NEAR INFRARED AERIAL PHOTO-DETECTION OF ZOSTERA JAPONICA
COMMUNITIES IN PACIFIC NORTHWEST ESTUARINE INTERTIDAL HABITATS.*
David T. Specht1, David R. Young1 and Patrick J. Clinton2
1 Coastal Ecology Branch, Western Ecology Division, NHEERL, US EPA,2 OAO Corp.,
2111 SE Marine Science Drive, Newport, OR 97365-5260
ABSTRACT
Near infrared color aerial photography (-1:7200) of Yaquina Bay, Oregon, flown at minus
tides during summer months of 1997 was used to produce digital stereo ortho-photographs
covering tidally exposed eelgrass habitat. GIS analysis, coupled with GPS positioning of ground-
truth data detected Zostera japonica communities (non-indigenous eelgrass), which are physically
separated by elevation in this and similar Pacific North-west coastal estuaries from Zostera marina
(native eelgrass) communities. The non-indigenous Z. japonica typically occurs at or near mean
high water while the native Z. marina is restricted to -0.7 m above mean low low water and below.
Recognition of Z japonica patches from adjacent bare sediment and emergent beach or marsh
grasses was aided by combining specifically tailored detection algorithms and "heads up" digitizing
with digital bathymetry. Further progress in algorithm development should allow areal delineation,
mapping and change analysis.
INTRODUCTION
In Pacific Northwest estuaries, tidal excursion maxima range from ~2 to 4+ m, allowing
the use of near infrared false color imaging of patches of intertidally exposed submerged aquatic
vegetation (S AV) communities, utilizing a variation of the standard Coastal Change Analysis
Protocol (C-CAP) (Dobson et al. 1995). Zostera marina (L.) is the native eelgrass species in the
Pacific Northwest (including Yaquina Bay); its habitat in central Oregon coast estuaries is
effectively restricted by exposure, dessication and other environmental influences from subtidal
to -0.7 m above mean low low water (MLLW) (Kentula and Mclntire, 1986, Phillips and Menez,
1988, Bayer, 1979). Individual plants are occasionally encountered above that elevation in
standing tidal water puddles (pers. obs.; Kentula and Mclntire 1986). Zostera japonica (Aschers.
and Graebn.) (first thought to be Zostera noltii [Hornemann]) was recognized by Bayer (1979,
1996) in the mid-1970s as a then recently established non-indigenous species in the Yaquina Bay
estuary. Bigley and Barreca (1982) determined the identity of this species invading Pacific
Northwest estuaries as Z. japonica. Phillips and Menez (1988) estimate its introduction to the
Pacific Northwest coast with Japanese oysters imported to Willapa Bay, Washington, in 1925;
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. Specific mention of trade names or commercial products does not imply or constitute endorsement or
recommendation for use by the U.S. Environmental Protection Agency.
* Presented at the Sixth International Conference on Remote Sensing for Marine and Coastal
Environments, Charleston, South. Carolina, 1-3 May 2000
-------
Bulthuis (1995) reported that Japanese oyster seed importation began as early as 1902 in Samish
Bay in northern Puget Sound, where Z, japonica is abundant. Harrison and Bigley (1982),
Gallagher et al, (1984) and Bayer (1996) established its expanding distribution along the Oregon
coast, and Posey (1988) documented the ecological influence of Z. japonica on estuarine benthic
communities. However, generalized habitat maps of Yaquina Bay (Cortwright et al,, 1987,
Gaumer et al., 1974, and Bureau of Sport Fisheries and Wildlife, 1968) do not recognize Z.
japonica as a distinct species. The area its communities now occupy was variously classified as
low salt marsh, algae on mud or mud; all "seagrass" subhabitats were located in subtidal or lower
intertidal areas (Cortwright et al., 1987). The two earlier mapping efforts were accomplished
before the presumed date of local invasion of Z. japonica, and lumped all SAV species into
"seagrass" or "algae" in very general areal delineations.
Distribution of Z. japonica is restricted in this estuary from mean higher high water
(MHHW) down to -2m above MLLW) in the intertidal zone, although we have occasionally
encountered it as isolated plants down to ~0.7m above MLLW, especially up bay in lower
salinity regimes with steeper mudflat gradients. Bayer (1996) characterizes the typical
distribution to be in linear patches parallel to the shore. They range from ~1 to -34 m wide,
typically are -10 m in length by ~3 m wide, but occasionally extend for several hundred meters,
with significant unoccupied stretches of bare sediment in between. Particularly in the lower bay,
Z. japonica is vertically separated from Z. marina communities by as much as several hundred
meters of vertically sloping bare sediment. In this estuary these occupied patches are perennial,
and the plants appear healthy year-round, although somewhat diminished in stature during the
winter months. Z. japonica seems to tolerate low tide exposures to freezing air temperatures (to
— 10°C) and prolonged inundation of rainfall runoff with no discemable effect; we have
observed both flowering and seed set locally in 1998 and 1999 (pers. obs.).
