Precise Geometric and Radiometric Corrections of Spot Pan & XS Data For Land Use Monitoring
4. Mergin of Pan and XS
After registering and resembling the XS image to the geometry and resolution of the geocoded Pan image, a multispectral image with 10 meters resolution can be created by means of data fusion. The standard approach is to transform the XS image from the RGB (Red, Green, Blue) to the HIS (intensity, Hue, Saturation) color space, to replace the intensity channel with the Panchromatic image, and finally to perform the reverse transformation to the RGB colour space. However, this does not work very well with SPOT imagery because there is very little correlation between the Pan and the XS3 (Infrared ) bands. A study [DARVISHEEFAT] compared several approaches and found a radiometric procedure to yield best results for vegetation (esp. forest) mapping. This procedure was initially developed by the Ditital Image and Remote Sensing Laboratory (DIRS) at the Rochester Institute of Technology [MUNECHIKA]. It is based on the overlap of the spectral responses of the sensor channels. First, a synthetic cannel SYN which represents the sum of spectral response overlaps of the XS bands with Pan is computed as follows:
The weights wi are w1 = 0.454
W2 = 0.546
W3 = 0.000
Second, the Panchromatic band is edge enhanced with a 5 x 5 average filter and its histogram adjusted to the synthetic band:
| a = |
sSYN
---------------
sPAN |
sSYN = standard deviation of synthetic band
sPAN = standard deviation of Pan banc
SYN = average of synthetic band
PAN = average of Panchromatic band
Finally, the merged (hybrid) bands are computed as follows:
| HYBi = XSi . |
PANadj
----------------
SYN |
The result is a color image with 10 m resolution
5. Outlook
In the forthcoming months, we will compare the imagery with the digitized land use maps of the 1980s on screen and try to identify areas of major and use changes. Subsequent field work will show us, what type of land use changes can be seen, and how good the reliability of identification is. A comparison with standard with standard processed imagery is also planned to determine whiter the extra effort is really worthwhile
Acknowledgements
The authors wish to thank Dr. H. Manthrithilake, Mr. R. White and Mr. T. Senanayake of the Environment and Forestry Division of Mahaweli Authority for supporting this study with DEM and satellite image data, the Dept. Of Meteorology for providing atmospheric data, and Dr. S. Sandmeier of Zurich University for advising on the radiometric corrections.
References
-
DARVISHSEFAT, AA (1995): Einsatz und Fusion von Multisensoraten Satellitenbilddaten zun Erfassung von Waldinventurn (in German)
Remote sensing Series Vol 24, Dept. of Geography, University of Zurich
-
ITTEN, KI & P MEYER (1993): Geometric and Radiometric Correction of TM Data of Mountainous Forested Areas
IEEE Transaction on Geoscience and Remote Sensing Vol. 31. No. 4
-
Munechika, CK (1990): Merging Panchromatic and Multispectral Image for Enhanced Image Analyis
MS thesis, Rochester Institute of Technology, NY USA
-
SANDMEIER, SR (1995): A Physically-Based Radiometric Correction Model : Correction of Atmospheric and Illumination Effects in Optical satellite data of rugged terran
Remote Sensing Series Vol. 26,Dept. of Geography, University of Zurich
-
SRI LANKA SWISS REMOTE SENSING PRO-JECT (1988): Final Report
Sri Lanka Studies Vol. 2, Dept. of Geography, University of Zurich
-
TURNER, STD & R White (1994):
Geographical Information System for natural resource management in South East Asia
Environment and Forest Conservation Division, Mahaweli Authority of Sri Lanka
-
VERMOTE et al (1994)L Second Simulation of the Satellite Signal in the Solar Spectrum Signal in the Solar Spectrum (6S); Users guide
NASA GSFC, Greenbelt MD, USA
-
ZEE, D van der & JA Cox (1988): Monitoring in Moneragala District, Sri Lanka with SPOT images
ITC Journal No 3, 1988
