|
|
|
Land Use
|
Polarimetric Radar Data Analysis of Key Landcover Features in the Coastal Panay Area: A Preparatory Tool for Polarimetric SAR Image classification
Results and Discussion
Figure 2. Comparison of mean backscatter coefficients among different landcover types taken from the sample areas. The cross-polarized L-band seem to possess the least backscatter value among all of the signals followed subsequently by all of the cross-polarized.
Figure 2 alone indicates strong correlation between backscatter coefficient values for seawater and the fishponds. It is also evident that both bodies of water have relatively smaller backscatter values due to the tendency of their surfaces to act as specular reflectors, returning electromagnetic waves away in a direction opposite to incidence wave. Higher backscatter coefficient values for the urban areas are likewise observed possibly because of a double bounce scattering mechanism which building, houses and other similar structures commonly found in built-up areas, acting as corner reflectors, exhibit. On the other hand, backscattering from forests appears to have two main components: one is a two bounce corner reflector mechanism and the other is a diffuse component which can be attributed to multiple scattering behaviour.
Figure 2. Comparison of mean backscatter coefficients among different landcover types taken from the sample areas. The cross-polarized L-band seem to possess the least backscatter value among all of the signals followed subsequently by all of the cross-polarized.
Figure 3. Different Polarization signatures for five different landcover types all plotted in co-polarized C-band.
Sample regions taken from the build-up areas have different co-polarization signatures shapes for the three bands. The L-band polarization signature of the urban areas demonstrates a non conforming cross-polarized signature, i.e. it did not represent the predicted reversal of its co-polarized counterpart which is common for the other landcover types.
As expected, the fishponds exhibited a co-polarization signature shape similar with of the seawater. This is a result which reinforce early results from the mean value plots of the training areas. There were however discernable difference in the cross-polarization plots of fishponds most prominent in the C-band. This can be attributed to the influence of sea winds on surface water where wave heights up to twentieth of a meter can be observed due to the seasonal winds at the time of observation.
Among the land cove types being described, the forested areas, which display a multiple scattering mechanism, has the highest coefficient of variation indication diversity in the scattering mechanism within the sampled area.
5. Summary and conclusion
Two methods to analyze AirSAR data were developed for this study. One technique is by computing for the ordinary statistical indicators and relate them to the backscattering responses of targets. The second method involves plotting the plarimetric signatures for the different landcover types, examining the plots visually and relating these to the coefficients of variation of the signatures. The comparison of the C,L and P-band SAR images in terms of their scattering responses and polarization signatures increase understanding of land cover polarimetric characterization and serve as a useful input for land cover discrimination. It is recommended that further research is needed mathematically relate the characterization of the polarization signatures to their respective landcover types to yield meaningful results when performing SAR image classification.
Acknowledgement
The authors wish to acknowledge the kind support of the Philippine Government, particularly the Philippine council for Advanced Science and Technology Research and Development of the Department of Science and Technology for making the acquisition of the polarimetric radar data possible through the Philippines -NASA AIRSAR Project and for generously funding the research.
Reference:
-
Basili, P,P. ciotti, G.D Auria, F.S. marzano, N. Pierdicca and P. Quarto (1996), "Assessment of polarimetric features of discriminate land cover from the MAESTRO I campaign" International Journal of Remote Sensing, 15(4), pp. 2887 -2899.
- Evans, Diane L., Tom G. Farrr, jakob J. Van Zyl and oward A. Zebker, (1988) " Radar Polarimetry: Analysis Tools and applications" IEEE Transactions on Geosciences and Remote Sensing, 26(6), pp. 774-789.
- Van Zyl, Jakob. J., Howard. A. Zebker, and c. Elachi, (1987) "Imaging radar polarization signatures: Theory and Observation, "Radio Science, 22(4), pp. 529-543.
- Yun Shao, Lin Hao, Lu Xinqiao, Gu Huadong (1997), "Effect of Frequency and Polarization of SAR system on Target Detection", Proceeding of the Chinese conference on Remote Sensing, Beijing, China, pp. 418-426.
- Zebker, Howard A., Jakob J. van Zyl, and Daniel N. Held (1997), "Imaging Radar Polarimeter from Wave Synthesis", Journal of Geophysical Research, 92(B1), pp. 638-710.
|
|
|
|
|
|
|
