Estuarine, Coastal and Shelf Science (1983) 16, 37-50

Remote Sensing Observations of Biological Material by LANDSAT along a Tidal Thermal Front and their Relevancy to the Available Field Data

Jacques Le F&vreO, Michel Viollier”, Pierre Le CorreC, CCcile Dupouy” and Jean-Rem? Grallt ” “Laboratcire d’Oce’anographie Biolcgique, Universitk de Bretagne Occidentale,

29283 Brest Cedex, France, bLaboratoire d’optique Atmospherique, Universitd

des Sciences et Techniques de Lille, B.P. 36, 59650 Villeneuve d’dscq, France, ‘Laboratoire d’oceanographie Chimique, Universitk de Bretagne Occidentale, 29283 Brest Ce’dex, France, and dStation Biologique, 29211 Roscoff, France

Received 13 April 1982 and in revised form 27July 1982

Keywords: hydrologic fronts ; remote sensing; plankton ; optics ; chloro-

phyll distribution; tidal variations; Celtic Sea; English Channel

Two images recorded on two successive summer days by LANDSA’I

satellite over the western approaches to the English Channel show bright

pattern of complex shape the origin of which is puzzling. Among the wavelength bands available on LANDSAT’s multispectral scanner, these

patterns are apparent only in the green region of the spectrum, and they are located towards the stratified side of a well marked tidal thermal front. Spectral signature analysis and available knowledge on hydrography and

plankton in the area are used to derive a proposed interpretation. Phyto- plankton would accordingly be the best candidate for being responsible for

the observed patterns.

Introduction

Among the coastal waters in the world’s ocean, north-west European seas are an area characterized by a combination of the considerable extension of the continental shelf and the outstanding importance of tidal processes which is rarely matched elsewhere. Perhaps as prominent a feature as the extent of such processes is that of their geographical variation over the area. Accordingly, the latter exhibits a wide range of conditions with respect to hydrographic structures, especially regarding whether or not a thermocline develops as a result of summer heating. Thermal fronts, often very sharp, are found at the boundary between stratified and homogeneous areas.

These contrasting hydrological structures have an important influence on biological processes, and especially on planktonic production. The classical pattern for the annual cycle in northern temperate waters, with two diatom outbursts and a less productive summer dinoflagellate maximum in between, is actually found where summer stratification develops, while in well mixed waters the tendency is to one season, namely summer, for maximal

tThe late Jean-Rent Grail.

37 o~7~~77~4/83/010037+17 $03.00/0 @ 1983 Academic Press Inc. (London) Limited

38 J. Le FPvre et al.

production, for which diatoms are responsible (Grall, 1972; Boalch et al., 1978; \Vafar, 1981). In addition to being a simple boundary between two contrasting areas, the fronts themselves appear as a type of environment of their own, which deserves special attention. The Ushant front, which runs across the whole entrance to the English Channel and off the west coast of Brittany, was among the first to be described in European waters (Dietrich, 1950, 1951, 1963), and to be considered a special ecosystem (Grall & Le Fevre, 1967; Le Fevre & Grail, 1970; Grall et al., 1971). More recent work on the Ushant front (and some similar environments elsewhere in European shelf seas) has been reviewed by Holligan

(1981). From a number of papers (e.g. Pingree et al., 1975, 1976, 1978; Holkan, 1979)) one of the main biological features of this front appears to be the existence of very high surface chlorophyll a values along the front outline. One species is usually responsible for these high phytoplankton biomasses, namely the dinoflagellate Gyrodinium aureolumHulburt, which, although apparently a newcomer to the area, has been reported several times under

red tide conditions (Pingree et al., 1975; Grall, 1976; Boalch, 1979; Le Fevre, 1979). In the last few years, our capabilities of studying the physical characteristics of such

structures as the fronts have been greatly enhanced by the advent of space-borne infrared radiometry, through which sea-surface temperature mapping was freed from the limitations in resolution and space-time distortions that were inherent to ship-borne measurements. Progress in collecting biological data lags somewhat behind. Zooplankton studies, as well as identifying and counting phytoplankton species, still rely largely on traditional discrete sampling. The situation is better, however, with respect to phytoplanktonbiomass. Field techniques are available (Lorenzen, 1966), which, together with elaborate statistical

methods (Denman & Platt, 1975; Denman, 1976; Fasham & Pugh, 1976; Horwood, 1978), can be used to resolve fine-scale structures. Remote sensing itself can be considered

to be on the point of providing as powerful a tool for phytoplankton studies as it does for mapping sea-surface ‘temperature, through the analysis of ocean colour from data provided by LANDSAT and NIMBUS 7 satellites. However, there are so far only a few published examples of such achievements.

There happens to be one instance of such satellite data available for two successive days in the Ushant region, which show patterns strikingly similar to recorded phytoplankton surface distribution. There is obviously a direct relationship of these patterns to the out- line of the Ushant front, and we think of these two images as good candidates for being an example of phytoplankton (or at least biological material) distribution observed from space. We present these data and discuss them both from the point of view of remote sens- ing methods, with respect to the likely nature of the phenomenon actually observed, and

from the point of view of their relevancy to available field data.

