Borstad Associates Ltd. - Mapping Surface Chlorophyll and Temperature

Mapping Surface Chlorophyll and Temperature in a Coastal Sound Using Remote Sensing Imagers

Borstad Associates Ltd., 1996

SUMMARY

As part of an airborne remote sensing field experiment staged during the period April 24 - May 5, 1987, to coincide with biological studies of marine salmon survival in Barkley Sound on Vancouver Island, British Columbia, we have mapped solar stimulated in vivo fluorescence of chlorophyll a using a state of the art imaging spectrometer, the 'Fluorescence Line Imager' (FLI) belonging to the Canadian Department of Fisheries and Oceans. Supporting in situ observations and thermal infrared image data from a Daedalus multispectral scanner were also obtained. This document presents an overview of the digital image data for the April 25 flight and a brief discussion of its relationship to the in situ data collected.

INTRODUCTION

Mapping phytoplankton distribution and growth is important in fisheries and physical oceanographic studies. The light absorbing pigments collectively known as chlorophyll a are commonly used by oceanographers and limnologists as an index of phytoplankton concentration increases. This phenomenon has enabled remote sensing of the pigment from aircraft and satellites (Clarke et al., 1970; Morel and Prieur, 1977; Gordon et al., 1983).

A second independent method for remote detection of chlorophyll has been developed at the Institute of Ocean Sciences (Neville and Gower, 1977; Gower, 1980; Borstad et al., 1980; Gower and Borstad, 1981; Borstad and Gower, 1984; Gower et al., 1984). This work has shown that in vivo fluorescence by the plant pigment chlorophyll a stimulated by direct or scattered sunlight, provides a distinctive signature for the presence of this pigment in near surface water.

For the 1987 experiment, remote sensing data acquisition was carried out under two separate contracts. Water colour image data was collected by Moniteq Ltd. of Toronto, using the DFO Fluorescence Line Imager, a programmable high spectral resolution imaging sensor designed to image chlorophyll fluorescence (Borstad et al., 1985). More conventional colour and thermal image data were obtained by Pacific International Mapping Ltd. of Victoria, who flew a Daedalus multispectral scanner in a separate aircraft. Flights were timed to coincide with maximum tidal flow, so that small scale tidal circulation could be imaged.

Figure1. Approximate location of the transects made by the aircraft carrying the Fluorescence Line Imager, April 25, 1987. Locations of in situ samples are indicated by crosses.

The following discussion of the relationships between the remote sensing data and in situ observations illustrates the potential of high spectral resolution image data. Some problems in interpretation of the data (possibly arising from processing) are pointed out. A more complete discussion of this work is to be found in Borstad et al., (1988). The Fluorescence Line Imager and its use are described in Borstad et al. (1985).

METHODS

1. FLUORESCENCE LINE IMAGER DATA

Digital data was acquired along 6 transects over Barkley Sound (Figure 1), from a nominal altitude of 5500m (18,000') on several days during the period April 23 to May 5, 1987. Only data for flight lines 2, 3, 5 and 6 of April 25 have been processed.

The FLI operates in one of two modes: 'spatial mode' in which it is a high spatial resolution pushbroom scanner with 8 spectral bands, the position and width of which are under keyboard control; and 'spectral mode' in which it can be thought of as a low spatial resolution pushbroom scanner, or as 40 spectrometers with 2.5 nm spectral resolution. Images can be formed in either mode. The flexibility of band selection has been used extensively during the first years of operation of the instrument to select optimum spectral bands for various uses.

Table 1 lists the spectral bands defined for the 1987 chlorophyll fluorescence mission.

