Module 7
Caribbean Regional Node

To learn how to derive a map of seagrass standing crop from a SPOT XS image.
 
 

Ecologists and managers of seagrass systems may require a range of data on the status of seagrass habitats. Those parameters which can be measured using remote sensing are listed in Table 7.1.

Of the studies listed in Table 7.1, the most detailed obtained a quantitative empirical relationship between seagrass standing crop (biomass of above-ground leaves and shoots) and remotely sensed imagery (Armstrong, 1993; Mumby et al., 1997b). Standing crop is a useful parameter to measure because it responds promptly to environmental disturbance and changes are usually large enough to monitor (Kirkman, 1996). The methods used to map seagrass standing crop are explored in this lesson.

Essentially, this involves five stages:

1. Measurement of seagrass standing crop in the field.
2. Relating seagrass standing crop to remotely-sensed image data using linear regression. The regression equation represents the calibration of imagery to seagrass standing crop and the coefficient of determination (r²) indicates how much of the variation in standing crop is explained by the remotely-sensed data.
3. Using the regression equation to convert (calibrate) the image data in each pixel to seagrass standing crop (g. m^-2)
4. Masking out land and non-seagrass habitats to leave a map of seagrass standing crop.
5. Assignment of a colour palette to standing crop for easy visual interpretation.

This lesson relates to material covered in Chapters 12 and 16 of the Remote Sensing Handbook for Tropical Coastal Management and readers are recommended to consult these for further details of the techniques involved. This lesson describes an empirical approach to mapping seagrass standing crop using SPOT XS. The seagrass was located on the Caicos Bank in shallow (<10 m deep) clear water (horizontal Secchi distance 20 - 50 m) and was dominated by the species Syringodium filiforme (Kützing) and Thalassia testudinum (Banks ex König).

Familiarity with Bilko for Windows 2.0 is required to carry out this lesson. In particular, you will need experience of using Formula documents to carry out mathematical manipulations of images and Histogram documents to calculate the area of particular features on the imagery. A familiarity with Excel is also desirable.

This lesson will use a SPOT XS image of the Caicos Bank as this imagery covers a large area while having a reasonably high spatial resolution (20 m). To obtain a quantitative relationship between image data and seagrass standing crop, it is important to compensate for the effects of variable depth. Therefore, the image you are provided with has undergone the depth-invariant processing described in Lesson 5 of this module. SPOT XS bands 1 (green ) and 2 (red) were processed to form a single depth-invariant bottom index band and you are provided with a subset of the Caicos Bank showing South Caicos (file XS_DI.DAT). This is a floating point file where each pixel value is represented by a real number stored in four bytes. To illustrate the importance of compensating for variable depth, when seagrass standing crop was regressed on reflectance (not corrected for depth), the coefficient of variation (r²) ranged from 0.05 - 0.32 for SPOT XS (and was similarly low for other sensors). These values are too low for adequate prediction of seagrass standing crop.

You are also provided with two mask files. The first is a land mask (LANDMSK7.GIF) in which all land pixels are set to zero and all submerged pixels are set to a value of 1. When another image is multiplied by the land mask, all land pixels are set to zero whilst submerged pixels are unaffected. The second mask removes all non-seagrass habitats in the imagery (SEAGMASK.GIF). This was created from an image of coral reef, algal, sand and seagrass habitats (see Lesson 6 for a similar example). Land, coral reef, sand and algal pixels were set to zero leaving only seagrass habitats set to a value of 1.

A variety of techniques exist for measuring seagrass standing crop including destructive and non-destructive approaches. We recommend non-destructive methods because they allow repeated monitoring, are less harmful to the environment and are generally faster than destructive methods. Good general texts on seagrass sampling include Unesco (1990), English et al. (1994) and Kirkman (1996). The non-destructive visual assessment method of Mellors (1991) for estimating seagrass standing crop in situ was modified for use here.

Standing crop is measured in units of dry weight organic matter per unit area (g. m^-2) and is correlated with shoot density and leaf area index. The visual assessment method uses a linear scale of biomass categories which are assigned to seagrass samples in 0.25m² quadrats. A six point integer reference scale of quadrats is used (1-6). The technique is summarised in Figure 7.1 and a thorough analysis of its errors and limitations is given by Mumby et al. (1997a).
 

