ALAN AUSTIN ROBERT ADAMS University of Victoria Victoria, British Columbia vaw 2Y2, Canada Aerial Color and Color Infrared Survey of Marine Plant Resources

Natural color proved to be the most useful for definition of submerged vegetation while color infrared and natural color together provided the best definition of above-water seaweed vegetation in an assessment carried out in the Georgia Strait area of the north-east Pacific.

INTRODUCTION velopmental pattern, such pre-use surveys are even more vital. NCREASING DEMANDS upon seaweed as raw As a result of the need for base-line infor­ I material for the growing extractives indus­ mation on marine plant resources of British try has indicated the need for gathering ac­ Columbia in the nOlth-east Pacific, the cur­ curate and complete inventory data upon rent program was designed to research, test, abundance and harvestability of pmticular and apply a methodology for aerial mapping seaweed (macroalgal) species. Rapid and of seaweed communities and inventory of

ABSTRACT; Color, color infrared, and water penetration aerial photographs at various scales were assessed for identification, map­ ping, and inventory ofmacroalgal vegetation along extensive shore­ line ofGeorgia Strait in the north-east Pacific. Natural color proved to be the most useful for definition of submerged vegetation to depths of 7 m, while CIR and natural color together proVided the best definition ofabove-water intertidal seaweed vegetation. Using both films, exposed under rarely occurring optimum weather and tide conditions, and with the aid ofground data, a total of11 vegeta­ tion units were classified and mapped at a scale of I: 10,000. Bound­ aries of the vegetation unit containing a valuable red seaweed re­ source, Iridaea cordata, could be defined equally well at a scale of 1:10,000 as at 1:2,500. A ground truth program was designed which revealed a close relation between the aerially mapped I. cordata veg­ etation units and field-observed vegetation units. Air photography with complimentary ground data collection was found effective in establishing baseline data for seaweed resource use, management, and conservation, and will be of value in environmental impact as­ sessment.

economical base-line surveys of wild plant selected valuable species. The coast of resources are a prerequisite to policy deci­ British Columbia is deeply indented, in ex­ sions that ensure use is continued on a sus­ cess of 25,000 kilometers, and is fringed by tained yield basis. With seaweed resources, rich populations of marine plants. Early virtually unknown in distribution and de- shipboard seaweed surveys were inaccurate

