographs were
taken during a series of low tide events, when the intertidal mudflats along the
drowned-river were exposed. The resulting images were digitally scanned and
georeferenced to commercially produced digital orthophotographs. Benthic
surveys of two transects (-3500 m2) oriented perpendicular to the estuary's
channel were conducted within about two weeks of each aerial survey.
Distributions of the perennial seagrass Zostera marina along the upper edge of
the channel were delineated from the aerial photographs taken in late spring
before extensive development of the green macroalgae beds. Summer
expansion and fall contraction of these algal beds (comprised principally of
Ulva spp. and Enteromorpha spp.) was documented via a series of four aerial
and ground surveys conducted between May and December. The study
demonstrated the usefulness of this approach in mapping blooms of green
macroalgae in Pacific Northwest estuaries.
Disclaimer: This information has been funded wholly by the U.S. Environmental Protection
Agency. It has been subjected to the Agency's peer and administrative review, and it has been
approved for publication as an EPA document. Mention of trade names or commercial
products does not constitute endorsement or recommendation for use.
* Presented at the Sixth International Conference on Remote Sensing for the Marine and
Coastal Environments, Charleston, South Carolina, 1-3 May 2000.
-------
INTRODUCTION
The steady growth in human habitation of the Pacific Northwest (PNW) has led to increased
concern regarding eutrophication of coastal estuaries in this region of the U.S.A. Nutrients
released by anthropogenic sources in coastal watersheds can stimulate growth of algae in
freshwater and estuarine receiving waters (Cerco and Seitzinger, 1997; Raffaelli et al., 1998).
Such eutrophication can affect natural populations of ecologically important organisms
(Isaksson and Pihl, 1992). The rapid development of benthic macroalgae on estuarine mudflats
during the summer growing season makes difficult the accurate documentation of spatial and
temporal distributions of such algae from ground surveys alone. This paper describes progress
made in developing a relatively inexpensive method of mapping estuarine intertidal vegetation
throughout the year using small format (35 mm) color infrared (CIR) aerial photography. This
approach is well suited for the dynamic and demanding conditions encountered in the coastal
environment of Oregon. The timing of monthly low tide cycles requires the capability to
acquire quality photos under variable light and weather conditions. CIR film was selected
because the resultant false color images can be used to delineate water, soil, and classes of
actively growing vegetation (Young et al., 1998; 1999). However, for consistent results CIR
film requires meeting special handling needs and working within a narrow exposure latitude (ca
0.5 f stop).
BACKGROUND
Digital orthophotographs had been obtained from large-format CIR aerial surveys conducted
on July 23, 1997 (photoscale 1:7200) and August 10,1998 (1:6000). Comparison of the 1997
images with results of associated ground surveys (Young et al., 1998; 1999) showed that,
during that summer of relatively low algae densities, major beds of native eelgrass (Zostera
marina) generally were distinguishable from those of green macroalgae (mainly Ulva spp. and
Enteromorpha spp.). Distributions of the introduced eelgrass species Zostera japonica near the
upper edge of the mudflats also were discernable in the 1997 and 1998 CIR aerial photographs
(Speeht et al., 2000). From these images there appeared to be a distinct increase in the 1998
distribution of green macroalgae over that mapped in 1997. The temporal disparity of these
surveys (two and one-half weeks) may explain the increased macroalgae in 1998. To address
this question, as well as concerns that high densities of green macroalgae among the seagrass
can mask its CIR signature, we initiated a small format aerial photography time series program
in concert with associated ground surveys at selected sites within the estuary.
-------
METHODS
AERIAL SURVEYS
Low oblique, 35 mm color infrared photographs were acquired over ground transects on six
mudflats of Yaquina Bay estuary near low tide during each of four survey dates between May
and December, 1999. The general study area, including Site 2 (Idaho Point) and Site 3
(Coquille Point) discussed here, is illustrated in Figure 1 of our companion paper (Specht et al,,
2000). Using CIR 35 mm slide film (Kodak Ektachrome infrared [EIR], 36 exposures),
photographs were taken from an open window of a Cessna 170 airplane using a hand-held
Nikon N90 camera equipped with an MF-26 programmable back. Two different Nikon AF
Nikkor lenses were used: a 50 mm 1:1.4 D and a 24 mm 1:2.8 D, both equipped with Tiff en
filters (Sky 1-A and #12 [yellow]). Usual altitudes averaged 400 m, resulting in photoscales of
1:8,000 to 1:16,666, respectively, for 50 and 24 mm lens. Photos were framed to include
permanent landmarks around each site. Typical camera settings for a photo survey on a cloudy
day were: auto-exposure with matrix metering, manual focus infinity (lens taped), shutter
priority, shutter speed 1/500 s, 3.6 fps auto-film advance, ISO 200 (for E-6 processing),
exposure compensation -0.5 stop, and 3-frame exposure bracketing of +/- 0.7 stop. During a
typical 30 minute flight, one 36-exposure roll was used. Each site was photographed twice,
once with each lens, resulting in 6 frames of mixed scales and exposures from which to select
one or more for continued analysis.