Table 1. Spectral band definitions for mapping of chlorophyll via the green /blue ratio
and in vivo fluorescence (F)

Band number	Wavelength range (nm)
0	467.7 - 476.9
1	540.9 - 550.1
2	586.6 - 595.7
3	633.5 - 643.9
4 F	659.5 - 672.5
5 F	673.8 - 686.9
6 F	708.8 - 713.9
7	746.2 - 757.8

Processing began with reading FLI bands 3 and 7 for all of the data from CCT onto disk with reduced spatial resolution as a 'quick look'. Two scenes of 512 scanlines at the south end of line 6 were selected for development of processing protocol. At the altitude and aircraft velocity used for this mission, the FLI oversamples in the cross-track direction by a factor of 4 (data can not be recorded fast enough). In order to correct for this, bands 4, 5 and 6, which are required for calculation of FLH images, were then read in, subsampling in the cross-track direction. This data was then `normalized' for camera gain differences and limb brightening using procedures developed earlier (Borstad et al., 1986, 1987). After normalization parameters were established which permitted calculation of FLH images reasonably free of instrument effects, the rest of the data were processed. For flight lines 5 and 6, the cross-track subsampling scheme referred to above was used to reduce the width and adjust the aspect ratio of the images. Data for lines 2 and 3 were processed using averaging, since this provided an increase in the signal to noise ratio of the data over subsampling.

As explained in earlier reports, determination of the normalization parameters is very time consuming because: (1). it requires the analyst find an area which is uniform across the entire swath of the image; (2). the FLI produces a very large amount of data; and (3). many of the instrument effects are not constant and have not yet been adequately characterized.

After calculation of the FLH images for each 372 pixel x 512 scanline scene, average FLH values were extracted for areas corresponding to a 300 x 300 m area surrounding the location of in situ samples. Since navigation data was not recorded electronically on the aircraft the geometric resampling required to remove distortions is crude. There was a variable 10 to 20º crab angle caused by a cross track wind at the time of data acquisition, as well as small variations in aircraft speed, yaw and roll. Evidence of uncompensated aircraft motion may be seen in Plate 1. These motions were so severe on flight lines 2 and 3 that we have not made a composite of images of the data for the central part of these lines (data for the north part of line 3 is missing).

Green/blue ratio images could also be calculated from the FLI data and might be expected to show strong patterns resulting from absorption at short wavelengths by dissolved organic compounds in the fresh water runoff, as well as absorption from chlorophyll. This work was in progress at the time of this note.

2. THERMAL IMAGERY

Daedalus multispectral imagery was obtained on April 25 only. Flight tracks were similar to those of the FLI, but slightly offset because of the asymmetrical field of view of the FLI. Because of an operator error the thermal imagery recorded was from the 3 µm band instead of the 11 µm band. Noise levels are higher than would be expected with the higher wavelength, but thermal patterns are easily discernible and the data are still useful
(Plate 2).

3. IN SITU OBSERVATIONS

In situ measurements are more fully described in Borstad et al., (1988). Table 2 summarizes observations made at 51 locations within the Sound on April 23, 24, 25 and May 5, 1987.

Table 2. In situ water observations of Barkley Sound water properties.

Parameter	Depth	Method
Salinity	0m	Water sample
Temperature	0m	Thermistor
Secchi depth		Secchi disk
Water colour	½ Secchi	Munsell scale
Chlorophyll & phaeopigments	0m	Water sample
Fluorescence	0m	Fluorometer
Dissolved organic matter	0m	Optical density
Suspended solids	0m	Filtered weight
Phytoplankton	0m	Cell counts & taxonomy

RESULTS AND DISCUSSION

1. WATER COLOUR IMAGERY

Fluorescence was highest in Trevor Channel (Plate 1) and lowest in the Pipestem Inlet at the head of Loudon Channel. A rapid change in FLH is seen across a front extending west of the small peninsula which points westward into Loudon Channel. This front corresponds to the location of a thermal front seen in the thermal imagery (Plate 2). Distortion in the imagery is due to uncompensated aircraft roll or yaw. Striping is evidence of continuing problems with instrument effects.

Figure 2. A comparison of Fluorescence Line Height and in situ chlorophyll for stations within the area imaged in Plate 1. Squares represent April 24, crosses represent April 25.