METHODS
The Yaquina River and estuary, located on the central Oregon coast, USA (described in
Young et a/., 1998), is characterized as a drowned river valley of -1,730 ha area, of which
-1,000 ha are intertidal mud flats (Fig. 1). Dominant SAV in Yaquina Bay are two eelgrass
species, Zostera marina and Z. japonica, and two green macroalgae, Ulva spp. and
Enteromorpha spp., with a number of other green, brown and red macroalgae (Phinney 1977).
These species occur partially or wholly in the intertidal mudflat zone, and daytime minus tidal
excursions during the summer months afford the opportunity for aerial photographic detection.
We conducted such aerial surveys in the summers of 1997, 1998 and 1999, observing C-CAP
protocols (Dobson et al. 1995) for sun angle, cloud cover and related criteria, substituting near
infrared color film instead of natural color emulsion. The resulting stereophotographs of 1997
were digitized to produce orthorectified images with a nominal ground pixel resolution of 0.20 m
(process described in Young et al, 1998; Young etal, 1999). We employed post-processed
differentially corrected GPS (DGPS, in UTM, NAD83) to locate community patches in the field,
with an average positioning accuracy of ±2.0 m; geodetic monuments for control were located to
±0.6m. Areas above MHHW (nominally emergent marsh and above) and below -1m above
-------
MLLW were masked out before applying generalized SAV and Zostera-specific detection
algorithms (Young etal., 1998; Clinton, etal., this vol.; Young, etal., this vol.).
We attempted to establish "user's and producer's accuracies" (sensu Congalton, 1991) by
independently locating known patches of Z. japonica habitat in the field (~ km 5, 6.5 and 13,
north shore, and km 6, south shore) and obtaining a DGPS position (± 2m). These points were
then located by GIS on the orthophotographs (resolution ±0.5m), and a determination was made
as to agreement with algorithmic detection using Arclnfo® software (Clinton et aL, this volume).
Figure 1. Yaquina Bay Study Site
i
| Newport
Yaquina
Bay Bridge
EPA LAB
Pacific
Ocean
RM 0 11 Idaho Point '/_,
11 /,-'
11 King Slouch
! i ^
I .•
If RiverBend-T
if
i{ Z. japonica
Sally's Bend
Z. japonica
jCoquille
Sawyer's
Landing Criteser's
Oregon Boone
Oyster Slough ,
Toledo
RM 1KS.
Craigie Pt.
Z. marina
Z. japonica
Yaquina Bay
Oregon
SAY CLASSMCATION
A combination of "heads up" digitizing, specific algorithm development, digital
bathymetry and GPS positioning of ground-truth data was used to detect Z. japonica from
adjacent communities and habitats (bare sediment, other algae and emergent marsh grasses) in
demonstation sites. Two algorithms were used, one for SAV in general and one for Z. marina.
RESULTS AND DISCUSSION
According to Congalton (1991), "user's accuracy" tells us how many of the pixels
identified as SAV on the image are really SAV on the ground, Le., how accurately is a given
-------
pixel classified? Errors of this type are of commission, or % false positives. "Producer's
accuracy" tells us how often SAV on the ground is correctly identified on the image as SAV, Le,,
how well is the area classified? Errors of this type are of omission, or % false negatives. Overall
accuracy tells us how many of all the stations (all types, including negative controls) are correctly
classified.
The most comprehensive published mapping ofZostera sp. in Oregon estuaries, and in
Yaquina Bay specifically, The Oregon Estuary Plan Book (Cortwright et a/., 1987) classified the
distribution of seagrass communities in Yaquina Bay as either subtidal or intertidal aquatic beds,
with further subclasses for mixed seagrass/algae and algae (Table 1.). Subhabitats classes of
"flat" and "shore" do not provide for the occurrence of seagrass - the only SAV provision near
the MHHW elevation is for "algae." Z. japonica is not recognized as an existing species, much
less mapped. The total area of Yaquina Bay classified as seagrass at the time of survey in all
subhabitats was -570 acres (-230 ha), with a further -150 ac. (-61 ha) classified as seagrass-
algae. Subhabitats area totals were estimated to within -0.1 acre, and mapped on a scale of
-1:37,000. Bulthuis (1995) has mapped seagrass areal distribution (to include Z. marina, Z.
japonica, Ruppia maritima, and the algae Ulva sp. and Enteromorpha sp.) in Padilla Bay, WA
(Puget Sound), using color aerial photographs at 1:12,000 on a minus tide and extensive ground-
truthing effort, and delineated with a transfer scope by hand in units of 0.1 ha and larger.