PHOTOGRAMMETRIC ENGINEERING AND REMOTE SENSING, 469 Vol. 44, No.4, April 1978, pp. 469-480. 470 PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING, 1978 and dealt only with the larger iloating kelps photography to map algal communities at a visible on the surface. scale of 1: 1,200 along short sections of coast Aerial remote sensing, with a variety of in northern \,yashington. color and color infi'ared (CIR) film, has been Little use has been made of aerial photo­ extensively used in surveys of terrestrial graphy for quantitative inventory of sea­ vegetation for quantifying distribution, weeds and it was apparent that further de­ biomass, and even plant vigor. Aerial photo­ tailed studies were required prior to exten­ grammetry has been used to survey aquatic sive aerial survey work in the north-east vegetation in lakes (Lukens, 1968; Vary, Pacific. The major objectives of the current 1971) and considerable attention is now work were to assess various films at several being paid to aerial survey of inland and photoscales for use in mapping seaweeds, to coastal marsh vegetation (Anderson and evolve systematic methods for ground data Wobber, 1973; Seher and Tueller, 1973; collection complimentary to the photo imag­ Klemas et ai., 1974; McEwen et ai., 1976). ery, and to evaluate the accuracy of these Similar surveys of rocky seashore, inteltidal, procedures for resource inventory. The de­ and subtidal vegetation appear to be un­ tailed mapping ofsuch sensitive and produc­ common. Difficulties inherent in such sur­ tive vegetation will provide base-line data veys inc:lude the imaging of small and vari­ for resource use and management and will ably colored species, which occur on irregu­ also serve to indicate the effects oflong term lar terrain in heterogeneous bands of vari­ pelturbations impinging from both landward able density and width. The vegetation is and seaward. uncovered and covered to various depths depending upon the tides, thus severely METHODS limiting accessibility and photography. Through-water photography requires atten­ AERIAL PHOTOGRAPHY tion to problems associated with specular re­ Mapping and inventory work was under­ f1ectance, water transparency, spectral taken in the Strait of Georgia, British Col­ transmittance, and scene reflectance. umbia, where commercially valuable red However, a limited number of experimen­ algae were known to occur. The area (Figure tal studies have been made to assess and de­ 1) has a characteristic algal flora determined velop remote sensing as a tool in marine by the enclosed nature of the water mass, plant reconnaissance. Such studies were first sheltered from open Pacific storms and documented in Scotland (Walker, 1950, swells. Gradually sloping shores provide 1954) and lova Scotia (Cameron, 1950). Walker's lengthy (1946-1953) survey esti­ mated quantities of Laminaria, a seaweed valuable for alginate extraction, and made observations on seasonal biomass fl uctua­ tions but gave no technical photogrammetric data and used photography only to differen­ tiate seaweed beds from sandy areas. Three specific communities \\.~re differentiated, using minimal ground truth survey, in Cameron's (1950) survey, where the use of various filters constituted a pioneer, ifprimi­ tive, attempt at multispectral photography. Dubois (1964), working on the French coast, made an interpretive key to imagery of sev­ eral subtidal algallseagrass associations and more recently Kelley and Conrad (1969), ...... 5 fathom using both aerial and satellite photography, ___ photo distinguished characteristics of seven com­ coverage o 2 .. 6 munities on soft bottoms on the Bahamas ~ Banks. Vadas and Manzer (1971) described MILES aerial and crR photography of intertidal biota affected by thermal effluents but used only small-format, hand-held cameras. The state ofremote sensing imagery, as it applies to intertidal vegetation, was reviewed in FIG 1. Map ofarea in Georgia Strait, British Col­ 1972 by Jamison who used large-format color umbia, encompassed by seaweed inventory. SURVEY OF MARINE PLANT RESOURCES 471 wide inteltidal expanses (Plate 3), which GROU:\lD DATA COLLECTIOI\ SUppOlt populations of benthic algae on sub­ Interpretation of photographic imagery strate composed primarily of surficial cob­ was accomplished using ground data col­ b~' bles and rocks in sand, deposited glacial lected for the definition of seaweed com­ action, with some protruding granitic bed­ munities. Vegetation was surveyed along rock. :\Iaximum vertical tidal range is 6 m. belt-transects to f~lcilitate both the classifica­ Along this coast two sensing flights, ac­ tion of vegetation units based on species companied by ground data collection, were composition and the location oftheir bound­ undertaken, one in 1972 and one in 1974. aries on air photographs used to produce The first exploratory (1972) flight used vegetation maps. A total of 22 transects, av­ natural color, CIR, and panchromatic B/vV eraging 500 m in length, were surveyed from film, at various altitudes. Aerial photographs Mav through August 1972 and 30 such tran­ of 23 cm format were taken at times, specifi­ sects were surveved ~Ia~' through August cations, and weather conditions shown in 1974, to accompany photography in respec­ Table 1. The photography was by ~1achair tive years. Prior to field work, transects were Company (Calgary) using a Wild RCB cam­ positioned at right angles to high water on era with a 15 cm f(lcallength lens, along a 45 existi ng standard B/W aerial photographs. km "test" coast (Elma Bay to Denman Is­ The transects ran seaward, at intervals along land, Fig. 1) at a height of 1500 m (approxi­ the coast, on a variety of shore types having mate photo scale = 1:10,000) with 60 percent different substrates, exposures to wave ac­ forward lap. Color and CIR were exposed at tion and currents, and degrees of steepness, additional scales of 1:5,000 and 1:2,500 in embracing most major environmental condi­ two 2.5 km test areas within this region. tions for seaweed growth. The second (1974) photo mission was un­ In the field, the distance between bound­ deltaken to extend the photo coverage to all aries of distinct seaweed communities were coastal areas in northern Georgia Strait recorded, the bottom contour mapped, and where dense beds of selected commercially seaweed samples collected along each valuable red seaweeds could potentially o~­ transect at ,50 m intervals. Position of tran­ cur. Conditions of exposure also are shown sects in the field was accurately recorded in Table 1. Films and photo scales proven with the aid ofa field compass and sextant so best for definition of this seaweed in the that they could later be precisely located on previous tests were used, and a total of52 km the aerial photographs. of CO,lSt were photographed with Kodak vVhile such observations are commonplace aero-negative 2445 film. On the B.C. main­ and relatively easily obtained in terrestrial land shore near Powell River additional surveys (Kuchler, 1967), much of our data photography, for comparison with the collection was carried out underwater and natural color film, was taken using Kodak presented severe difficulties. When possi­ experimental color film for water penetra­ ble, transect observations were made by tion which has two emulsion lavers covering wading or using wet suits in shallow water at the maximum water transmission spectra low tide. Observations were continued by (Specht et ai., 1973). SCUBA divers into the subtidal, where TABLE 1. AERIAL PHOTOGRAPHY EXPOSURE PARAMETERS IN SEAWEED SURVEY FLIGHTS, 1972 AND 1974

DATE: August 6 & 7,1972 Tll'dE: 0830 - 1000 P.S.T. CLOUD: sunny, light high altitude haze WIND: 5 mph TIDE: +0.6 m above ehart datum Film Type f. Stop Exposure

Aeroeolor 2334 5.6 1/400 C.I.R. 2443 5.6 11200 B/W 5.6 1/500

DATE: Julv 21,1974 TIME: 1200 - 1300 P.5.T. .. CLOUD: broken eumulus, 40% overeast WilD: 7 mph TIDE: +0.7 m above chart datum Film Tvpe f. Stop Exposure