Following development by E-6 processing, the uncut film frames were scanned digitally
using an Epson 836x1 transparency scanner at a resolution of 6400 dpi, and the digital
photographs were georeferenced using ground control points from the digital orthophotos
obtained from the 1997 and/or 1998 large-format CIR survey. Due to variation in photoscale,
resulting ground pixel sizes of the georeferenced imagery varied between 15 and 25 cm.
Geographical Image Processing software (ERMapper v. 6.0) was used to classify S AV
reflectance in the georeferenced imagery in an iterative process based on band ratio algorithims
of image pixel intensity values (Clinton et al. ,2000). Due to photometric variation between
images, a standard algorithim for use with all images was not achieved. In some cases, more
than one algorithm would be used in order to classify both submersed and emmersed SAV's, or
to distinguish between SAV taxa. Photointerpretation training was derived from previous SAV
photointerpretation projects, and ground survey data from these two sites was reserved for an
independent comparison with the results obtained from the digital classification (Figure 1).
GROUND SURVEYS
Ground survey zones were established at the six study sites whose range incorporated most
of the intertidal range of native eelgrass in Yaquina Bay estuary. A "permanent" transect 100
meters in length was established by placing 11 -12 wooden stakes along a line perpendicular to
the river channel, generally at 10 meter intervals (in narrow eelgrass beds a 5 m line was
added). Each transect was situated so that the first three or four stakes (labeled 0 m, 10 m,
-------
20 m,...) were within the upper section of the eelgrass bed, with the next stake (30 m or 40 m)
just outside (upslope) of the dense section of the bed. These transect stakes were
geopositioned using a backpack Differential Global Positioning System (DGPS) (CMT Model
PC5-L) with a ~ + 0,6 m horizontal accuracy (Young et al, 1998). From each transect stake a
lateral line ~ 35 m long (station) was established perpendicular to the transect line, and three
randomly selected distances along this line (approximating a bathymetric contour) were marked
by numbered sample stakes. This established a fixed ground survey zone of ~ 3500 m2 at each
of the six sites, with the lowest three to four of the 11-12 stations situated within the dense
upper portion of the native eelgrass bed. Between early June and mid-August, 1999, during
each ground survey three randomly-selected distances from the transect stake were preselected
for each station (excluding a two meter interval around a fixed station stakes where the
vegetation may have been altered by the stake's presence). Beginning in late August, at Sites 2
and 3 only stations Om, 10 m, 20 m, 50 m, and 90 m were sampled. Percent cover values for
native eelgrass, green macroalgae, and bare substrate were obtained from each of these samples
(and usually a fourth sample collected near the transect stake, outside the range of apparent
disturbance). These samples were taken during low tide when the stations were exposed, using
a 0.5 m square quadrant containing two perpendicular sets of 5 equally-spaced taut wires. The
values were obtained by measuring the frequency of occurrence under the 25 intercept points
and/or by visual estimation. Results from the two approaches have been found to be highly
correlated (r2 = 0.97; Young et al, 1998).
PERCENT COVER CALCULATIONS
To obtain percent cover values from the digital classification, rectangular grids 10m wide
and 35 meters long centered around the ground station lines were established in the EPA
Geographical Information System (GIS) and overlayed on the SAV classifications. The percent
area within a given rectangle classified as SAV then was determined. Average percent cover
values for native eelgrass, green macroalgae, and their sum (grouped as SAV) also were
determined from the values for the three to four quadrat samples obtained at each station during
a given ground survey. Because the aerial survey conducted in early November fell midway
between the dates of the ground surveys conducted in late October and late November (a period
of rapid change in the SAV distributions), the percent cover measurements made during those
two months were combined to obtain the November average (n = 8) for each station.