Table 1. Aquatic vegetation in Yaquina Bay, by subclass (adapted from Cortwright, et ai, 1987)
Subclass
Estuary area
Seagrass (subtidal)
Algae (subtidal)
Seagrass (lower intertidal)
Seagrass/algae (lower intertidal)
Algae (lower intertidal)
Low salt marsh
High salt marsh
Area (acres - hectares)
-4,300 --1,730
44.7- -18.1
5,9 - -2.4
525.1 - -212.5
152.4- -61.7
125.4- -50.7
143.8 - -58.2
475.3- -192.4
% of estuary area
100.0
1.0
0.1
12.1
3.5
2.9
3.3
10.9
We originally designed algorithms for near-infrared color imagery to detect Zostera
marina specifically, and SAV (in general) on tidally exposed mudflats. We employed these
algorithms to detect Z. japonica and other vegetation or bare sediment patches in the upper
intertidal areas adjacent to the MHHW line. We evaluated 39 data points, classified as sparse,
medium and dense cover, to establish "user's" and "producer's" accuracy (Table 2). The SAV
algorithm correctly classified known Z. japonica patches (i.e., an identified pixel was within 2 m
of a field-identified patch) with an accuracy of 84 % in this test; with an overall classification
-------
accuracy of 72 %. The Z. manna-specific algorithm had a Z japonica-specific accuracy of 73%,
and an overall accuracy of 56 %.
Table 2. Algorithm classification accuracy assessment using ground-truth data.
Zostera marina algorithm
classified as Z japonica
classified as not Z japonica
Ground-truth stations
Zostera
japonica
(22)
8
14
marsh grass
(7)
0
7
brown
algae
(3)
3
0
Filament.
green algae
(1)
0
1
bare
sediment
(6)
0
6
algorithm
classification
totals
11
28
User's accuracy (# correctly classified as Z. japonica 1 total # identified by algorithm as Z japonica)
Producer's accuracy (# correctly classified Z japonica I total # ground-truthed Z japonica)
Overall accuracy (Total # stations correctly classified / total all stations)
73%
37%
56%
SAV algorithm
classified as Z japonica
classified as not Z japonica
Ground-truth stations
Zostera
japonica
(22)
16
6
marsh grass
(7)
0
7
brown
algae
(3)
3
0
Filament
green algae
(1)
0
1
bare
sediment
(6)
0
6
algorithm
classification
totals
19
20
User's accuracy (# correctly classified as Z. japonica I total # identified by algorithm as Z japonica)
Producer's accuracy (# correctly classified Z japonica / total # ground-truthed Z japonica)
Overall accuracy (Total # stations correctly classified / total all stations)
84%
73%
77%
Although Z. japonica occurs in considerably smaller average patches than Z. marina, its
occurrence was verified by remote aerial sensing using methods developed for this survey. Our
field verification showed that 77 % of all test community patch types detected by the generalized
SAV algorithm were correctly identified (i.e., within 2 m of an algorithm-specified pixel), while
the Z. marine-specific algorithm was more conservative with respect to Z. japonica patches,
with an overall accuracy of 56 %. The Z. manna-specific algorithm may not have worked as
well for Z. japonica because of the considerable difference in algal epiphyte coverage of leaf
blades (at the optimal time for obtaining this imagery, Z. marina leaf blades would typically have
a dense coverage of epiphytes, increasing distally; Z, japonica tends to have very little). Thorn
(1990) assigns -50% of net above-ground primary production in the Padilla Bay system to
epiphytes on Z. marina; one would expect the spectral signature to differ significantly as a result.
With further development, these algorithms should allow delineation of the Z. japonica
-------
patches with accuracy sufficient to allow areal estimation of the communities at a considerably
finer scale than exists now, and contribute to the knowledge of changing habitat structure caused
by its introduction and spread (Posey, 1988, Baldwin and Lovvorn, 1994).
ACKNOWLEDGMENTS
Ground reference surveys were conducted by employees of DynCorp/TAI, Inc., stationed
at Newport, OR, and EPA scientists and staff of the Western Ecology Division's Coastal Ecology
Branch. The aerial photography was conducted by Bergman Photographic, Inc. (Portland, OR);
the orthoreetified digital photographs were produced by Photogrammetric Digital Services, Inc.