At least 30 seagrass sites should be surveyed and these should represent the entire range of standing crop in the area of interest. The size of each site will depend on the spatial resolution of the remotely-sensed imagery; larger sites are required to represent the standing crop in larger pixels. As a general guideline, we found a 10 m diameter site size to be acceptable for SPOT XS imagery. At each site, the mean standing crop of seagrass is estimated from a number of randomly-placed replicate quadrats. For medium to dense seagrass beds, a sample size of 6 quadrats is probably acceptable. However, a larger sample size should be used in patchier seagrass areas (for further discussion see Downing and Anderson, 1985).

This lesson assumes that stage 1, the collection of field data, has already been undertaken using the guidelines set out above. The first practical activity will be the second stage of preparing a map of seagrass standing crop, namely, relating field data to image data to derive a calibration graph.

The calibration equation calculated above was based on only five samples of field data. When we used 110 seagrass sites, we obtained the following equation which is not dissimilar to that which you have already calculated.

This is the standard equation of a straight line where 38.6 is the intercept of the line with the y-axis (i.e. the value of y when x = 0), 5.6 is the magnitude of the gradient and the minus sign indicates that the slope is negative (i.e. as x gets bigger, y gets smaller). As the y-axis variable was the square root of seagrass standing crop, the left hand side of the equation must be squared to give values in original units of seagrass standing crop (g.m^-2).This can be re-written in a slightly easier format to give:

The calibration will be implemented at the same time as applying the land mask. Fortunately, seagrass standing crop does not exceed 255 g. m^-2 at the study site and therefore, it can be represented directly on an 8-bit image with 256 possible values for each pixel (if this was not the case, standing crop would have to be scaled to 255 values).

The next step is to create a colour palette to display the seagrass standing crop as five colour-coded categories corresponding to the top five classes on the visual assessment scale. The palette applies to the display image values (mapped by the Look Up Table - LUT) as opposed to the underlying data values. We want to map the underlying data values to colours and so must make sure there are no stretches in operation. Thus before proceeding further we should clear any stretches of STEP2.GIF.

Now that you have created a map of seagrass standing crop, the final objective is to calculate the area of seagrass in each standing crop category. In a practical situation, this might be repeated every few years to monitor changes in the coverage of particular types of seagrass bed.

A map of seagrass standing crop could be used in several ways. The simplest of these would be the provision of extra information on the status of seagrass beds in a particular area of interest. However, the maps might also be used to identify the location and extent of important fisheries habitats. For example, the juvenile conch (Strombus gigas) is often associated with seagrass habitats of low standing crop (Appeldoorn and Rolke, 1996). Monitoring of seagrass standing crop over time would also be possible, particularly if the remotely sensed data had a high spatial resolution. Changes in standing crop would be displayed in 2-dimensions and possibly related to the causes of the change (e.g. point sources of pollutants). However, if monitoring is going to be undertaken, it is important to quantify errors and assess the confidence with which remote sensing predicts seagrass standing crop. For further details of these methods, readers are referred to Mumby et al. (1997b). Alternatively, a simpler monitoring programme could compare the coverage of seagrass classes from, say, year to year (as outlined in this lesson).

Appeldoorn, R.S., and Rolke, W., 1996, Stock abundance and potential yield of the Queen Conch resource in Belize. Unpublished report of the CARICOM Fisheries Resource Assessment and Management Program (CFRAMP) and Belize Department of Fisheries, Belize City, Belize.

Armstrong, R.A., 1993, Remote sensing of submerged vegetation canopies for biomass estimation. International Journal of Remote Sensing 14: 10-16.

Cuq, F., 1993, Remote sensing of sea and surface features in the area of Golfe d'Arguin, Mauritania. Hydrobiologia 258: 33-40.

Downing, J.A, and Anderson, M.R., 1985, Estimating the standing biomass of aquatic macrophytes. Canadian Journal of Fisheries and Aquatic Science 42: 1860-1869.

English, S., Wilkinson, C., and Baker, V., 1997, Survey Manual for Tropical Marine Resources, 2nd Edition. (Townsville: Australian Institute of Marine Science).

