Qualitative Analysis of Remote Sensing Datasets for Vegetation Discrimination

Visual analysis of the imagery indicated that two categories of vegetative ecosystems can be discriminated in TM data: uplands and wetlands. Many sub-environments exist within each of these categories.

The wetlands of KSC consist of salt, fresh, and brackish water marshes in addition hardwood swamps consisting of willows, red maple, cabbage palms and red bays. Currently, KSC has classified the freshwater marshes as graminoid marshes which may include the following species; Spartina bakeri (sand cordgrass), Cladium jamiensis, Juncus rosmerianus (black rush), and Typha domingensis and T. latifolia (cattail). Of these vegetative types, cattail is believed to be confidently distinguished from the remaining types. For those other types, too much uncertainty exist to classify these types into their own separate classes and therefore have been grouped together as graminoids. Many of the graminoid marshes occur as ephemeral swales which are positioned between old relict dunes. Many cattail marshes are typically associated with the mosquito impoundment or in freshwater marshes where anthropogenic activities has altered the natural system.

Saltwater marshes at KSC are non-tidal and typically higher marshes that remain unflooded for most of the year with the exception of isolated ponds and creeks. Most of the saltwater marshes are impounded and are typically composed of Spartina backeri (sand cordgrass), Juncus roemerianus (black rush), Distichlis spicata (salt grass), and other salt tolerant grasses. Spartina alterniflora (Spartina) which approaches its southern limits in the vicinity of KSC is only a small component in these marshes but in recent years seems to have become a more obvious species. Many of the saltwater marshes were invaded by Avicennia germinans (black mangrove) and Laguncularia racemosa (white mangrove), especially along the earthen dikes built during the impoundment program of the 1950's. Mangrove is also common along shorelines at KSC.

Upland vegetated areas on KSC include oak and saw palmetto scrub, slash pine flatwoods, and hardwood hammocks. Pure oak scrub often occurs in the best drained soils which are typically inland sites, whereas saw palmetto scrub becomes more dominant on sites that are less well drained. Historically, these two scrub areas have been mapped as a single unit due to the difficulties in discrimination between them using aerial photography. Slash pine flatwoods occur in the central portion of Merritt Island and can have either open or dense canopy. The undergrowth is dominated by oaks, saw palmetto, and herbaceous vegetation. Hammocks are closed forests with a typically higher canopy than surrounding scrub vegetation and are dominated by evergreens (oaks). Cabbage Palm hammocks are dominated by a canopy of Sabal palmetto (Cabbage palm) and may or may not have an undergrowth of saw palmetto.

In a visual comparison of the Landsat-TM and SPOT XS data, some differences are apparent in the reflectance values of the wetland areas. The Landsat-TM data has a stronger response in the red band than SPOT; however, it is not known whether these differences are due to the spectral characteristics of the data or the response of the vegetation at different times of the year. For an accurate comparison of these data sources, both scenes should be acquired at the same time of year to account for seasonal variations.

1987 TM (Bands 3,2,1)	1987 TM (Bands 4,3,2)
1987 TM (Bands 4,5,7)

In further comparison of wetland areas, another advantage of Landsat-TM is that the thermal infrared channel indicates some 'hot spots' in the saltwater marsh and mud flats which indicate areas of shallow, standing water. The improved spatial resolution of SPOT XS can potentially provide information for better identification and delineation of boundaries between inundated and non-inundated areas.

Upland areas are distinguishable in both the Landsat-TM and the SPOT XS data, although boundaries can be discriminated better in SPOT XS. Both the Landsat-TM and SPOT XS can separate areas such as oak/saw palmetto scrub from slash pine flatwoods. Areas containing oak hammocks, cabbage palm hammocks, and hardwood areas are not as readily apparent from visible inspection of the imagery.

The SIR-C radar data appears to contain different information than the optical Landsat-TM and SPOT XS imagery. While the optical/infrared sensors respond to spectral characteristics of the vegetation, the radar responds to its structural characteristics. In the radar imagery, many upland vegetation, such as slash pine flatwoods, are distinguishable from oak scrub/saw palmetto cover. The taller pine trees have a bright response due to the 'double bounce' of the wave in longer wavelengths. Differences in marsh areas are also quite noticeable in the SIR-C data. Typically, the marsh vegetation and soil appear as dark responses in the radar imagery. However, some freshwater marsh areas, which could contain mangrove or willow, have a bright response in both L-Band and C-Band. Overall, even considering that the SIR-C data was acquired during heavy rain and is only a single polarization, the SIR-C data is an interesting and potentially useful dataset for vegetation mapping .

Using the current datasets, it is difficult to identify individual species within a particular sub-environment due to the heterogeneity of the landscape. For example, KSC contains several brackish marshes which contain vegetation that are tolerant in near saline to near fresh conditions. A method to identify 'species level' vegetation cover will require data with superior spatial resolution such as those acquired by airborne sensors or the next generation of satellites. Of existing sensors, Landsat-TM and the SPOT XS data are useful in identifying broader environmental groups and to some extent sub-environment types, but airborne imagery, such as AIRSAR, CAMS, and AVIRIS is necessary for detailed mapping.

Last Modified: Wed Apr 14, 1999
CSR/TSGC Team Web