(Eugene, OR). We thank S. Rumrill, S. Cline, L. Brophy and R. Bayer for constructive reviews.
REFERENCES
J.R. Baldwin and J. R. Lovvorn, "Expansion of seagrass habitat by the exotic Zostera japonica, and its
use by dabbling ducks and brant in Boundary Bay, British Columbia," Mar. Ecol. Progr. Ser.,
Vol. 103, pp. 119-127, 1994.
R.D. Bayer, "Intertidal zonation of Zostera marina in the Yaquina Estuary, Oregon," Syesis, Vol. 12,
pp.147-154, 1979.
R.D. Bayer, "Macrophyton and Tides at Yaquina Estuary, Lincoln County, Oregon," J. Oregon
Ornithology, No. 6., 1996.
R.E. Bigley, and J.L. Barreca, "Evidence for synonymizing Zostera americana den Hartog with Zostera
japonica Aschers. & Graebn," Aquat. Bot., Vol. 14, pp.349-356, 1982.
D.A. Bulthuis, "Distribution of seagrasses in a north Puget Sound Estuary: Padilla Bay, Washington,
USA," Aquat. Bot. Vol. 50, pp. 99-105, 1995.
Bureau of Sport Fisheries and Wildlife, "Preliminary Survey of Fish and Wildlife in Relation to the
Ecological and Biological Aspects of Yaquina Bay, Oregon," Fish and Wildlife Service, U.S.
Department of the Interior, Portland, Oregon, 23 pp., maps, appendices, 1968.
P. Clinton, D. Young, B. Robbins and D. Specht, "Issues in Digital Image Processing of Aerial Photo-
graphy for Mapping Submersed Aquatic Vegetation," Proceedings: Sixth International
Conference on Remote Sensing for Marine and Coastal Environments., Charleston, SC, 1-3 May
2000, this vol.
R.G. Congalton, "A review of assessing the accuracy of classifications of remotely sensed data," Remote
Sensing of the Environment, Vol. 37, pp. 35-46, 1991.
R. Cortwright, J. Weber and R. Bailey, "The Oregon Estuary Plan Book," Oregon Department of Land
Conservation and Development, Salem, OR. 126 pp, ill., charts, maps, 1987.
J.E. Dobson, E.A. Bright, R.L. Ferguson, DW. Field, L.L. Wood, K.D. Haddad, H. Iredale m, J.R.
-------
Jensen, R.J. Kelmas, R.J. Orth, and J. P. Thomas, "NOAA Coastal Change Analysis Program (C-
CAP): Guidance for Regional Implementation," NOAA Technical Report NMFS 123. U.S.
Department of Commerce, 1995.
J.L. Gallagher, H.V. Kibby and K.W. Skirvin, "Detritus processing and mineral cycling in sea-grass
(Zostera) litter in an Oregon salt marsh," Aquatic Botany, Vol. 20, pp. 97-108, 1984.
T. Gaumer, D. Demory, L. Osis and C. Walters, "1970-71 Yaquina Bay Resource Use Study," Fish
Commission of Oregon; US ACE Contract No. DACW 57-72-C-0138; NOAA NMFS PL 88-309
Contract Nos. N208-0073-72(N) and N04-3-208-55, p. 30, 1974.
P.O. Harrison and R.E. Bigley, 'The recent introduction of the seagrass Zostera japonica Aschers. and
Graebn. to the Pacific Coast of North America," Can. J. Fish. Aqua:. Sci. Vol. 39, pp. 1642-
1648, 1982.
M.E. Kentula and C.D. Mclntire, "The autecology and production dynamics of eelgrass (Zostera marina
L.) in Netarts Bay, Oregon," Estuaries, Vol. 9, No.3, pp.188-199, 1986.
R.C. Phillips and E.B. Menez, "Seagrasses," Smithsonian Contributions to the Marine Sciences, Number
34. Smithsonian Institution Press, Washington, D.C. 104pp., 1988.
H.K. Phinney, "The Marine macrophytic Algae of Oregon," In: R.W. Krauss, (Ed.) The Marine Plant
Biomass of the Pacific Northwest Coast. Oregon State University Press, Corvallis, OR. pp. 93-
115, 1977.
M.H. Posey, "Community changes associated with the spread of an introduced seagrass, Zostera
japonica" Ecology, Vol. 69, No. 4, pp. 974-983, 1988.
R. M. Thorn, "Spatial and Temporal Patterns in Plant Standing Stock and Primary Production in a
Temperate Seagrass System," Bot. Mar., Vol. 33, pp. 497-510, 1990.