Aeroeolor 2445 5.6 1/250 Kodak Water Penetration 5.6 1/200 472 PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING, 1978 weighted buoys were placed every 50 m out H.W.L. features discernable on the air to a maximum depth of 10 n1. A collapsible photographs and the boundaries ofthe major metal quadrat frame was dropped by a boat vegetation units and the +0.6 m water-line operator mid-way between the 50 m buoys contour were traced from the photographs and two divers were towed to a given site. onto the acetate sheet. vVhereas these maps Samples of all macroscopic seaweeds (ex­ were uncontrolled, they were sufficiently cept the smallest filamentous and encrusting accurate for the representation of seaweed forms) and notes on abundance were taken, vegetation units and have been prepared 2 within the large quadrat (10m ), such that similarly for lake vegetation (Lukens, 1968). divers in low visibility waters explored as­ Higher precision in drafting boundaries was sociations sufficiently large to be resolved not warranted because ofthe in herent grada­ on the aerial photographs. The depth and tion observed between many seaweed com­ substrate type at each quadrat site, the visi­ munities. ble changes between sample quadrats, and the distances between distinct vegetation FIELD VERIFICATION OF INVENTORY ESTIMATES boundaries and the nearest buoy were also One species of commercially valuable recorded for identification on aerial photo­ mari ne plant was found in sufficient abun­ graphs. dance and density to warrant inventory for Actual depth, relative to chaIt datum, was commercial interests. This was a leafy calculated for each sample site, as was the (0.lm2 ) red alga, Iridaea cordata (Turner) slope of shore between successive quadrats Bory, yielding an extract used widely in food and the distances of vegetation boundaries processing and pharmaceutical products. fi'om high water. Scale profiles of each tran­ The number of hectares occupied by dense sect were drawn (Figure 2) showing sample 1. cordata was measured planimetrically sites, the major vegetation zones, and tide hom the photographically prepared maps. levels. Distances on these profiles fi'om high Additional ground data were required, how­ water to each vegetation boundary were ever, to estimate the accuracy of this area transferred to the survey photographs. measurement and to make subsequent ad­ Phytosoc:iologic:al methods of community justments in order to derive a final estimate analvsis commonly used in terrestrial work ofthe area covered by dense I. cordata. Such wer~ applied to the seaweed species data. methods are used for inventories of terres­ The quadrat species lists were compiled trial crops as outlined by Benson et aZ. (1971). with an IBM 360 computer, and a Zurich Ground truth data were sought to deter­ I\lontpellier sOlting method was used which mine the relation between mapped bound­ simultaneously groups species which co­ aries of the 1. cordata vegetation unit occur in sample quadrats and groups sample (LV.D.) and field-observed LV.D. bound­ quadrats with co-occuring species (Ceska aries. The mapped coastline of Georgia and Roemer, 1971). The final sOlted associa­ Strait was divided into roughly 2 km strata tions were used to help define and classify (sections) and within each, 3 to 10 transects, the vegetation units observed on aerial running nearly perpendicular to shore, were photographs prior to mapping. randomly positioned. Each transect was lo­ cated to run through two clear landmarks MAPPING (e.g., a house near high water and a large Seaweed vegetation maps for the entire boulder in the intertidal) which could be study were prepared at the same scale identified both on aerial photographs and in (1: 10,000) as the survey aerial photographs. the field. The distances from the landmarks Source maps for the survey coast were to the landward and seaward LV.U. bound­ photographically enlarged to the desired aries were measured along each survey 1: 10,000 scale and the high water line transect, both on the ground using survey in­ (H.W.L.) was traced fi'om the source-map struments (± 1 percent error) and on the onto transparent matte acetate. Boundaries maps using dividers (± 3 percent average er­ ofthe vegetation units, the +0.6m water line ror). Because the 1. cordata vegetation units contour (low water line) and the H.\V.L. were long and narrow, parallel to the coast, debris-or storm-line were, with the aid of a the widths of the mapped LV.D. and field­ stereoscope, outlined on the aerial photo­ observed 1. V. U. were considered propor­ graphs. These photographs (color and CIR, tional to their respective areas. The ratio of 1: 10,000) were laid one after another under­ these widths was used to adjust the mapped neath the acetate sheet, matching H.vV.L. as area by the formula closely as possible. The tracing from the source-map was corrected for the smaller A." = Alii X R SURVEY OF MARINE PLANT RESOURCES 473 where A" is the adjusted area estimate, Alii is seaward edge ofthe intertidal and the upper the LV.U. area measured fi'om the maps, and portion of the subtidal marine vegetation. R is the ratio of the field-observed width Dominant species and vegetation zones are measurements to the mapped width mea­ illustrated in Figure 2. A ground level illus­ surements. The variance of R provided an h'ation of a red algal community, at low tide estimate ofthe accuracy ofthe adjusted area. in Kye Bay, dominated by Iridaea cordata The widths ofmapped and field-identified adhering to cobble substrates is given in LV.U. and also I. cordata biomass were Plate 3. A total of eleven major vegetation sampled along a total of58 randomly located units between high water and -7 m were transects in 11 coastal strata. An average of identified and mapped along the 45 km test two to three transects were surveyed per day coast in Georgia Strait. by a crew of three SCUBA divers and a total Color and CIR photographs were infin­ of 25 days were required to complete the itely superior to black-and-white (B/W) ground truth transects in this survey. Biolog­ photographs for definition of the vegetation ical observations and biomass data have units both above and below the sea surf~lce. been reported elsewhere (Austin and 'vVith B/\"; films lIsed during surveys by Adams, 1973, 1974, 1975) and will not be in­ Cameron (1950) and Walker (1954), bound­ cluded here. aries of beds were apparently noted to