The relationship between FLH radiance extracted from the large scale 372 x 512 pixel scenes and in situ chlorophyll a concentration (not including phaeopigments), for 33 stations within the imaged area is illustrated in Figure 2. The standard deviation of FLH for a 300m square area around each of the stations was in the range .0044-.0067 W/m2 sr µm. Agreement between FLH and chlorophyll was good, especially if the data are separated into two groups according to their temperature salinity characteristics (see later discussion of fluorescence yield). Lines joining two values for stations 12, 19 and 20 indicate different FLH obtained for the same locations on neighbouring transects. On average, FLH on transect 5 was .020 W/m2 sr µm higher than on line 6. This problem is not yet fully investigated, and may relate to sun elevation (although the effect is not removed when reflectance FLH is calculated), look direction, real diurnal variations of fluorescence or unresolved problems with our normalization procedures.

2. THERMAL IMAGERY

The thermal image in Plate 2 shows that the eastern side of Sound was considerably warmer than the side open to the Pacific. A diffuse front crossing the middle of the Sound separated the warm and cold water. Local warming in enclosed bays and small inlets, colder river plumes and turbulence behind islets can also be seen.

3. TEMPERATURE / SALINITY

The TS plot in Figure 3 shows that surface waters of Barkley Sound on April 24/25 were essentially a mixture of 2 water types: coastal water of a temperature near 10ºC and salinity near 30‰, and warmer, fresher water (temperature near 12ºC, salinity near 18‰) influenced by river runoff. The upper group of points are from the heads of the two channels where surface waters had warmed slightly. Because we wanted a wide synoptic coverage we did not take subsurface samples or make vertical profiles as part of the in situ sampling. However, temperature measurements were also made at 1.5m at 5 stations on April 25. The vertical lines in Figure 3 join the surface and sub-surface values. That the five subsurface values all fell on the main mixing line for surface data reinforces our supposition that the upper group represents more stratified waters. Deeper vertical profiles are required for a more complete description.

Figure 3. Temperature versus salinity for surface samples collected April 24 (squares) and 25 (crosses).

4. DISSOLVED ORGANIC MATERIAL ABSORPTION

Freshwater runoff in this area exhibits a strong absorption at short wavelengths by dissolved organic materials (tannins, lignins and other products of plant litter breakdown), and waters of local creeks and rivers are visibly brown. Figure 4, which shows the relationship between salinity and optical density at 350 nm of filtered water, demonstrates that local rivers, and possibly the pulp mill on the Alberni Canal at the head of Trevor Channel, are important sources. These geographic variations in OD₃₅₀should be visible in blue/green ratio images. Elevated OD₃₅₀at stations 6, 8, 9 and 21 have not been satisfactorily explained, but may be related to a lab measurement error.

Figure 4. Optical density at 350 nm of filtered surface water versus surface salinity, illustrating that fresh water was a source of dissolved organic material. April 24 (squares) and 25 (crosses).

Figure 5. Fluorescence radiance (FLH) versus surface temperature, line 5 data adjusted to agree with line 6. April 24 (squares) and 25 (crosses). Adjusting line 5 data increases the spread of points.

5. PHYTOPLANKTON, CHLOROPHYLL AND FLUORESCENCE

The greatest phytoplankton and chlorophyll concentrations were near the mouth of the Sound. It is now clear that both the green discolouration and the low Secchi transparencies observed during the field work were a result of strong absorption by dissolved organic materials from freshwater runoff. These can be expected to interfere with remote sensing of chlorophyll using the green-blue method. Generally, the concentration of surface phytoplankton and chlorophyll as well as fluorescence decreased with increasing temperature (Figure 5), suggesting that stratification resulting from warming of low salinity surface waters and resulting nutrient depletion was limiting surface phytoplankton growth in the eastern part of the Sound at this time. There was also some variation of phytoplankton composition across the Sound, a with higher proportion diatoms in the west and a greater proportion of chrysophytes in the east. Fluorescence yield decreased with increasing % diatoms and decreasing % chrysophytes.

The separation of stations on the basis of fluorescence yield is also shown in a temperature-salinity plot in which the available fluorescence yield values (fluorescence uncorrected for differences between flight tracks) are plotted (Figure 6). Most of the yields greater than .04 are in the waters on the upper line representing warmer waters which we interpret as more stratified.