D.R. Young, S. Cline, D. Specht. B. Robbins and J. Lamberson, "Mapping Spatial/Temporal
Distributions of Green Macroalgae in a Pacific Northwest Coastal Estuary via Small-Format
Color Infrared Aerial Photography," Proceedings: Sixth International Conference on Remote
Sensing for Marine and Coastal Environments., Charleston, SC, 1-3 May 2000, this vol.
D.R. Young, D.T. Specht, P.J. 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.," Proceedings, Fifth International Conference on Remote Sensing for Marine
and Coastal Environments, San Diego, California, 5-7 October, 1998, 1998.
D.R. Young, D.T. Specht, B.D. Robbins and P.J. Clinton, "Delineation of Pacific Northwest SAVs From
Aerial Photography: Natural Color or Color Infra-red film?," Proceedings: From Image to
Information. 1999 ASPRS Annual Conference, Portland, OR, May 17-21, 1999, 1999.
-------
Filename: c:\wordp\mss\eco\zzjapnmsq.wpd
final revision, submitted copy: 2/2/00
plus addendum: worksheet for Table 2
Actual GPS field station identification on screen (post-processed jobs Z012021A, Z012023A,
Z012023E, Z012100A, Z012101A)
Z. marina
algorithm
is japonica
IK not
japomcu
disallowed
(in shadows,
etc.)
SAV
algorithm
is japonica
is not
japonica
disallowed
(shadows,
etc.)
Z, japonica
± 13,34,43
m: 8, 10,26,33,
(21, @2.7m)
sp:
d: 38,39,41,42
ir: ',9,15,16,17,
23,25,28,29,32
sp:
6 (zj, sp)
36 (zj, d)
37 (zj, d)
d: 13,34,38,39,
41,42,43
m: 8,10,17,25,
26,28,29,33(21)
sp:
d:
m: 7,9,15,16,
23,32
sp:
6 (zj, sp)
36 (zj, d)
37 (zj, d)
Marsh grass
1,2,4,18,22,
27,35
1,2,4,18,22,
27,35
brown algae
12,30,31
12,30,31
filgr
algae
19
19
bare sediment
5,11,14,20,24,
40
5,11,14,20,
24,40
3
Note: # 27 is dense marsh grass - double classified on GPS file by keying error.
-------
TECHNICAL REPORT DATA
(Please read instructions on the reverse before completing)
1. REPORT NO, 2.
EPA/600/A-00/027
4. TITLE AND SUBTITLE Near infrared aerial photo-detection of Zosters Japonica
communities in Pacific Northwest estuarine intertidal habitats
7, AUTHOR{Sl David T, Specht1, David R, Young1, Patrick J. Clinton2
9. PERFORMING ORGANIZATION NAME AND ADDRESS
'Coastal Ecology Branch 2OAOCorp
US EPA NHEERL WED US EPA NHEERL WED
2111 SE Marine Science Drive 2111 SE Mariene 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.
13. TYPE OF REPORT AND PERIOD
COVERED
14. SPONSORING AGENCY CODE
EPA/600/02
15. SUPPLEMENTARY NOTES:
1 6. ABSTRACT: Near infrared color aerial photography (-1 :7200) of Yaquina Bay, Oregon, flown at minus tides during summer months of 1 997
was used to produce digital stereo ortho-photographs covering tidally exposed eelgrass habitat. GIS analysis, coupled with GPS positioning of
ground-truth data detected Zostera japonica communities (non-indigenous eelgrass), which are physically separated by elevation in this and
similar Pacific North-west coastal estuaries from Zostera marina (native eelgrass) communities. The non-indigenous Z, japonica typically occurs
at or near mean high water while native Z. marina is restricted to -0.7 m above mean low low water and below. Recognition of Z. japonica
patches from adjacent bare sediment and emergent beach or marsh grasses was aided by combining specifically tailored detection algorithms
and "heads up" digitizing with digital bathymetry. Further progress in algorithm development should allow areal delineation, mapping and
change analysis.
1 7. KEY WORDS AND DOCUMENT ANALYSIS
a. DESCRIPTORS b. IDENTIFIERS/OPEN ENDED
TERMS
GIS, near infrared, estuaries, Pacific
Northwest, aerial photography, Zostera
japonica, mapping.
1 8. DISTRIBUTION STATEMENT 1 9. SECURITY CLASS (This Report}
20. SECURITY CLASS (This page)
c. COSATI Field/Group
21. NO. OF PAGES: 8
22. PRICE
EPA Form 2220-1 (Rev. 4-77J PREVIOUS EDITION IS OBSOLETE
------- |