RESULTS AND DISCUSSION depths of7 111. Similar depth penetration was achieved in our B/W prints but again only COLOR AND COLOR INFRARED PHOTOGRAPHY the boundary characteristics ofplant associa­ Results of the exploratory study in 1972, tions could be determined. using large format color and CIR photog­ A comparison of color and CIR for raphy taken at a scale of 1:5,000, can be above-water imagery indicated that neither seen in Plates 1 and 2 of Kye Bay, Van­ one was superior for definition of maximum couver Island. In this test area the cobble numbers of communities, but they could be and sand shore was very extensive, having used together with advantage. In some shore an intertidal width of 200-1200 m, and the areas the vegetation un its dominated by photography at this scale imaged only the brown species, especially Sargasslim

~ 0 ~

'"> -n ~ '" 0 '"~ -2 sample plot vegetation unit o ~ • pos; tions -4 boundar; es ~ 0- -6 0'" . 0 50 100 150 200 250 300 350 400 450 Distance (m.) a10n9 transect from conspicuous landmark (bolder)

!!!. Ul va fenes tra. Crypt. wood; i Iridaea cord. 5 Iridaea cord. 4 Iridaea cord. 2 lamin. sacchar. ~ Crypt. wood;; Iridaea hetero. Ploc. coccineum 1 Ploc. coccineum 1 Plac. coccineum 4 Alarla tenuif. ~ Prion. lyal1i Prion.lyalli; Iridaea hetero. 1 Prion. lyallii 1 Prian.lyallii 1 Castaria cost. c: Iridaea cord. Iridaea cord. Cons t. subu1if. 1 Cymathere trip. Polyneura lat. Iridaea hetero. 1 ~ Crypt. woodi i 1 '"u ,----'-----, ,-----'-----, 0- '" Crypt. woodi i Sargasso mut. Iridaea cord. 4 Iridaea cord. 4 lamin. sacchar. 5 Ul va fenes tra. Iridaea cord. Lamin. sacchar. 4 P1oc. caccineum 1 iridaea cord. 1 Iridaea hetero. Plac. coccineum 2 Prian.lyallii 2 Ploc. ca.:cineum 1 Po lyneuro 10 t. PriDn.lyonii r CDnst. sublif. 2 PriDn.lyonii r Botryog. rup. Canst. subul if. 1 Rhod. pertusa 1 Iridaea hetera. 1 Iridaea hetero. 1 FIG 2_ Typical profile of survey transect (Kye Bay) showing depth and distance scales, quadrat sample locations, cover ofdominant species (abbrev.) in sample plots and vegetation boundaries, used as an aid to photo-interpretation. Vegetation Units: A = Ulva spp_ B = Cryptosiphonia woodii C = Sargassum muticwll 0 = Iridaea cordata E = Laminaria groenlandica. Species cover values; I = <5%; 2 = 5-25%; 3 = 25-50%; 4 = 50-75%; 5 = 75-100%. 474 PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING, 1978

PLATE 1. Portion of natural color (2445) print of intertidal and subtidal vegetation at low tide (Kye Bay, Vancouver Island) at scale of 1:5,000. Overlay indicates survey transect location and seaweed vegetation units defined fi'om ground data. Legend: e = Cryp­ tosiphonii woodii I = Iridaea cordata L = Laminaria groenlandica S = Sargassllm muticU11l U = Ulva spp. Dashed line = water line (+0.6 m).

PLATE 2. POItion ofcolor infi'ared (2443) print of transparency of same area as Plate 1 at scale of 1:5,000. Tide is at same level but elR film renders all sub-water surface vegeta­ tion dark blue. Refer to legend of Plate 1 for above water vegetation units. SURVEY OF MARINE PLANT RESOURCES 475

PLATE 3. Low intertidal shore area in Kye Bay with dense Iridaea cordata seaweed population adhering to cobble substrate.

PLATE 4. POItion of natural color print of B.C. mainland shore (near Powell River, B.C.) at low tide for comparison with water penetration film (Plate 5). Maximum depth penetration = 5 meters. Scale = 1:10,000. Yellow color is enhanced.