CONCLUDING REMARKS

The combination of large area coverage by remote sensing devices and surface in situ collection from a rapidly moving small boat allowed a good description of the surface character of the Sound. Vertical profiles at a few locations would have allowed a more complete discussion, however this would have required a second boat or helicopter. The patterns visible in thermal imagery allow flow visualization, even in the poor quality 3 µm band. At the time of this operation the spatial pattern of surface temperature and salinity was much more complex than could be adequately sampled from a vessel, although FLH was more slowly varying.

This document reports work conducted during the 1987 DSS contract 06SB.FP941-7-0399/01-SB to G.A.Borstad Associates Ltd. entitled "Surface data and remote sensing in support of the MASS program".

Figure 6. Surface temperature versus salinity for the April 24 (squares) and 25 (crosses) stations, labelled with fluorescence yield (FLH/Chl) x 100.

REFERENCES

Borstad, G.A., R.C. Kerr, L.L. Armstrong and D.N. Truax. 1988. Remote sensing and surface data in Barkley Sound in support of the 1987 Marine Salmon Survival Program. Unpubl. report by G.A. Borstad Associates Ltd., Sidney, B.C. for Dept. Fisheries and Oceans, Sidney, B.C. DSS 06SB.FP941-7-0399/01-SB.

Borstad, G.A., R.C. Kerr, D.N. Truax and D. Pan. 1987. Using an imaging spectrometer to map phytoplankton chlorophyll. Unpubl. report by G.A. Borstad Associates Ltd., Sidney, B.C. for Dept. Fisheries and Oceans, Sidney, B.C. DSS 06SB.FP941-6-0259.

Borstad, G.A. 1986. Analysis of test and flight data from the Fluorescence Line Imager, 1985. Unpubl. report by G.A. Borstad Associates Ltd., Sidney, B.C. for Dept. Fisheries and Oceans, Sidney, B.C. DSS 06SB.FP941-4-3696.

Borstad, G.A., H.R. Edel, J.F.R Gower, and A.B. Hollinger. 1985. Analysis of test and flight data from the Fluorescence Imager. Can. Spec. Publ. Fish. Aquat. Sci. 83: 38 p.

Borstad, G.A. and J.F.R. Gower. 1984. Phytoplankton distribution in the Eastern Canadian Arctic. Arctic 37: 224-233.

Clarke, G.L., G.L. Ewing and C.J. Lorenzen. 1970. Spectra of back-scattered light from the sea obtained from aircraft as a measure of chlorophyll concentration. Science 167: 1857-1866.

Gordon, H.R., D.K. Clark, J.W. Brown, O.B. Brown, R.H. Evans and W.W. Broenkow. 1983. Phytoplankton pigment concentrations in the Middle Atlantic Bight: comparison of ship determinations and CZCS estimates. Applied Optics 22: 20-37.

Gower, J.F.R., G.A. Borstad and D.N. Truax. 1984. Optical imaging of the sea surface with high spectral resolution. In Proceedings, 9th Can.Symp. on Remote Sensing. St.John's, Newfoundland. August 14-17, 1984.

Gower, J.F.R. 1980. Observations of in situ fluorescence of chlorophyll a in Saanich Inlet. Boundary Layer Meteorol. 18: 235-245.

Gower, J.F.R. and G.A. Borstad. 1981. Use of the in vivo fluorescence line at 685nm for remote sensing surveys of surface chlorophyll a, p. 329-338. In J.F.R. Gower [ed.], Oceanography from Space. Plenum.

Morel, A. and L. Prieur. 1977. Analysis of variations in ocean colour. Limnol. Oceanogr. 22: 709-722.

Neville, R.A. and J.F.R. Gower. 1977. Passive remote sensing via chlorophyll a fluorescence. J. Geophys. Res. 82: 3487-3493.

PLATES

Plate 1. Distribution of phytoplankton fluorescence (Fluorescence Line Height or FLH) in Barkley Sound as calculated from digital data from the DFO Fluorescence Line Imager, April 25, 1987.

Plate 2. Surface thermal patterns in Barkley Sound from digital 3 µm band data from a Daedalus Multispectral Scanner, April 25, 1987.

Click HERE to return to Papers on Aquatic and Coastal