Remote sensing observations

The one remote sensing device intended, among other purposes, for surface chlorophyll assessment is the CZCS (Coastal Zone Color Scanner) aboard the experimental satellite NIMBUS 7 (Hovis et al., 1980). Data from this have been published in a number of methodological papers, especially on which kind of algorithms can or cannot be used to extract significant information on phytoplankton pigments from the radiometric data (for reviews, see Viollier, 1980; Gordon & Morel, 1981). Reports on results obtained in this way within the framework of environmental studies, are still, however, relatively few (Gordon et al., 1980; Smith & W’ilson, 1981; Mueller & Laviolette, 1981; Viollier et al., 1981). None of them is as yet relevant to the Ushant frontal ecosystem or the western approaches

Remote sensing observations of biological material

Figure I. Location of LANDSAT patterns with respect to the thermal structure. Infrared patterns are derived from a NOAA 5 very high resolution radiometer (VHRR) scene recorded on 3 July 1977, about I h earlier than the LANDSAT picture [Plate I(b)] from which the location of the bright patches (stripes) is taken. The scale of infrared (IR) numerical counts is about ro units to I “C. No reliable calibration with respect to actual temperatures is available; a reasonable guess would be about 15 “C for a numerical count of 70. The inshore leg of the front is well marked, but no bright pattern is found there [cf. Plate x(a)]. Infrared data by courtesy of the Centre de T&detection et d’Analyse des Milieux Naturels de dIEcole des Mines de Paris.

40 J. Le F&e et al.

to the English Channel. Two of us (M.1:. and C.D.), however, have work in preparation together with P. M. Holligan (MBA, Plymouth) on English Channel data from summer 1981.

Another sensor can also, under special conditions, be used for tentative estimates of surface chlorophyll content namely the Multi-Spectral Scanner (MSS) of the routine operated satellites LANDSAT, as exemplified by the work of Maul & Gordon (1975) and Gower et al. (1980). There are very few available cloud-free LANDSAT images over west Brittany. Two of them, however, taken on successive days (z and 3 July 1977) exhibit

striking features close to the Ushant front location. Plate x(a) and (b) gives the photographic reproduction of the result of data processing from MSS 4 band, corresponding to the LOO- 600 nm spectral range. Two bright patterns of complex shape, about 30 km wide, show up in these pictures. Although the satellite orbit is shifted somewhat to the west on the second day, a common area (50% of the coverage) can be observed on both images. This shows that, while the general shape of the bright patterns remains broadly unchanged, modifications

occur with respect to details, which provide a first clue to the nature of the phenomenon observed. An eddy can for instance be seen on Plate I(a) at about 48”ro’N, 5”3o’W, which exhibits on Plate x(b) unmistakable signs of evolving into a double vortex system. This, at the scale of the pattern, is a very typical characteristic of frontal dynamics (James, 1981). Further evidence of the close link between these patterns and the Ushant front is provided by NOAA 5 infrared data of 3 July 1977, the same day as the image in Plate I(b). Figure I

is an attempt to match the main features of the visible light image from LANDSAT and the infrared image from NOAA on the same day. Although the comparison is somewhat hindered by the limited extent of common coverage of the pictures and by the much coarser

resolution of VHRR data, Figure I shows a good agreement between the LANDSAT patterns and the outline of the Ushant front. It also shows that the LAKDSAT features are found on the warmer side of this front i.e. on the edge of the stratified area. A third characteristic to be observed is that the bright patches are confined along the offshore (west) leg of the front, no trace of a similar pattern being found along its southern end coming ashore near Pointe du Raz.

Interpretation

Further interpretation can be only tentative, due to the lack of direct confirmative evidence from simultaneous field data. A number of clues to the nature and significance of the pheno- menon observed are nevertheless available and can be synthesized in an attempt to build up a coherent interpretation. The necessary information can be obtained in various ways, namely spectral signature analysis, study of the hydrographic background of the area and search for consistency with appropriate available field data.

Spectral signature analysis LANDSAT provides not only spatial and temporal information on the observed target, but also spectral information, since the scene is registered through four spectral bands. The definitions of these bands and the corresponding calculated reflectances for the bright- est points of Plate I patterns are given in Table I. Calculations were made on the digital data provided by the reception station of Telespazio (Italy) according to a procedure previ- ously designed (Viollier & Baussait, 1979) to overcome some limitations of LANDSAT data in coastal waters studies. The main characteristics of this procedure are: averaging the signal over 6 line x 6 point areas to improve the signal to noise ratio, retrieving the original radiometric calibration somewhat obscured by routine processing, conversion of radiance

Plate I. (a) [Facingg. 40

Plate I. LANDSAT images of (a) 2 Jull- and (b) 3 July 1977. Positive photographic documents for the joo~-6oo nm channel, communicated by GDTA--Telespazio. Clear areas offshore correspond to high reflectance values interpreted as concen- trations of biological material, probably phytoplankton. The white spots in (b), both in the bottom right corner and at the top edge west of 6’W, are clouds. (a) is actually a mosaic of two adjacent standard photographic documents from Telespazio.