PLATE 5. Portion of Kodak experimental water depth penetration print made from transparency. Same location as Plate 4 for comparison. 476 PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING, 1978 muticum and Fucus distichus, were more creased ease in measuring the areas occu­ readily distinguished on the CIR photo­ pied by these vegetation units. Photo-scales graphs as were living seaweeds in areas smaller than 1: 10,000 would probably show where sparse populations (especially Fucus insufficient detail judging from our re­ distichus and VIva spp.) contrasted little sults, particularly when the algal com­ with underlying substrate on the color munities are narrow, such as found on more photographs. Vadas and Manzer (1971) also steeply sloping shores. FUlthermore, photo­ noted that CIR was superior for separating graphs of the study area at a scale of 1:5,000 Phaeophytes (brown algae) from Rho­ often recorded only a pOltion of the shore dophytes (red algae). In some areas of our between high water and the lower limit of study, however, red species such as the Cryp­ depth penetration. In such areas imagery of tosiphonia woodii vegetation unit were more the entire shore at a scale of 1:5,000 or larger readily differentiated from neighbouring would require two or more flight paths side vegetation on the natural color photographs. by side with standard 30 percent sidelap, in­ Submerged seaweeds could not be dif­ volving considerably more time, film, and ferentiated on CIR taken in Georgia Strait, cost than a single 11 ight at a higher altitude. A and the boundaries of vegetation units on further serious limitation of photography at these photographs could only be recognized larger scales was the difficulty of providing in shallow water when surrounded by a re­ ground control for tlights made entirely over flective substrate such as sand. The shallow water areas with no coastline or high maximum depth at which such boundaries water line for reference. could be discerned was 3 to 4 m, while Presence of scattered cumulus cloud be­ across much of the CIR photograph little or tween the sun and camera severely reduced no penetration was observed. The surface tex­ seaweed community definition along por­ ture and color of seaweeds under water tions of coast in the 1974 photographs. In could not be discerned even in beach areas these areas, contrast between seaweed where there were very shallow drainage communities was poor and maximum ob­ rivulets down the shore, much less in deeper served depth at which highly contrasting ob­ pools where only sand could be differen­ jects could be defined was only 1 to 2 m tiated from the dark blue colored algae. In belovv the water surhlce. Additional factors this respect, CIR was good for locating the observed to reduce seaweed definition in level of the tide and for locating areas of some photographs, even in cloud-free areas, drainage. The floating seaweeds, Sargassum were water turbidity and incorrect color en­ muticum and Nereocystes Iuetkeana, were hancement in processing. More contrast be­ very conspicous on CIR. However, in this tween seaweed communities was obtained survey, the lack of penetrance observed on if yellow-red hues were enhanced during CIR photographs severely limited their use­ printing. 'Water transparency was greatest on fulness for subsurface definition. an outgoing tide and also varied daily with The natural color film was found to give storms and plankton blooms. much better depth penetration in this sur­ A comparison of the natural color photo­ vey, defining object~ at depths of 7 m below graphs taken in cloud-hee areas with Kodak the water surface. At depths to 2 m some red water penetration film photographs (Plates 4 colors could still be discerned, and between and 5) revealed little differences in the 2 m and 7 m surface texture and boundary maximum depth (4 to 5 m) at which highly characteristics of plant associations could be reflectant objects, such as sand, could be de­ seen. It is important that this color film be fined below the water surface. However, the overexposed by one to two stops for greatest contrast between different algal vegetation penetration (Anderson, 1971; Jamison, types and also between algal beds and bare, 1972). sandy bottoms was very poor on the water Comparisons of photo-scales revealed penetration film, both above and below the considerably more land form detail at the water surhce. On this film, all features large scale (1:2,500) than at the smaller test below the water surface retained a rather uni­ scale (1: 10,000), but the depth penetration form, dull green tone which varied only and definition ofvegetation associations was slightly in density between the different only slightly better at the 1:2,500 scale. The seaweed populations. In comparison, the distinctness of boundaries was not particu­ 2445 film showed a color range from deep larly better at the 1:2,500 scale than at the blue to light red on below-surface details. 1: 10,000 scale due to the graduation inher­ FUIthermore. above-water features showed a ent between our vegetation units. The main wider range of colors, while the water pene­ advantage of the larger scale was the in- tration film was limited to shades of green SURVEY OF MAlUNE PLANT RESOURCES 477

TABLE 2. MEAN NUMBER OF DAYS ± S.E. PER YEAR WHEN DAYLIGHT Low TIDES ARE COINCIDENT \-\lITH SUITABLE WEATHER FOR PHOTOGRAPHY (WINDS"" 10 MPH AND CLOUDS < 1110 CIRRUS AND 2/10 Low CLOUD lOT OBSCURING SUN) IN GEORGIA STRAIT, B.C. CALCULATED FOR PERIOD 1966 TO 1975.

Selected Restrictions 1ay June July Aug. Total per Year

Tide level < + 1.0 m 1.1 ± 0.4 1.4 ± 0.5 2.1 ± 0.5 1.3 ± 0.3 5.9 ± 1.0 sun angle 35° to 60° Tide level < + 0.5 m 1.4 ± 0.2 0.7 ± 0.3 0.8 ± 0.3 0.3 ± 0.2 2.2 ± 0.4 sun angle 35° to 60° Tide level < + 1.0 m and 0.4 ± 0.2 0.5 ± 0.2 0.9 ± 0.4 0.9 ± 0.4 2.1 ± 0.5 sun angle 45° to 50°