Ferguson, R. L., Wood, L.L., and Graham, D.B., 1993, Monitoring spatial change in seagrass habitat with aerial photography. Photogrammetric Engineering and Remote Sensing 59: 1033-1038.

Greenway, M., and Fry, W., 1988, Remote sensing techniques for seagrass mapping. Proceedings of the Symposium on Remote Sensing in the Coastal Zone, Gold Coast, Queensland, (Brisbane: Department of Geographic Information) pp. VA.1.1-VA.1.12.

Kirkman, H., 1996, Baseline and monitoring methods for seagrass meadows. Journal of Environmental Management 47: 191-201.

Kirkman, H., Olive, L., and Digby, B., 1988, Mapping of underwater seagrass meadows. Proceedings of the Symposium Remote Sensing Coastal Zone, Gold Coast, Queensland. (Brisbane: Department of Geographic Information) pp. VA.2.2-2.9.

Lennon, P., and Luck, P., 1990, Seagrass mapping using Landsat TM data: a case study in Southern Queensland. Asian-Pacific Remote Sensing Journal 2: 1-6.

Luczkovich, J.J., Wagner, T.W., Michalek, J.L., and Stoffle, R.W., 1993, Discrimination of coral reefs, seagrass meadows, and sand bottom types from space: a Dominican Republic case study. Photogrammetric Engineering & Remote Sensing 59: 385-389.

Mellors, J.E., 1991, An evaluation of a rapid visual technique for estimating seagrass biomass. Aquatic Botany 42: 67-73.

Mumby, P.J., Edwards, A.J., Green, E.P., Anderson, C.W., Ellis, A.C., and Clark, C.D., 1997a, A visual assessment technique for estimating seagrass standing crop. Aquatic Conservation: Marine and Freshwater Ecosystems 7: 239-251.

Mumby, P.J., Green, E.P., Edwards, A.J., and Clark, C.D., 1997b, Measurement of seagrass standing crop using satellite and airborne digital remote sensing. Marine Ecology Progress Series 159: 51-60.

Robblee, M.B., Barber, T.R., Carlson, P.R., Jr, Durako, M.J., Fourqurean, J.W., Muehlstein, L.K., Porter, D., Yarbro, L.A., Zieman, R.T., and Zieman, J.C., 1991, Mass mortality of the tropical seagrass Thalassia testudinum in Florida Bay (USA). Marine Ecology Progress Series 71: 297-299.

Savastano, K.J., Faller, K.H., and Iverson, R.L., 1984, Estimating vegetation coverage in St. Joseph Bay, Florida with an airborne multispectral scanner. Photogrammetric Engineering and Remote Sensing 50: 1159-1170.

Sheppard, C.R.C., Matheson, K., Bythell, J.C. Murphy, P., Blair Myers, C., and Blake, B., 1995, Habitat mapping in the Caribbean for management and conservation: use and assessment of aerial photography. Aquatic Conservation: Marine and Freshwater Ecosystems 5: 277-298.

Unesco, 1990, Seagrass Research Methods. Monographs on Oceanographic Methodology, 9. (Paris: Unesco).

Zainal, A.J.M., Dalby, D.H., and Robinson, I.S., 1993, Monitoring marine ecological changes on the east coast of Bahrain with Landsat TM. Photogrammetric Engineering & Remote Sensing 59: 415-421.

1. The calibration equation derived from the 5 data points is y = 42.237 - 6.1762x. The y-axis represents the square root of seagrass standing crop. The x-axis represents the pixel value of the depth-invariant bottom-index image derived from SPOT XS bands 1 and 2.
2. The coefficient of determination (r²) = 0.98 [0.9777].


In the block of 41 x 36 seagrass pixels starting at coordinates 338, 143:

3. The lowest value of the seagrass pixels is 7, representing a standing crop of 7 g.m^-2. One pixel has this value.
4. There are 3 pixels with a standing crop of 253 g.m^-2.
5. The modal (most frequent) pixel value is 142, indicating a modal standing crop of 142 g.m^-2. Eighty-one pixels have this value.

6. 136 pixels have a value between 201 and 254. This represents an area of 7.2 ha of high density seagrass.

Back to main page

Back to Module 7

USDOC | NOAA | NESDIS | CoastWatch