Remote sensing observations of biological material 4’

TABLE I. Reflectance6 of the bright patterns observed on the LANDSAT image

Band : MSS 4 S 6 7 Spectral range (nm) 500-600 600-700 700-800 800-1100 Reflectance 0’040 0.006 0’000 0’000

The values are given for the four spectral bands of the LANDSAT sensor (MSS). Signal to noise ratio has been improved by averaging the numerical counts over 6 line x 6 point areas, molecular atmospheric effects have been removed and the zero level has been adjusted to background values in marine areas outside the bright patches.

into reflectance, and correction of molecular atmospheric effects. Furthermore, since the uncertainty in the value of the offset of these data are relatively important, the figures given in Table I correspond in fact to the difference between the reflectances of the pixels (or groups of pixels) taken into consideration, and the reflectances of the background in the same spectral bands. These values demonstrate that the frontal features are absolutely invis- ible in MSS 6 and MSS 7 bands, slightly detectable in MSS 5 band, and highly significant in MSS 4 band, as already shown by the photographic documents.

These spectral characteristics can yield further valuable information by considering them with respect to models of marine and atmospheric optics. The radiation reaching the satellite sensor can, accordingly, be split up into three components:

(i) scattering by the atmosphere; (ii) reflection by the sea surface (sun glitter and white cap effects);

(iii) diffuse reflection within seawater.

The last component is known (see Figure 2) to exhibit a typical response, which is maximal in the blue-green region of the spectrum (400-600 nm) and negligible beyond 700 nm, due to the strong absorption of water in the red and infrared. Atmospheric and air-water interface effects, on the other hand, give way beyond 700 nm to rather high reflectance values, quite comparable to those they produce at shorter wavelength. In the set of data considered here, reflectance values in MSS 6 and MSS 7 bands are nil. Any atmospheric or interface effect, including that of floating material, can therefore be ruled out. Accordingly, the spectral signature observed for the bright patches on the LANDSAT images can only be explained by light scattering within the bulk of seawater, at least in the upper layers.

0 02

i? /-Lo\

e :

/iY. ‘\

g 0.01

i .A-

. . . . . . -- . . . . . . . . . . . . . . . s;t----w

&- ;?.,y-.\ . . . .

. . . . . ‘\

\ I,‘--

,,

-.-

0.OOV I 600 700

Wavebngth (nm)

Figure z. Typical spectral reflectance of ocean water as a function of chlorophyll concentration C (mg m+). For comments, see text.

42 J. Le F&we et al.

Seawater diffuse reflectances presented in Figure 2 have been calculated according to the following model : wavelength-dependent diffuse reflectance ~~(2) of water in a homogen- eous medium can be considered proportional to b’(n)/u(n), where b’(l) and a(;() are respec- tively the wavelength-dependent coefficients of backscattering and absorption. The formula used is

b’(i) p&)=0.17 -9 4)

(1)

which approximately corresponds to the formula ~~‘0.33 b’/a, established by Morel & Prieur (1977) for underwater measurements. For the spectral variations of b’ and a, we used a model close to the one suggested by these authors and by Morel (19&x), which gives the following equations :

(a) for the scattering properties :

b’(i)=o.oIz b,(A)+o*5 6, (A), (2)

where b,, proportional to L-l, is the scattering coefficient for suspended particles; 0.012

corresponds to a mean backscattering ratio for particles; and b, is the molecular scattering coefficient.

(b) for the absorption properties:

44=~,(4+C~,(4+b, (PO nm>44, (3)

where a, is the absorption coefficient for pure water; C is the content in chlorophyll-like pigments (mg mm3); a, is the specific absorption coefJicient for chlorophyll-like pigments; a, represents the absorption due to all particulate and dissolved matter other than chlorophyll-like pigments.

The values a, are taken from Bricaut (1979), and correspond to an average for five phyto- plankton species; those of a, are taken from Morel & Prieur (1977). In addition, we intro- duce a linear relationship between C and b,:

b, (500 nm)=o.o2+o.2 C. (4)

All these equations or coefficient values are representative of in situ measurements carried out by Morel and his team in the Sargasso Sea and in north-west African upwelling areas. They allow the calculation of the typical spectral signature of ocean water, as a function of chlorophyll concentration, which is presented in Figure 2. A large amount of variation in spectral signature is, however, attributable to the nature of coastal waters. Interpretation generally requires an accurate knowledge of the optical class of water involved, and of the hydrological background of the area studied. Indeed, dissolved matter brought in by land runoff can yield additional absorption terms in equation (3), while various suspended materials, such as sediment, can increase the value of the backscattering coefficient. The validity of equations (2)-(4) can often become doubtful, and independent determinations of b, and C be required.

Two different spectral regions show up in Figure 2:

(i) At the shorter wavelengths, the upper limit of which usually lies between 480 and 560 nm, reflectance decreases as C increases. This is due to high values of u,(i), according to the spectral characteristics of photosynthetic activity.

(ii) Beyond about 500 nm, reflectance increases together with b,, and therefore indirectly with C, provided equation (4) is valid.

Remote sensing observations of biological material 43

In some cases, for which the relationship between 6, and C does exist and has been worked out, the detection of chlorophyll content of seawater by LANDSAT is feasible. It should, however, be borne in mind that LANDSAT, which has no channel in the first spectral

region (< 500 nm), cannot detect properties of chlorophyllian pigments associated with photosynthetic activity.