RESULTS
For each site and survey period, the average percent cover values for SAV in a 10 m x 35 m
grid, determined via the digital classification and the ground survey methods, are compared
(Figure 1), We note that a relatively sharp spatial gradient in the percent cover values for native
eelgrass was observed during the first ground survey in early June, before the development of
dense green macroalgae cover that could mask this eelgrass boundary zone. Specifically, at Site
2 the average (+ 1 std. error) values for percent cover by eelgrass at the 20 m and 30 m stations
were 48 ± 11 and 0 + 0 percent, respectively. At Site 3 the corresponding values for the 30 m
-------
and 50 m statons were 81+4 and 0 + 0 percent. These upper boundary zones for the dense
sector of the native eelgrass bed at Site 2 and at Site 3 are indicated by the vertical dashed lines
(Figure 1).
The late October and late November ground survey averages for the 50 m station at Site 2
were 96 + 1 and 51 + 11 percent, respectively. Corresponding values for the 90 m station were
88 + 5 and 19 + 3 percent. For Site 3, the October and November 50 m station values were 95
+ 3 and 12 + 6 percent, respectively. Corresponding values for the 90 m station were 86 + 4
and 1 + 1 percent. For each of these stations, the two-month average and 95 % confidence
interval is illustrated in Figure 1.
DISCUSSION AND CONCLUSIONS
The two methods of measuring S AV percent cover were compared; in general, they
agreed within about 20 percentage points. However, somewhat greater discrepancies were
observed for the lower stations at Site 3 during August, and at Sites 2 and 3 during December.
In both cases, the percent cover values obtained from the digital classification were lower than
those obtained from the ground survey. The differences in the December values may reflect the
fact that the ground survey was conducted early in the month, while the aerial survey was
conducted two weeks later. Thus, the lower digital classification values may indicate a
measurable loss of algal coverage over that period. November's digital classification values for
station 50 m at Site 2, and 50 m and 90 m at Site 3, also were lower than those obtained from
the ground surveys (although only the first difference was significant). A review of the
November aerial photographs after this comparison revealed inundation of the stations up to
and including the 80 m station. This suggests that, at the lower stations, the high percent cover
values from the digital classification could have been caused by floating eelgrass blades;
upslope of these beds, the CIR signal from the benthic green macroalgae could have been
diminished by absorption of light in these spectra via passage through the overlying water.
In any case, the large drop in percent cover of green macroalgae (which constituted the entire
SAV signal at the 50 m and 90 m stations) between the late October and late November ground
surveys illustrates the need for a practical method of frequently monitoring intertidal SAV
coverage in PNW estuaries. As noted in the previous section, at each site a relatively sharp
boundary was measured along the upper edge of the native eelgrass bed during early June, when
there was insignificant coverage by green macroalgae. Results of various eelgrass distribution
surveys conducted in the estuary indicate that this boundary does not undergo a major shift up-
slope during the summer growing season. This suggests the strategy of accurately mapping the
upper edge of the native eelgrass beds between late fall and early spring, and then masking out
these areas in the GIS analysis. This would allow temporal changes in the exposed intertidal
distributions of SAV, obtained from frequent surveys using small-format CIR aerial
photography, to be interpreted with confidence as representing temporal changes in green
macroalgae coverage outside the relatively fixed upper boundary of the native eelgrass beds.
-------
ACKNOWLEDGMENTS
We thank W. Bengal, EPA Western Ecology Division's Coastal Ecology Branch, and employees of
DynCorp/TAI, Inc.(Newport, OR), for assistance in conducting the ground surveys. Aerial survey
flights were provided by J. Newell, Central Oregon Coast Air Services (Newport, OR). C. Doyle and
B. Coleman, OAO Corp., provided image analysis and technical support.
REFERENCES
C.F. Cerco and S.P. Seitzinger, "Measured and modeling effects of benthic algae on eutrophication in
Indian River - Rehoboth Bay, Delaware," Estuaries, Vol.20, No. 1, pp. 231-248, 1997.
Clinton, P.J., D.R. Young, B.D. Robbins, and D.T. Speeht, "Issues in digital image processing of aerial
photography for mapping submersed aquatic vegetation." In Proceedings of the Sixth Interntional
Conference on Remote Sensing for Marine and Coastal Environments, ERIM Intemation, Inc., Ann
Arbor, Michigan, 2000.
I.. Isaksson and L. Phil, "Structural changes in benthic macrovegetation and associated epibenthic faunal
communities," Netherlands Journal of Marine Research, Vol. 30, pp. 131-140, 1992.