and blue with some pink tones. Thus, both VEGETATION MAPS color and density characteristics could be By using the Zurich Montpelier computer used to greater advantage in defining and sorting technique, a total of ten species identifying seaweed populations, above and groups were diHerentiated which we consid­ below the water surface, on the Kodak 2445 ered to correspond closely with our intui­ mm. These results indicate that the water tive definitions ofspecies associations in the penetration film would not be as useful as field. Table 3 lists the mean depth and sub­ the Kodak 2445 natural color mm for defin­ strate of each association calculated by av­ ing red seaweed populations occupying the eraging the depths of the quadrats which lower intertidal and upper subtidal with were grouped in each association. Eight of water conditions prevalent in Georgia Strait. the ten vegetation units in Table 3 could be V\leather, tide, and sun angle requirements differentiated on air photos at a scale of for air-photography of marine algae within 1:10,000. Two groups, Desmarestia aculeata the inteliidal and upper subtidal proved to and Alaria tenuifolia, were found in com­ severely restrict the time available for munities too small and too diffuse to be de­ photography during summer in British Col­ fined accurately on the air photos. In addi­ umbia. Wind should be less than 5-10 mph tion three other biological units were map­ with cloud cover less than 10 percent cirrus ped including a small high intertidal red and no low cloud (Lukens, 1968) as con­ algal Porphyra spp. population, a deep sub­ finned by our results. Hunter (1971) noted tidal floating Nereocystes luetkeana (kelp) that specular reflectance is minimal and bot­ population, and maritime vegetation consist­ tom illumination maximal when the sun is ing primarily of Salicornia sp. (glasswort). between 45° to 50 ° above the horizon. For None of these three were sampled in quad­ color rendition of any subtidal vegetation, rats but all were conspicuous on air photo­ tides must be less than + 1.0 m or, better, less graphs. In total, 11 seaweed vegetation units than +0.5 m. Using wind, cloud, and tide were mapped using the 1: 10,000 scale color data hom Comox, Vancouver Island, aver­ and CIR photos taken under close to ideal aged over 10 years, we have calculated the conditions (1972). probability of the occurence of daylight low The minimum size vegetation unit map­ 2 tides with suitable weather in Georgia Strait ped at the 1:10,000 scale was 500 m . In­ (Table 2). For tides less than +0.1 m and a terpretation characteristics were not well de­ sun angle of 45° to 50°, the probability was fined in units smaller than this and only 90 percent that one suitable day for ae­ moreover, in the field, small populations rial photography would occur during any were often atypical and fragmentary in one summer (May-August). V\'e have not species composition compared to larger considered here the seasonal characteristics populations. of many seaweed species, which would Plate 1 shows boundaries of algal vegeta­ flllther limit photography if it had to be car­ tion units prepared from a selected air ried out during a restricted time when plant photograph and ground data. The color, sur­ density was high. All of these parameters face texture, and boundary characteristics of must be weighed carefully prior to a photo seaweed beds, along with information con­ mission with the understanding that a flight cerning the littoral distribution and the may not be possible in anyone year. major species associations of algae in the 478 PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING, 1978

TABLE 3. VERTICAL DISTRIBUTION AND TYPICAL SUBSTRATE OF MARINE PLANT ASSOCIATIONS SORTED FROM GROUND DATA SAMPLES IN GEORGIA STRAIT, B.G.

Association Mean Depth (m) Range Substrate

Rhodomela larix Gelidium sinicola Crassostrea gigas Mytilus edulis +2.8 +2.0 to +3.9 pebbles and sedimentary rock Fusuc distichus Gigartina stellata + 1.2 -3.5 to +3.9 boulders and sedimentary rock Enteromorpha intestinalis Gracilariopsis sjoestedii +0.3 -2.0 to +3.2 pebbles in sand Sargassum muticum +0.3 -2.8 to +2.9 cobbles in sand Cr~fptosiphoniawoodii Iridaea heterocarpa 0.0 -2.5 to +2.2 cobbles Iridaea cordata Plocamium coccineum Prionitis lyallii Laminaria saccharina -0.6 -6.3 to +2.9 cobbles in sand Laminaria groenlandica -1.1 -3.2 to +0.9 cobbles Desmarestia aculeata Agardhiella tenera -1.3 -4.8 to +3.4 pebbles in sand A/aria tenuifolia Costaria costata -2.0 -4.8 to +0.3 large boulders Zostera marina -2.0 -6.3 to + 1.4 sand