According to the theoretical calculations presented above, the values measured here in RISS 4 band, i.e. a reflectance of 0.04, would correspond to a very high chlorophyll content in the bright patches: IOO mg m-3 or more. Values in this range have indeed been obtained from field measurements in the same area, at the appropriate time of year (Pingree et al.,

1977, 1978). Without the direct confirmation of field measurements at the very same time as the remote sensing data, we must, however, consider such high values unrealistic over such extended areas. In order for the model to produce more conservative chlorophyll estimates consistent with the radiometric measurements, higher values of the backscattering coefficient than those derived from equations (2) and (4) are required. Such higher values would not be unrealistic according to Wilson & Kiefer (r97g), given the proper phyto-

plankton species and the proper stage of development. Typically, the latter would correspond to a bloom going on for about IO days. These results, however, were obtained from culture batches and still need to be confirmed by in situ measurements. Resuspended sediment could also, given the proper hydrographic conditions, account for the observed high values of reflectance.

The results from spectral signature analysis can therefore be summed up as follows:

(a) The phenomenon observed is due to light scattering within the upper water layer, and not to atmosphere or surface (i.e. interface) effects.

(b) The observed reflectances appear to be higher than those observed over chlorophyll- rich phytoplankton blooms in upwelling areas studied to date. However, chlorophyll values actually recorded from field work in the area under study appear to be higher, by an order of magnitude or more, than those from any oceanic water, major upwelling regions included, which casts some doubt upon the extrapolation of models previously

adjusted for calculating chlorophyll contents hardly exceeding IO mg rnp3. (c) The hypothesis of resuspended sediments accounting for the features observed cannot,

however, be ruled out from the spectral signature analysis alone,

Hydrographic background Available information, both on the usual hydrographic behaviour of the area and on the hydrological situation prevailing at the time of the remote sensing observations, can be put at work to provide a further clue to the nature of the phenomenon observed. Various models have been put forward by various authors to account for the contrast between the stratified and the well mixed areas. All of these models, and especially the most up to date of them (Pingree & Griffiths, rg78), point out the interaction between water depth and strength of tidal currents as the main governing factor of the hydrographic regime. Vertical homogeneity is found wherever the water column is shallow enough and tidal currents strong enough to overcome the density gradients which tend to develop as a result of the upper layer’s heating. It should also be noticed that these two conditions namely shallowness and strong currents, are precisely the ones required for the presence of resuspended sediment close to the surface to be likely to any extent. Vertical stratification, on the other hand is found where the effect of warm water buoyancy prevails over stirring effects of bottom friction, i.e. potential energy over kinetic energv effects. The outline of the Ushant front is usually quite consistent

44 3. Le F&e et al

with the predictions of the model of Pingree & Griffiths (1978) for a value S- 1.5 of theFc authors’ variant of the stratification parameter, defined according to the formula

S=h&ll~lcD( I 24 I “>I, where S is the stratification parameter, Cn is the bottom drag coefficient (N o.ooa5), u is the vertically averaged current velocity, and h is the height of the water column.

Given a typical water depth of about IOO m in the vicinity of the front, s~1.5 would correspond to a critical value of about I knot for the amplitude of the tidal current (con- verted from cgs units used in the calculation). In an area where stratification is found to any extent, the figure should, therefore, be somewhat lower, a condition under which a noticeable amount of sediment could hardly be held in suspension IOO m or so above the bottom. Since the particulate matter responsible for the bright patches in Plate I is observed (see also Figure I) towards the frontal edge of the stratified area, this rules out the hypothesis of material on the process of being resuspended from the underlying bottom, and raises the question of where this material originates from.

The hypothesis of material being brought in from a more coastal area has to be discarded as well, since no such material is found closer to the shore, where depth and current condi- tions are the most favourable for its being observed. Furthermore, such structure as the front are well-known (e.g. Pingree et al., 1974, figure 6) for being affected by convergent motion. The particulate matter observed beneath the surface in the stratified area is therefore moving from offshore towards the front, where it is in the process of being concentrated near the surface or sinking with the downwelling water, and not westwards from the coastal area. In Plate I, the southernmost patch indeed appears to have narrowed somewhat between 48”oo’ and 48”30’N on 3 July with respect to the previous day.

Yet another possible origin should, however, be taken into consideration, One could imagine a situation where the front outline has recently moved eastwards, corresponding to the appearance of some stratification over a fringe of previously well mixed water. Some sediment could be still present in this recently stabilized water, while being in the process of settling, or otherwise being recycled through the frontal circulation system. However, there are difficulties inherent in such a hypothesis. This, for instance, would still not explain why no such material, or at least a much smaller amount of it (taking into account the fact that the zero reflectance level has been adjusted to background turbidity) is found in the area where vertical mixing still prevails. These difficulties, however, are unimportant, since the hypothesis has to be rejected on other grounds. Frontal position, once the summer situation is established, has been shown (Simpson, 1981) to vary only to a negligible extent in response to wind forcing, so that tidal conditions alone shouId be taken into consideration. Advection related to the M, tidal cycle can, furthermore, be neglected as a first approxi- mation with respect to the position of the Ushant front, at least between 48”oo’ and 48”3o’N, since this has been shown (Mariette et al., rg8o), to take place there primarily along a north-south direction, i.e. roughly parallel to the front outline. We are, therefore, left with one main cause of change in frontal position, namely neaps-springs adjustment. This occurs only to a limited extent, due to a feedback mechanism (Simpson & Bowers, rg&r), and exhibits a phase lag of about 2 days with respect to the tide amplitude cycle. The trend is for a retreat of the front towards the stratified area when the tide increases from neaps to springs, and a spread of stratification towards the well mixed area for a tide amplitude variation in the opposite sense. July 2 and 3, 1977, were days of maximal spring tides, and, accordingly, the front couId only have been retreating westwards. Indication of such a shift can indeed be obtained through comparison of Plate r(a) and (b) for almost the whole of

