Figure 1. Location of the ERS-1 SAR scene of the agricultural area of the Eastern Darling Downs, Queensland, Australia.
Figure 1 shows the 100 x 100 km location of the SAR scene of the Eastern Darling Downs, Queensland, Australia. The Darling Downs is an important agricultural area of Australia. A fast-delivery image (FDI) was purchased from the Australian Centre for Remote Sensing (ACRES). This image was obtained from a descending pass at 23:53 UTC on 19 October 1992 (orbit 6602). This FDI was a four-look level-1.0 product with 12.5 x 12.5 m pixels without radiometric or geometric corrections in CEOS format.
Figure 1. Location of the ERS-1 SAR scene of the agricultural area of the Eastern Darling Downs, Queensland, Australia.
As part of this investigation the FDI was geometrically corrected using a cubic convolution based on twelve ground control points using software called ERDAS by GeoImage Pty Ltd.
When the photographic print was examined to find the signature of a corner reflector that had been placed in the image area, a very large 'cross' pattern was observed at a location that had not been ground truthed (Fig. 2). The size of the 'cross' was approximately 15 km across track and 8 km along track (the original FDI data confirmed this). It was interesting to note that there were three dark regions in the centre of the 'cross' (the FDI revealed more than three regions in a line at a small angle to the west of north). The FDI also revealed that these dark regions were in fact formed from digital numbers with a maximum value of 65535 (which is in fact the 2's complement of -1).
Figure 2. A photograph of an area (approx. 18 x 13 km) of the geometrically corrected SAR image containing the unexpected 'cross' signature (black lines are northings and eastings; white lines are latitude and longitude)
It was decided to physically locate the position of the 'cross' centre using a GPS to find out what had caused this unusually large signature. On 24.1.94 at latitude -27 degrees 45'53.8'' and longtitude 151 degrees 32'26.0'' the cause was found at a totally unexpected site just south of a small forest of native trees (Fig. 3, Fig. 4,Fig. 5). The three photographs show different angle shots of a 'chook' farm (a farm having thousands of hens laying eggs in huge sheds). Figure 3 was taken facing southwest showing the six large sheds (each shed approximately 100 x 30 m). Figure 4 was taken looking south and shows a length shot of the first shed. Figure 5 is looking west along the first shed and shows the angle of the rooves at approximately 11 degrees (seven of these on most of the sheds). In all there were six of these sheds aligned east-west and lined-up in a north-south direction separated by about 50 m between each shed. A further shed of slightly smaller size was located about 200 m to the west of the fourth and fifth sheds also aligned east- west.
Although not present at the time of taking the photographs, it was reported by the property owner (of Cathnan Pty Ltd.) that the surrounding dams and channels were reasonably full of water at the date of the satellite pass.
Figure 3. A photograph looking south-west showing the six sheds of the Cathnan 'chook' farm.
Figure 4. A photograph looking south showing the length view of the first shed.
Figure 5. A photograph looking west of the first shed showing two of the seven rooves.
As this has been only a side issue from my main research project, I can only speculate (after brief discussions with a number of remote sensing researchers) that the large 'cross' signature is a result of multiple scattering between the sheds (walls and rooves) and surrounding water at the site. The dark regions would indicate an overloading of a processor (somewhere in the data processing sequence) and correspond with each shed. I can only speculate as to whether this signature would have disappeared if the image was radiometrically corrected or a higher level product had been available. With a few other large 'chook' farms in the area with similar buildings and alignments, it is surprising that they did not show up as large 'crosses' like this one.
ESA EOQ Nr. 48.