study area, were used to interpret aerial­ the 1: 10,000 scale maps, suggesting that photographs and map the various vegetation photo scale did not severely influence in­ units. However, both the littoral distribution terpretation of this seaweed type. and the color rendition of some algal units were observed to have considerable varia­ INVENTORY ACCURACY tion along the coast. Thus, generalized vege­ The ratios ofthe field-identified I. cordata tation characteristics on aerial-photographs vegetation unit (LV.U.) widths to the map­ were used cautiously and ground-truth ped LV.U. widths were calculated for each checks were sometimes necessary to confirm coastal stratum to obtain area adjustment correct interpretation. ratios (Benson et al., 1971). The adjustment The maps prepared in this study were use­ ratio (11) was calculated by the formula: ful for the classification of shoreline on the basis of mari ne plant vegetation. For inven­ tory purposes, large sections of coast, which L Wfi R =,_.=_1__ did not contain sufficient densities of the resource-valuable plants, could be elimi­ " L. W m nated hom more detailed surveys in which i = 1 map accuracy and such biological parame­ ters as veItical and horizontal biomass vaIia­ where Wfi are the individual fleld identified tion were estimated hom ground-truth sur­ transect widths and W", are the correspond­ veys. ing mapped bed widths. Considerable emphasis was given the ac­ The correction factor R varied between curate mapping of the vegetation unit con­ 0.63 ± 0.15 and 1.32 ± 0.17 in different stra­ taining the principle resource-valuable alga ta. A correction factor larger than 1.00 indi­ Iridaea cordata along this coast in northern cated that the mapped LV.U. area under­ Georgia Strait. The total area covered by the estimated the field-observed LV.U. area in I. cordata vegetation unit as measured on that stratum and a small correction factor the 1: 10,000 scale vegetation maps was 179 (e.g., 0.63) indicated that the area of the hectares. The area of the I. cordata unit on a mapped LV.U. overestimated the field­ test map of a portion of the coast, drawn observed LV.U. in that stratum. using air photos at a scale of 1:2,500, varied We must point out that the similarity in only 5 percent from that area estimated hom size of the mapped vegetation unit and the SURVEY OF MARINE PLANT RESOURCES 479 field-identified vegetation unit was not as vantage for the identification of major vege­ important as the precision of the ratio be­ tation units but requiring more than one tween the two. In fact, the field boundaries flight line, resulting in disproportionately may be as subjectively located as are the high costs. Poor definition of the se,lweed boundaries on the photographs. Biomass or vegetation units would be provided by other inventory data must, however, be col­ scales smaller than I: 10,000. lected in the field within a defined area Photography of most seaweed was re­ which bears some calculable relation (map stricted to low tides; furthermore, suitable adjustment ratio) to the mapped area. In the weather conditions during photography Georgia Strait survey the mapped LV.U. did were critical. These requirements may hold correspond closely (3 percent smaller) to the photographic crews on standby for many field-identified LV.U. when averaged for the days, which some agencies are unable to do. entire 80 kilometers of coast mapped. The Only three to seven days with moderate to adjustment ratio averaged for the coast was good photographic conditions are likely to 1.03 ± 0.06. vVhen we applied the adjust­ occur during anyone summer, with the pos­ ment ratio of 1.03 to the mapped area, the sibility that no optimal days will occur in a sample error (coefficient of variance) was es­ pcuticular year. Thus, survey programs must timated to be 6 percent. This error in area be designed to accomodate possible delays. measurement, obtained with the combined The use of vegetation maps prepared from use of aerial photography and ground-truth aerial photographs for inventory of data, represented a considerable improve­ resource-valuable seaweeds was investi­ ment over many previous shipboard surveys. gated using the red alga Iddaea cordata as a test species. The area of the vegetation unit SUMMARY AND CONCLUSIONS containing this spe<.:ies varied little when Procedures for mapping and inventory of mapped at I: 10,000 scale or 1:2,500 scale. seaweed vegetation using aerial photog­ vVhen cost was balanced against accuracy, raphy and ground surveys were examined the I: 10,000 scale was considered to be the for coastline in Georgia Strait, British Col­ most suitable for mapping extensive umbia. stretches of coast. Ground data for classification of marine Ground truth survevs were conducted to vegetation included collection of quadrat sample the width ofth~ I. cordata vegetation samples at 50 m intervals along transects unit and I. cordata biomass. The mapped positioned perpendicular to shore and these unit corresponded closely with the field­ data (a) eliminated long stretches of observed unit. An adjustment ratio of 1.03 coastline from the photographic surveys (calculated from the ratio of mapped to where potential resource-valuable plants field-observed I. cordata vegetation unit were at low density, and (b) were used to widths) was applied to the mapped area and interpret aerial photographs obtained along the sample error of the adjusted area was es­ selected coastline. timated to be ±6 percent. This error was Altogether, 11 vegetation un its were de­ low compared to those presented by many fined in the inteltidal and upper subtidal to 7 shipboard surveys. m below the water surface using large f(mllat