Remote sensing observations of biological material 45

both bright patches. This shift is very likely to be representative of neaps-springs adjust- merit:Both LANDSAT images were recorded at the end of the ebb, 44 min before low tide on 2 July and about I h 30 min before low tide on 3 July. The Lagrangian measurements carried .out on both sides of the front by Mariette et al. (1980) show that whenever surface drift exhibits an east-west component, the latter is slightly westwards at the relevant time of the tide. Tidal amplitude was larger at the time of LANDSAT images than it was at the time of Mariette’s measurements, which makes likely a larger extent of tidal advection. But would this be responsible for the east-west shift, the front ought to be found at a more easterly position on the image from 3 July, taken earlier with respect to the tide cycle, than on that from 2 July. Just the opposite is actually observed. Changes in frontal position related to tidal advection bear anyway little relevancy to the origin of the particulate matter and the likelihood of resuspended sediment being present, while those related to neap- springs adjustment do, and on this instance contribute to rule out an eastern origin for a material observed to the west of the front.

The only possibility left for the origin of the particulate matter observed in the bright patches is, therefore, the stratified area west of the front. A bottom origin is obviously ruled out on the very grounds of the existence of a stratification and the consideration this implies with respect to current speeds (see above). The deeper water layer also is a very unlikely source. This consists of a stable cold water mass termed ‘bourrelet froid’ by Vincent & Kurt (1969). This water mass is characterized by a low dissolved oxygen content, well under the saturation value, as observed for instance in June 1969 (Grall et al., 1971).

Thisresult, found again repeatedly in 1979, 1980 and 1981, at different periods of summer time, varying from mid-June to late September, and under any tide conditions (Le Corre, unpublished data), clearly shows that this bottom water is severed from major exchange with the upper layers for a matter of months.

Taking into consideration the present state of knowledge on hydrography in the vicinity of the Ushant front leads to the conclusion that the suspended material responsible for the bright patches showing up in the LANDSAT images necessarily comes from the stratified area west of the front, where the thermocline level and the upper layers are the sole possible sources. Only a biological origin can be consistent with such considerations.

Consistency with biological field data The best candidate for responsibility for the LANDSAT features is obviously biological material the abundance. of which is best correlated with what could most neutrally be called ! turbidity’. There is indeed one such parameter for which years of records in the Ushant region do show a high correlation to turbidity, expressed as Secchi disc reading, namely chlorophyll content (Holligan, personal communication). Phytoplankton should, therefore, be given prime consideration in the search for the material responsible for the features ‘observed, although other possibilities are not ruled out, and will be taken into account in the discussion. According to the optical model presented above, records of chlorophyll content in the range 10-100 mg m- 3 should be looked for if a match to the situation observed OQ the LANDSAT pictures is to be found on the basis of pigment abundance, even taking into‘account a higher than usual value for the backscattering coefficient b’. There happen to, be a few available accounts of such high values in the Ushant region. One of them is the map of surface chlorophyll distribution on 27-28 July 1975 first published by Pingree et ‘al. (1976). Isolines are drawn by these authors for values equal to powers of 2, the highest of which is 64 mg m- 3. With the limitations in spatial resolution inherent in field work, the pattern obtained is strikingly similar to what is observed here in Plate I and bears an

46 J. Le FPare

identical relationship to the surface isotherms and outline of the front. A second example is provided by a situation observed by the same team on 31 July-2 August 1976, the map of

which was first published by Pingree et al. (1977). The system of representation is the same as for the previous case, with the difference that the highest value for an isoline is rz8 mg m-3.

Spatial relationship to the thermal structure is again similar. The third example is provided by our own results, obtained on cruise SATIR aa, 3-8

September 1979 (acronym from SATellite and IRoise Sea, the name of the area studied). Surface chlorophyll distribution is shown in Figure 3 and temperature distribution in Plate 2, from a TIROS N infrared scene. While our values are lower than those from the two previous cases, chlorophyll contents higher than 25 mg rnp3 were nevertheless observed. The data responsible for the chlorophyll maximum as an offshore rather narrow band were obtained on 6-7 September, together with temperature measurements. Infrared data are also available for the same two days; they show that the front outline had changed little from one day to the other, and that isotherms derived from field measurements over these two days agree quite well with the instantaneous patterns actually observed on satellite images (for a comparison see Le Fevre et al., 1979; Grail et al., 1980). Chlorophyll patterns are therefore likely to be just as realistic, again with the limitations in resolution inherent to field data. High values are again found along the front outline, and in the same position

40' 20'

Figure 3. Surface chlorophyll a (mg mm3) in the Iroise Sea on the first neck of September 1979. Chlorophyll values were obtained on cruise SATIR aa; solid circles indicate where the samples \vere collected.

