Riparian vegetation has been disappearing from the Swan Coastal Plain since
the beginning of European settlement over 200 years ago. Vegetation
around Peel-Harvey has been cleared to provide farming and grazing land for
the growing population. Later, more vegetation was cleared to make
way for spreading urban areas. Recently the different hydrologic
regime, due to opening of Dawesville Cut may eventually alter the vegetation
density and in particular the associations of vegetation fringing the estuary
and incoming rivers. Erosion due to greater tidal flushing and higher
average salinities throughout the year may contribute to the overall shift towards
salt tolerant species (McComb et al, 1995).
An outline map of the study area and various locations is shown in Figure
1.
Existing maps of vegetation of Peel-Harvey system vary in the detail, scale, resolution and type of vegetation mapped. McComb et al, (1995) describes the saltmarsh flats of the estuary, including those of Serpentine River up to Lake Goegrup in the north. All saltmarsh areas were mapped using aerial photography of 1994 and using historical photographs going back to 1957 Glasson et al (1995). This study mapped all areas where saltmarshes occur and identified individual plant species along transects established in 10 locations around the estuary (Figure 1).
Previous large scale mapping of vegetation in the area includes Siemon et al (1993) which covered in detail the banks of the Serpentine River. Aerial photography interpretation and field work formed the basis of vegetation association maps. Six major categories were identified in the study: saltmarsh, estuarine fringing forest, fringing vegetation, sandy rise vegetation, freshwater fringing vegetation and disturbance related plant assemblages (Table 1). Within each major category between 3 to 4 vegetation associations were mapped. Trudgen’s (1991) survey covered the coastal strip south of Mandurah, including the northern section of Lake Clifton.
Kinnaird et al (1979) compiled a vegetation map, which identifies eight vegetation associations, however their definitions have not been included. Comparing just one association of saltmarsh from McComb et al, (1995) and Kinnaird et al (1979), we noticed a large discrepancy. For example the area of Creery Wetlands was mapped by Kinnaird et al (1979) as “Saline Wetland Complex (salt paperbark, swamp sheoak)”. Previous vegetation mapping projects of the Peel-Harvey area have been summarized in Table 1.
It should be evident from Table 1 that no single study exists for the entire study area. In completion of each individual study above no consideration is given by the respective authors to a single consistent methodology over the entire study area. Indeed comparison of the respective studies is made difficult by the differing classes used in each. This may simply be due to the inability of the methods and resources at the time, to cover such a large area. For this reason remote sensing methods are eminently suitable to this task of coverage of a large area using a classification system that can be applied consistently. Remote sensing may employ subjective analogue methods such as air photo interpretation or objective digital classifications using airborne scanner or satellite data. An understanding of the method of collection of data, is critical in understanding the end classification system.
There are existing vegetation monitoring programs in the Peel-Harvey area by
CALM and PIMA involving transects and these are shown in Figure
1. Most of these transects have been established and surveyed recently
for saltmarsh distribution (McComb et al,
1995).
Mausel et al (1992) have reviewed currently available systems in airborne videography. They have addressed common technical problems of such systems. Some of those which were found in that study were of radiometric and geometric calibration. Mausel et al (1992) report on a growing number of studies which develop calibration models for the data and allow video digital count data to be converted to reflectance. Image registration (geometric calibration) has also been mentioned as one requiring attention (Mausel et al 1992)
Airborne video imagery has been used for vegetation mapping for over a decade, for example Everitt and Nixon (1985); King and Vlcek (1990); Nixon et al (1985). Digital videography is a large and growing field in environmental mapping and monitoring with special workshops and conferences (for example the ASPRS Special Workshop on Videography in 1988 and the ASPRS Biennial Colour IR Photography and Videography Conferences).
Aerial images are especially suited to the study of vegetation, water resources and shoreline mapping (A.S.O.P., 1968). Aerial images provide a perspective of the earth’s geographical features that are generally visually intuitive and can be spatially compared to an individuals own knowledge of a given area. Aerial images frequently provide working documents for planners and managers. They are however limited in spatial accuracy as these images suffer geometric distortions particularly near the margins or when the terrain is varied in height. Also, a single aerial image rarely covers an entire study area. There are manual and computer assisted techniques of joining images and also eliminating the geometric distortions present within and between the individual images. The assembly of aerial photographs into a single image is called a photo mosaic. Photo mosaics can be controlled or uncontrolled, the former having the geometric distortions removed.
The amount of geometric distortions depends on many factors including the physical optics of the camera and the orientation of the camera at the instant of exposure. Where the camera’s optical axis is near vertical (<3° from vertical) then the image is said to be vertical. The point where the optical axis of the camera meets the earth’s surface is said to be the principal point of the image. Vertical images are the most common type of metric aerial image, or one that is used for deriving information about spatial measurements of various geographical features. Geometric distortion on aerial images increases outwards from the principal point towards the margin. The distortions are greater if images are acquired at low altitude, or with cameras using wide angle (>70°) and super-wide angle (>100°) lenses. Gross distortions of scale occur on individual images when terrain slope suddenly changes, such as when a scarp, cliff or portion of a mountain is included in the image (Maling, 1989). Some of the distortion in vertical air images can be minimised by using only central portions of each image.
Aerial images are acquired in flight lines called runs. Each image within a run overlaps the previous by about 60% (endlap) to enable sufficient imagery to be acquired that is relatively distortion free, and to enable stereoscopic viewing of the imagery to aid in feature identification and interpretation. Each run is overlapped (sidelap) to the adjacent run although usually only 25-30% overlap is obtained (Figure 2). Physical assembly of a image mosaic by cutting and pasting is an exacting process where failure can be very costly, given individual colour photographs can cost about $25.00 and a mosaic may include around 50 photographs. Digital photo mosaics can be constructed and repeated as necessary without fear of costly failure. This investigation used the digital equivalent of a photo mosaic using computer software and DMSV imagery of the Peel-Harvey Estuary.
Air photos and mosaics have frequently provided planning resources but increasingly the use of remote sensing data is replacing this traditional media. The use of satellites has enabled data capture over large contiguous areas in limited time and provided rapid and timely data. The ease of use of remote sensing data is due to its digital format which enables rapid transmission and processing. It is however costly and this limits its use to larger projects or long term studies. A limitation of satellite sensors (passive sensors) is their periodicy over target areas and their exclusion from data capture if extensive cloud cover is present during the overpass. There is a growing requirement for platforms which can provide digital data with rapid response times and data processing. This has led to increasing use of airborne scanners, however many of these are also costly to operate and maintain, and justify usage when smaller projects or studies are undertaken. SpecTerra systems DMSV is designed to suite the requirements of rapid response and data processing using cost efficient light aircraft.
Remote sensing as a tool for any investigation involves a number of steps. The steps taken can generally be described as including:
A previous study using digital air photographs mosaics covering the period
1957 to 1994 to determine temporal changes in the extent of samphire coverage,
formed the baseline data for this study (Glasson
et al, 1995). The selected areas for long term monitoring
for this study are the same as those used for the samphire study.