The seaweed maps ofthe I. cordata vege­ natural color (Kodak 2445) and CIR (2443) tation unit prepared for 80 kilometers of photographs together. These vegetation coast will serve as an inventory base for po­ units were interpreted on the aerial photo­ tential commercial harvesting and regenera­ graphs according to their color, surhlCe tex­ tion ofthe resource but, more palticularly, as ture, boundary characteristics, substrate re­ a baseline condition of an extensive marine lations, and littoral distribution. littoral biota hom which environmental al­ During trial surveys of inteJtidal and sub­ teration can be detected. tidal communities, CIR and color com­ plimented each other, providing more inter­ ACKNOWLEDGMENTS pretative information than use of one alone. vVe thank research assistants Adrian Jones, For subtidal imagery, CIR was not useful Kent Anders, Bruce McInnis, Garry Harsh­ and Kodak water penetration film was found man, and Heather Ashton for their help as to be less satisfilctOlY than natural color. Bet­ SCUBA divers and in the collection and ter contrast was obtained on natural color analysis of ground data. Thanks are due to prints by yellow-red enhancement. Op­ Adolf Ceska for his assistance in computer timum scale was determined to be I:7,500 to SOlting procedures and to Drs. T. vViddow­ 1: 10,000. Scales larger than this (to I:2,500) son and D. V. Ellis of the University of Vic­ were not recommended, providing little ad- toria for their help and suggestions regarding 480 PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING, 1978 marine ecological surveys. Credit for an im­ (Application au Bassin de Thau et au littoral aginative fusion of Government needs and de la Region de Sete). In conf. Principes University capability belongs to the Marine Methodes Integrat. Etudes Explor., Aerienne Resources Branch, Ministry of Recreation Resources Nat. en Vue Possibilites Mise en and Conservation, Provincial Government Valeur. Toulouse, S.l. 3 p. of British Columbia. Support from the latter, Hunter, G. T. 1971. Contour mapping ofLambert the Canadian Federal Fisheries Branch, and Channel, British Columbia from inshore and also from the National Research Council of foreshore ecology. Preliminary draft repolt prepared for Canadian Hydrographic Service. Canada and University of Victoria Faculty 50 pp. Research Committee, hlllded the work, much Jamison, D. W. 1972. Aerial remote sensing as a of which will be reported in biological jour­ tool in seaweed surveys. In Proc. Int. Sea­ nals. \,Ve thank these agencies for their coop­ weed Symp. VII Univ. of Tokyo Press, pp eration and help. 351-357. REFERENCES Kelly, M. G., and A. Conrod, 1969. Aerial photo­ Anderson, RR 1971. Multispectral analysis of graphic studies of shallow water benthic aquatic ecosystems in the Chesapeake Bay. In ecology. In: Johnson, P. L. (ed.) Remote Sens­ Proc. Int. Symp., Remote Sensing of Envi­ ing in Ecology. University of Georgia Press. ronment, VII Univ. of Michigan, Ann Arbor, Athens. pp 173-184. p. 2217 to 2227. Klemas, V., et al. 1974. Inventory of Delaware's Anderson, R R, and F. J. Wobber, 1973. Wetlands Wetlands. Photog-ram. Eng. 40(4): 433-439. mapping in New Jersey. Photogram. Eng. Kuchler, A. W. 1967. Vegetation mapping. Ronald 39(4): 353-358. Press Co. N.Y. 471 pp. Austin, A. P. and R W. Adams, 1973. Develop­ Lukens, .T. E. 1968. Color aerial photography for ment ofa method for surveying red algal re­ aquatic vegetation surveys. In Proc. Symp., sources in Canadian Pacific waters. A repOlt Remote Sensing of Environment, V Univer­ submitted to the Federal Minister of sity of Michigan, Ann Arbor, pp 441-446. Fisheries and the Provindal 1inister of Rec­ reation and Conservation. May 1973. 173 pp. McEwen, R B., W. J. Kosco, and V. Calter, 1976. Coastal wetlands mapping. Photogram. Eng. 1974. Red algal resource studies in 42(2): 221-232. Ca11adian Pacific waters. A report submitted to the Federal Minister of Fisheries and the Seher, J. S., and P. T. Tueller, 1973. Color aerial Provincial Minister of Recreation and Con­ photos for marshland. Photogram. Eng. 39(5): servation. June 1974.256 pp. 489-499. _____ 1975. Red algal resource studies in Specht, M. R, D. Needles, and N. L. Fritz, 1973. Canadian Pacific Waters; Carrageenophyte New color film for water-photography pene­ inventory and experimental cultivation tration. Photogram. Eng. 39(4): 359-369. phase. A report submitted to the Federal Vadas, R L., and F. E. Manzer, 1971. The use of Minister of Fisheries and the Provincial aerial photography for studies on rocky inter­ Min ister ofRecreation and Conservation. Nov. tidal benthic marine algae. In Proc. Third 1975. 216 pp. Biennial Workshop on Aerial Color Photog­ Benson, A. S., W. C. Draeger, and L. R Pettinger, raphy in the Plant Sciences and Related 1971. Ground data collection and use. Photo­ Fields, American Society of Photogrammetry, gram. Eng. 37(11): 1159-1166. pp. 255-266. Cameron, H. L. 1950. The use of aerial photog­ Vary, W. E. 1969. Remote sensing by aerial photog­ raphy in seaweed surveys. Photog-ram. Eng. raphy for water depth penetration and ocean 16: 493-498. bottom detail. In Proc. Int. Symp., Remote Ceska, A., and H. Romer, 1971. A computer pro­ Sensing of Environment, VI University of gram for identifying species-releve groups in Michigan, Ann Arbor, pp 1045-1059. vegetation studies. Vegetatio 23(3-4): 255­ Walker, F. T. 1950. Sublittoral seaweed survey of 277. the Orkney Islands.]. Ecol. 38: 139-165. Dubois, A. 1964. Les possiblies de la photog­ -- 1954. Distribution of Laminaria raphie et de I'observation aeriennes pour around Scotland.]' Cons. Exp. Mar. 20(2): I'etude des peuplements vegetaux marins. 1960-1966.