Plate 2. Surface temperature (C) in the Iroise Sea shown as an advanced very high resolution radiometer scene recorded on 7 September ry;y. The correspon- dence between the grey scale and actual temperature is derived from SATIR za field measurements. Both offshore (a) and inshore (b) legs of the front are pointed at by arrows. Infrared image by courtesy of the Centre de MSorologie Spatiale, Lannion.

Remote sensing observations of biological material 47

with respect to the overall thermal structure, i.e. towards the frontal edge of the stratified area. The highest values are found exactly where the front is the sharpest, a situation which strongly recalls that of the LANDSAT images.

All three instances presented above share with the LANDSAT case not only spatial characteristics, but temporal ones as well, namely being observations made within 3 days of springs maximal tidal amplitude. The highest chlorophyll values (July 1975, 31 July- 2 August 1976) were observed shortly after maximal tides, while SATIRza’s were collected just before. At the same place as their 31 July-z August 1976 observations, Pingree et al. (1977) indeed obtained values quite similar to the ones presented here in Figure 3, on 27-28 July 1976, i.e. in the same situation as SATIR 2a with respect to the neaps-springs cycle. These similarities, both spatial and temporal, between several actually observed patterns of surface chlorophyll distribution and the LANDSAT image argue in favour of the latter’s being relevant to the same kind of phenomenon.

Discussion and conclusion

Difficulties in accounting for the observed reflectance values on the ground of the usual relationships between chlorophyll content and optical phenomena cast some doubt upon the kind of phytoplankton involved, if indeed phytoplankton is responsible for the features observed on the LANDSAT images. Other forms than the typical diatoms or dinoflagellates would account better for unusual optical properties. The genus Halosphaera (Prasino- phyceae) could be a possible candidate, since it has been found in dense near-surface patches that can produce a green discoloration of seawater (Boalch & Mommaerts, 1969). However, these observations were made in April, and other records of large numbers of this organism in the relevant area were even earlier: March (Le Fevre, 1971) or mid-winter (Parke & den Hartog-Adams, 1965). Coccolithophorids would be another quite distinct possibility, since these cells, armoured as they are with minute calcareous plates, are likely to exhibit an optical behaviour recalling that of sediment, and since there is at least one record (Holligan, 1978) of their abundance at the proper place and time of the year.

The alternative hypothesis of material other than phytoplankton being, in fact, respon- sible for the bright patches on the LANDSAT images should finally be taken into con- sideration. The possibilities are few however. One of them, which would account for the high reflectance values, is that of a milky water, similar to the one reported in the North Sea by Volkman et al. (1980). This phenomenon was due to the presence of decaying material rich in wax esters, originating from dead zooplankton. However, dead zooplankton is uncommonly encountered in the sea, and appears to be found mainly in the case of some unusual phenomenon resulting in a mass mortality. In the instance reported by Volkman et af. (1980) the factor put forward was a sudden fall in salinity, which is very unlikely to occur in the case of our region. A kill by a dinoflagellate red tide would be more appropriate, but this would be a less economical hypothesis than the one proposed.

Even considering that phytoplankton is merely a fairly likely candidate responsible for the LANDSAT patterns, but not a certainty, the images still provide very valuable infor- mation. Distribution of some particulate material of biological origin is observed in a synoptic way, the spatial characteristics of which are seen with an accuracy and a resolution that field data are unlikely to match in a predictable future. Relationships of the distribution of this material to the turbulent water circulation regime are shown with respect to both space and time, and processes such as convergence and neaps-springs adjustment of frontal position are visualized, which necessarily influence any material suspended within the

48 J. Le FPwe et al.

water mass, plankton included. These processes, therefore, should be taken into considera- tion in any conceptual model of the relationships between the pelagic ecosystem and such structures as tidal fronts.

Nc;te

Although some other posthumous work by Jean-Rem4 Grail is still due to be published, this paper is the last one to which he has contributed before he died in September 1980. Significant changes have subsequently been made to the text, especially with respect to the discovery in the appropriate archives of the second LANDSAT image. We cannot, therefore, be sure that he would have endorsed all the details of the final version. We feel, however, that his name deserves to remain on the authors’ list, in view of the part he did play in this work.

Acknowledgements

The authors are indebted to Dr P. M. Holligan for providing valuable information on his own data, as well as to Drs Y. Camus, V. Mariette, R. Maze and R. D. Pingree for critical reading of the manuscript. The SATIR cruises were made possible through ship’s time allocation by the Centre National de la Recherche Scientifique and funding by the Centre National pour 1’Exploitation des Oceans and by the Etablissement Public Regional de Bretagne. The remote sensing work was supported by a joint contract from the Centre National d’Etudes Spatiales and the Centre National de la Recherche Scientifique.

References

Boalch, G. T. 1979 The dinoflagellate bloom on the coast of south west England, August-September 1978. journal of the Marine Biological Association of the United Kingdom 59, 515-517.