D.G. Raffaelli, J.A. Raven, and LJ. Poole, "Ecological impact of green macroalgae blooms." In
Oceanography and Marine Biology: An Annual Review, eds. A.D. Ansell, R.N. Gibson, and M. Barnes,
UCL Press, 1998.
D.T. Speeht, D.R. Young, and P.J. Clinton, "Near infrared aerial photo-detection of Zostera japonica
communities in Pacific Northwest estuarine intertidal habitats." In Proceedings of the Sixth Interntional
Conference on Remote Sensing for Marine and Coastal Environments, ERIM International, Inc., Ann
Arbor, Michigan, 2000.
D.R. Young, D.T. Speeht, PJ. Clinton, and H. Lee n, "Use of color infrared aerial photography to map
distributions of eelgrass and green macroalgae in a non-urbanized estuary of the Pacific Northwest
U.S.A." In Proceedings of the Fifth International Conference on Remote Sensing for Marine and
Coastal Environments, ERIM International, Inc., Ann Arbor, Michigan, Vol. n,
pp. 37-45, 1998.
D.R. Young, D.T. Speeht, B.D. Robbins, and PJ. Clinton, "Delineation of Pacific Northwest SAVs from
aerial photography: Natural color or color infrared film?" In ASPRSAnnual Conference Proceedings,
American Society of Photogrammetry and Remote Sensing, Bethesda, Maryland, 1999.
-------
WED-00-069
TECHNICAL REPORT DATA
(Please read instructions on the reverse before completing)
1. REPORT NO. 2.
EPA/600/A-00/028
4. TITLE AND SUBTITLE Mapping spatial/temporal distributions of green macroalgae
in a Pacific Northwet coastal estuary via small format color infrared aerial photography
7. AUTHOR(S) D.R. Young!1, S.P. Cline1, D.T. Specht1, P.J. Clintonr2 B.D. Bobbins1,
J.O, Lamberson1
9. PERFORMING ORGANIZATION NAME AND ADDRESS
'US EPA NHEERL WED 2OAO Corporation
2111 SE Marine Science Drive 2111 SE Marine Science Drive
Newport, OR 97365-5260 Newport, OR 97365-5260
12. SPONSORING AGENCY NAME AND ADDRESS
US EPA ENVIRONMENTAL RESEARCH LABORATORY
200 SW 35th Street
Corvallis, OR 97333
3. RECIPIENT'S ACCESSION NO.
5. REPORT DATE
6. PERFORMING ORGANIZATION
CODE
8. PERFORMING ORGANIZATION REPORT
NO.
10. PROGRAM ELEMENT NO.
1 1 . CONTRACT/GRANT NO.
1 3. TYPE OF REPORT AND PERIOD
COVERED
14. SPONSORING AGENCY CODE
EPA/600/02
15, SUPPLEMENTARY NOTES:
ABSTRACT: A small format 35 mm hand-held camera with color infrared slide film was used to ma
mudflats of Yaquina Bay estuary on the central Oregon coast, U.S.A. Oblique photographs were tal
the intertidal mudflats long the drowned-river were exposed. The resulting images were digitally sc
produced digital orthophotographs. Benthic surveys of two transects (-3500mm2) oriented perpenc
conducted within about two weeks of each aerial survey. Distributions of the perennial seagrass Zo
channel were delineated from the aerial photography taken in late spring before extensive developrr
expansion and fall contraction of these alqal beds (compromised principally of Ulva SDD. and Ente
series of four aerial and ground surveys conducted between May and December. The st
this approach in mapping blooms of green macroalgae in Pacific Northwest estuaries.
j blooms of benthic green macroalgae upon
cen during a series of low tide events, when
anned and georeferenced to commercially
icular to the estuary's channel were
stera marina along the upper edge of the
ent of the green macroalgae beds. Summer
rcomorpha Spp.) was documented via a
udy demonstrated the usefulness of
17. KEY WORDS AND DOCUMENT ANALYSIS
a. DESCRIPTORS b. IDENTIFIERS/OPEN ENDED
TERMS
remote sensing, color infrared, arial
photography, mapping, estuary, green
macroalgae, Pacific Northwest.
1 8. DISTRIBUTION STATEMENT 1 9. SECURITY CLASS (This Report)
c. COSATI Field/Group
21. NO. OF PAGES: 7
------- |