ved from
alues.
or water in
it of 0.992
1)
roximately
DN value.
nent in the.
N for TM3,
to 26 at the
terms of
ig/l. across
rms of the
Omg/l. SS
oise due to
1 high pass
of median
ing a 3*3-
values was
iteratively
e Standard
| currently
shown in
e data
>monstrates
d the local
boundaries
difference
Mapper is
reflectance
to a greatly
also more
pared with
mogeneous
| a window
ated by by
nses within
the window. The S/N ratio is then calculated after Smith and
Curran (2000) as follows:-
SNR - R/R
(Eq. 2)
The Signal-to-Noise ratio for four homogeneous areas within
the image for the raw unprocessed image (image 1), the image
after removal of the 16" line banding (image 2) and the image
after removal of the vertical coherent noise (image 3) is shown
in Table 1. The mean values are 35, 36 and 113 respectively.
IAPRS & SIS, Vol.34, Part 7, “Resource and Environmental Monitoring", Hyderabad, India,2002
DN —— after destriping
after noise removal
"Lo MULA LU
"Y "m
24
Distance along image transect
Figure 4. Transect showing DN values across the sediment
plume shown in Figure 1 after removal of the 16-line banding
(destriping) and after additional removal of the vertical
coherent noise.
Sample Image Ra Rsd SNR
No. Type
Image 1 23.8155 | 0.6300 | 37.8023
Sample 1 | Image 2 23.8431 | 0.6269 | 38.0331
Image 3 24.0078 | 0.0882 | 272.1693
Image 1 25.1108 | 0.6555 | 38.3093
Sample 2 | Image 2 25.1215 | 0.6546 | 38.3796
Image 3 25.0905 | 0.2869 | 87.4629
Image 1 31.4246 | 0.9675 | 32.4810
Sample 3 | Image 2 31.4145 | 1.0009 | 31.3858
Image 3 31.3962 | 0.7492 | 41.9042
Image 1 34.3528 1.0448 | 32.8802
Sample 4 | Image2 34.3382 | 1.0349 | 33.1806
Image 3 34.3760 | 0.6398 | 53.7307
Table 1. Signal-to-Noise Ratios for the raw unprocessed image
(image 1), the image after removal of the 16" line banding
(image 2) and the image after removal of the vertical coherent
noise (image 3) (Ra = mean radiance within a 10*10 pixel
window; Rsd= Standard Deviation of radiance values within
this window; SNR = Signal-to-Noise ratio)
3. CONCLUSION
At the launch of LANDSATSs 4 and 5, an element of image
noise less than 2 DN values was regarded as acceptable and was
not addressed as part of basic image quality control at ground
receiving stations. Moreover, the noise discussed in this paper
is only readily apparent over homogeneous surfaces
characterized by few quantisation levels in the LANDSAT TM
sensor, such as water. However, during the last 25 years, fears
of sea level rise and rapid human-induced changes to coastlines
and coastal ecosystems worldwide have increased the need for
sensing systems capable of recording subtle spatial and
temporal differences in coastal waters. There are many
instances where sharp, rather than diffuse boundaries within
waterbodies are required. These include determination of the
land-water boundary on shallow or vegetation-infested coasts
using LANDSAT TM near infra-red waveband (Melsheimer
and Liew, 2001); détermining precise bathymetric limits for
navigation purposes (ibid.); and determining the position of the
saline-freshwater interface for extraction of water for various
types of use and the management of coastal ecosystems which
are sensitive to changes in salinity levels.
As more countries become aware of imposing water quality
standards for different types of use (eg. a maximum SS content
is generally specified for drinking water quality and 80 mg/l.
for fish culture and aquatic resources), then the magnitude of
error of +/- 10mg/1.SS suggested for the vertical coherent noise
investigated in this paper, may be regarded as unacceptable.
4 REFERENCES
References from Journals:
Crippen, R.E., 1989. A simple spatial filtering routie for the
cosmetic removal of scan-line noise from Landsat TM P-
Tape imagery. Photogramm. Eng. Remote Sens. 55(3), 327-
331,
Murphy, J.M., Ahern, F.J. and Duff, P. 1985, Assessment of
radiometric accuracy of LANDSAT 4 and. LANDSAT 5
Thematic Mapper data products from Canadian production
systems, Photogrammetric Engineering and Remote
Sensing 51(9), p.1359-1369.
Poros, D. J. and Petersen, C.J. 1985 A method for destriping
LANDSAT Thematic Mapper images: a feasibility study
for an online destriping process in the Thematic Mapper
image processing System (TIPS). Photogrammetric
Engineering and Remote Sensing 51(9), p.1371-1378.
Xia, Li, 1993, A united model for quantitative remote sensing
of suspended sediment concentration. International Journal
of Remote Sensing, 14(14):2665-2676
References from books:
Pease, C.B. 1991 Satellite imaging instruments. New York,
Ellis Horwood.
Smith, G.M and Curran, P. J. 2000 Methods for estimating
image Signal-to Noise Ratio (SNR), in Atkinson and Tate
(eds.) Advances in remote sensing and GIS analysis.
Wiley, Chichester, UK. p. 61-74.
Other references:
Melsheimer, C. and Liew, S.C., 2001, Extracting bathymetry
from multitemporal SPOT images. In Proc. of 22" Asian
Conference on Remote Sensing, Singapore, 5^-9th
November, 2001, pp. 104-109
Taylor, D.M., Couturier, Sanderson, P., Lee 2001 A
methodology for determining variations in the quality of
coastal waters off southeast Sumatra, Indonesia using
SPOT-XS data. Reference pending