Boalch, G. T., Harbour, D. S. & Butler, E. I. 1978 Seasonal phytoplankton production in the western English Channel, 1964-1974. Journal of the Marine Biological Association of the United Kingdom

5% 943-953. Boalch, G. T. & Mommaerts, J. P. 1969 A new punctate species of HalosphaeraJournal of the Marine

Biological Association of the United Kingdom 49, I 29-139. Bricaut, A. 1979 Absorption, diffusion (et retrodiffusion) de la lumiire par les substances influant sur

la couleur des eaux de mer. These de Doctorat de Specialit& Universite Pierre et Marie Curie (Paris

6). ISI PP. Denman, K. L. 1976 Covariability of chlorophyll and temperature in the sea. Deep-Sea Research 23,

539-550. Denman, K. L. & Platt, T. 1975 Coherences in the horizontal distributions of phytoplankton and

temperature in the upper ocean. Mimoires Sociite’ Royale des Sciences de Liige, 6e se’& 7, 19-30.

Dietrich, G. 1950 Die anomale Jahresschwankung der Warmeinhalts im Englischen Kanal, ihre Ursachen und Auswirkungen. Deutsche Hydrographische Zeitschrift 3, 184-201.

Dietrich, G. 1951 Influences of tidal streams on oceanographic and climatic conditions in the sea as exemplified by the English Channel. Nature, London 168, 8.

Dietrich, G. 1963 General Oceanography, an Introduction. Wiley-Interscience, New York. 588 pp. Fasham, M. J. R. & Pugh, P. R. 1976 Observations on the horizontal coherence of chlorophyll (I and

temperature. Deep-Sea Research 23, 527-538. Gordon, H. R., Clark, D. K., Mueller, J, L. & Hovis, W. A. 1980 Phytoplankton pigments from the

Nimbus-7 Coastal Zone Color Scanner: comparisons with surface measurements. Science 210, 63-66.

Gordon, H. R. & Morel, A. 1981 Water colour measurements. An introduction. In Oceanography from Space (Cower, J. F. R., ed.). Plenum Press, New York. pp. 207-212.

Gower, J. F. R., Denman, K. L. & Holyer, R. J. 1980 Phytoplankton patchiness indicates the fluctua- tion spectrum of mesoscale oceanic structure. Nature, London 288, 157-159.

Grail, J. R. 1972 Recherches quantitatives sur la production primaire du phytoplancton dans les parages de Roscoff. These de Doctorat es-Sciences, Universite de Paris 6. 259 pp.

Remote sensing observations of biological material 49

Grail, J. R. 1976 Sur une “eau coloree” B Gyrodinium aureolum Hulburt observee en Manche. Travaux de la Station Biologique de Roscoff 23, 19-22.

Grail, J. R., Le Corre, P., Le F&e, J., Marty, Y. & Tournier, B. 1980 Caracteristiques estivales de la couche d’eau superficielle dans la zone des fronts thermiques Ouest-Bretagne. Oceanis 6, 235-249.

Grall, J. R. & Le Fhre, J. 1967 LJne “eau rouge” a Noctiluques au large des &es de Bretagne. Penn ar Bed 6,152-163.

Grall, J. R., Le Fevre-Leho&fI, G. & Le Fevre, J. 1971 Observations sur la distribution du plancton a proximite d’ouessant en juin 1969 et ses relations avec le milieu physique. Cahiers OC~?UZO- graphiques 23, 145-170.

Holligan, P. M. 1978 Patchiness in subsurface phytoplankton populations on the northwest European continental shelf. In Spatial Pattern in Plankton Communities (Steele, J. H., ed.). Plenum Press, New York. pp. 22x-238.

Holligan, P. M. 1979 Dinoflagellate blooms associated with tidal fronts around the British Isles. In Toxic Dinoftagellate Blooms (Taylor, D. L. & Seliger, H. H., eds). Elsevier, New York. pp. 249256.

Holligan, P. M. 1981 Biological implications of fronts on the northwest European continental shelf. Philosophical Transactions of the Royal Society of London Series A 302, 547-562.

Horwood, J W. 1978 Observations on spatial heterogeneity of surface chlorophyll in one and two dimensions. Journal of the Marine Biological Association of the United Kingdom 58, 487-502.

Hovis, W. A., Clark, D. K., Anderson, F., Austin, R. W., Wilson, W. H., Baker E. T., Ball, D., Gordon, H. R., Mueller, J. L., El Sayed, S. Z., Sturm, B., Wrigley, R. D. & Yentsch, C. S. 1980 Nimbus-7 Coastal Zone Color Scanner: system description and initial imagery. Scicncv 210,60-63.

James, I. D. 1981 Fronts and shelf-circulation models. Philosophical Transactions of the Royal Society of London Series A 302, 597-604.

Le Fevre, J. 1971. Evaluation des caracteristiques d’emploi d’un echantillonneur de plancton haute vitesse, suivie d’exemples d’application a l’etude du zooplancton de la pointe de Bretagne. Th&e de Doctorat de Specialite, Universite de Paris 6. 179 pp.

Le Fevre, J. 1979 On the hypothesis of a relationship between dinoflagellate blooms and the “Amoco Cadiz” oil spill.~ournal of the Marine Biological Association of the United Kingdom 59, 5z5--527.

Le Fevre, J. & Grall, J. R. 1970 On the relationships of Noctiluca swarming off the western coast of Brittany with hydrological features and plankton characteristics of the environment. Journcrl o.f Experimental Marine Biology and Ecology 4, 287-306.

Le Fkre, J., Quiniou-Le Mot, F. & Tournier, B. 1979 Structures thermiques et distribution de certains organismes planctoniques: nouvelles methodes d’approche a partir de l’exemple du site de Plogoff. DeuxiknesJourne’es de la Thermoecologie, Names. pp. 229-244.

Lorenzen, C. J. 1966 A method for the continuous measurement of in vivo chlorophyll concentration. Deep-Sea Research 13, 223-227.

Mariette, V., Le Saos, J. P., Pichon, A. & Girardot, J. P 1980 Resultat des mesures effect&es lors de la campagne. DYNATLANT 80. Rapport scientifique, Laboratoire d’oceanographie Physique, Universite de Bretagne Occidentale, Brest. 93 pp.

Maul, A. & Gordon, M. R. 1975 On the use of the earth resources technology (Landsat I) in optical oceanography. Remote Sensing of Environment 4, 95-128.

,MoreI, A. 1980 In water and remote measurements of ocean color. Boundary-(ayer Meteorology 18,

177-201.

Morel, A. & Prieur, L. 1977 Analysis of variations in ocean color. Limnology and Oceanography 22,

709-722.

Mueller, J. L. & Laviolette, P. 1981 Color and temperature signatures of ocean fronts observed with the Nimbus-7 CZCS. In Oceanography from Space (Gower, J. F. R., ed.). Plenum Press, New York. pp. 295-302.

Parke, M. & Den Hartog-Adams, I. 1965 Three species of Halosphaera. Journal of the Marine Bio- logical Association of the United Kingdom 45, 537-557.

Pingree, R. D., Forster, G. R. & Morrison, G. K. 1974 Turbulent convergent tidal fronts. Jotrr?lul of’ the Marine Biological Association of the United Kmgdom 54, 469-479.

Pingree, R. D. &. Griffiths, D. K. 1978 Tidal fronts on the shelf seas around the British Isles. Jour>r~ll of Geophysical Research 83,46x5-4622.

Pingree, R. D., Holligan, P. M. & Head, R. N. 1977 Survival of dinoflagellate blooms in the western English Channel. Nature, London 265, 2666269.

Pingree, R. D., Holligan, P. M. & Mardell, G. T. 1978 The effects of vertical stability on phyto- plankton distributions in the summer on the northwest European shelf. Deep-Sea Research 25, 1011-1028.

Pingree, R. D., Holligan, P. M., Mardell, G. T. & Head, R. N. 1976 The influence of physical stahilit! on spring, summer and autumn phytoplankton blooms in the Celtic Sea. Journal af the Marine Biological Association of the United Kingdom 56, 845-873.

50 J. Le Fivre et al.

Pingree, R. D., Pugh, P. R., Holligan P. M. & Forster, G. R. 1975 Summer phytoplankton blooms and red tides in the approaches to the English Channel. Nature, London 258,672-677.

Simpson, J. H. 1981 The shelf-sea fronts: implications of their existence and behaviour. Fhi/oro$hica/ Transactions of the Royal Society of London, Series A 302, 531-546.

Simpson, J. H. & Bowers, D. 1981 Models of stratification and frontal movements in shelf seas. DC+- Sea Research Series A 28, 727-738.

Smith, R. C. & Wilson, W. H. 1981 Ship and satellite bio-optical research in the Califcrnia Bight. In Oceanography from Space (Cower, J. F. R., ed.). Plenum Press, New York. pp. 281-294.

Vincent, A. & Kurt, G. 1969 Hydrologie. Variations saisonniires de la situztion the1mic.t e du Golfe”’ de Gascogne en 1967. Rewue des Travaux de 1’Institut des Plches Maritimes 33,79-96.

Viollier, M. 1980 T&detection des concentrations de seston et pigments chlorophylliens contenus dans l’ocean. These de Doctorat &s-Sciences, Universite des Sciences et Techniques de Lille. I 92 pp. + appendices.

Viollier, M. & Baussart, N. 1979 Enhancement of Landsat imagery for monitoring coastal waters. Application to the southern part of the North Sea. Proceedings qf the 13th Symposium on Remote _ _ _ Sensing of Environment, Ann Arbor. pp. 1~9331105.

Viollier, M., Baussart, N. & Deschamps, P. Y. 1981 Preliminary results of CZCS Nimbus-7 experi- ment for ocean colour remote sensing: observation of the Ligurian Sea. In Oceanography from Space (Gower, J. F. R., ed.). Plenum Press, New York. pp. 387-393.

Volkman, J. K., Gatten, R.. R. & Sargent, J. R. 1980 Composition and origin of milky water in the North Sea. Journal of the Marine Biological Association of the United Kingdom 60, 759-768.

Wafar, M. 1981 Nutrients, primary production, and dissolved and particulate organic matter in well- mixed temperate coastal waters (Bay of Morlaix, Western English Channel). These de Doctorat de Sptcialite, Universite Pierre et Marie Curie (Paris 6). 238 pp.

Wilson, W. H. & Kieffer, D. A. 1979 Reflectance spectroscopy of marine phytoplankton. Part 2: a

simple model of ocean color. Limnology and Oceanography 24, 673-682.