The NASA/JPL AIRSAR is a side-looking imaging radar system which utilizes the Synthetic Aperture principle to obtain high resolution images which represent the radar backscatter of the imaged surface at different frequencies and polarizations. The image coordinates are usually slant or ground range and azimuth. Slant range is a measure of the radial distance of the image object from the radar. Ground range is a measure of the distance of the object from a point directly below the aircraft. Near range in the images is closest to the aircraft and far range is furthest away. The azimuth dimension in the images is parallel to the line described by the track of the aircraft (see figure). The spatial resolution of the AIRSAR images, which determines how far apart two objects have to be in order to separate them in the image, is on the order of 12 x 12 meters. Note that this is not the same as the range or azimuth pixel spacing which is contained in the variable format header of the AIRSAR data.
The information contained in each pixel of the AIRSAR data represents the radar backscatter for all possible combinations of horizontal and vertical transmit and receive polarizations (i.e. HH, HV, VH, and VV). The data is in compressed Stokes matrix format, which has 10 bytes per pixel. From this data format it is possible to synthesize a number of different radar backscatter measurements. The MacSigma0 and MacMultiView software packages allow the synthesis of images and analysis of data for some of the more important radar backscatter measurements for AIRSAR data. Each pixel contains information about the radar backscatter for a single frequency only. Other image files may be available which contain information about the radar backscatter at different frequencies. The radar backscatter measurement is a strong function of incidence angle, which is defined as:
where h is the altitude of the aircraft (typically about 8 km), R0 is near range in the image (typically 8-10 km), [[Delta]]R is the size of the (slant) range pixels in meters and i is the line number of the object in the image. The incidence angle is defined with respect to the normal to the Earth's surface. The above expression is strictly valid for a flat Earth only. Radar backscatter usually has a tendency to decrease as the incidence angle increases. This effect can often be seen in AIRSAR images, where the incidence angle is typically in the range of 0 to 70 degrees. The smaller incidence angles correspond to near range in the AIRSAR data, while the larger incidence angles correspond to far range.
Much of the data produced by the AIRSAR is now calibrated, so that the radar backscatter measurements are in normalized radar cross section format (m2/m2) or [[sigma]]0. [[sigma]]0 is the radar cross section (measured in m2) normalized by the area of the measurement, which in this case is the pixel area in square meters. Results obtained using uncalibrated AIRSAR data should be treated with caution. If a trihedral corner reflector is deployed within the imaged scene and a few simple measurements about that reflector are known, then there are facilities in the MacSigma0 and POLCAL software to lend assistance.
The direct measurements of radar backscatter which are made by the AIRSAR systems are the elements of the scattering matrix: Shh, Shv, Svh and Svv, which represent the backscatter for each pixel in response to the 4 different combinations of H and V transmit and receive polarizations. Shv is the scattering matrix term for H transmit, V receive, for example. The scattering matrix measurements are complex numbers, with amplitude and phase.
The Shv and Svh measurements are equal by the principle of backscatter reciprocity. In AIRSAR data, only the Shv term is retained. The AIRSAR data is issued in compressed Stokes matrix format. An intermediate step in forming the Stokes matrix for each pixel is to form the cross-products:
These cross-products can be recovered from the Stokes matrix data format, which can be considered as a type of coding. The first three are measurements. The last three are complex numbers, with amplitude expressed in and a phase term which can be expressed in degrees. The phase is between -180 and 180 degrees. In displaying the ShhSvv* term using MacSigma0, for example, the amplitude is displayed as a grey scale, the phase as a color representation.
Experience with AIRSAR data has shown that the first four of the above cross-products contain the most information about the imaged surface.
The reader interested in more background information on radar polarimetry is referred to the papers listed.
The following diagram may clarify some of the geometric terms mentioned previously:
The compressed Stokes matrix data format is composed of 10 bytes per pixel preceded by two ASCII headers, the variable format header and the "old" header. Both the variable format header and the "old" header are composed of 50-character fields where one character is saved per byte. In general, a compressed Stokes matrix data file looks like the following:
The following information is contained in the variable format header:
1 RECORD LENGTH IN BYTES = NNNNNNNN
2 NUMBER OF HEADER RECORDS = NNNNNNNN
3 NUMBER OF SAMPLES PER RECORD = NNNNNNNN
4 NUMBER OF LINES IN IMAGE = NNNNNNNN
5 NUMBER OF BYTES PER SAMPLE = NNNNNNNN
6 JPL AIRCRAFT SAR PROCESSOR VERSION XXXX.XXXX
7 DATA TYPE = CCCCCCCCCC
8 RANGE PROJECTION = CCCCCC
9 RANGE PIXEL SPACING (METERS) = XXXX.XXXX
10 AZIMUTH PIXEL SPACING (METERS) = XXXX.XXXX
11 BYTE OFFSET OF OLD HEADER = NNNNNNNN
12 BYTE OFFSET OF USER HEADER = NNNNNNNN
13 BYTE OFFSET OF FIRST DATA RECORD = NNNNNNNN
14* UPPER LEFT CORNER X (0-1023) = NNNNNNNN
15* UPPER LEFT CORNER Y (0-1023) = NNNNNNNN
16* AVERAGING (1,2,4) = NNNNNNNN
where `NNNNNNNN' represents a right-justified integer value, `XXXX.XXXX' represents a right-justified real number, and `CCCCCCCC' represents a character string. The number of bytes per sample (field 5) must be 10 and fields 11 and 13 must point to the "old" header (described below) and the beginning of the data respectively.
The contents of fields 14 through 16 vary depending on which AIRSAR processor was used and may not exist at all. It was originally intended that they would be used to keep track of the relative location of the data set being examined with respect to the "original" data set.
The "old" header is composed of approximately 160 50-character fields and contains information recorded during the acquisition of the data and information added during correlation and possibly calibration. In versions of the SAR correlator prior to 1990, this is the only header saved with each data set (it is saved in the 1025th line of the file). With the addition of the variable format header, this header became the "old" header.
The user header does not usually exist on data obtained direct from JPL.
For information about polarimetry, see:
Zebker, H. and vanZyl, J., "Imaging Radar Polarimetry: A Review", Proceedings of the IEEE, Vol. 79, No. 11, November 1991.
Zebker, H., vanZyl, J., and Held D., "Imaging Radar Polarimetry From Wave Synthesis", Journal of Geophysical Research, Vol. 92, No. B1, pp 683-701, January 10, 1987.
Other sources of information on polarimetry are the following:
Born. M. and E. Wolf, Principles of Optics, Sixth Edition, pp 24-32, Pergamon Press, New York, 1980.
Freeman, A., SAR Calibration: An Overview, IEEE Trans. on Geoscience and Remote Sensing, Vol. 30, No. 6, pp. 1107-1121, November 1992.
Giuli, D., Polarization Diversity in Radars, Proceedings of the IEEE, Vol. 74, No. 2, pp 245-269, 1986.
Huynen, J. R., Phenomenological Theory of Radar Targets, Ph. D. Dissertation, Drukkerij Bronder-Offset, N. V. Rotterdam, 1970.
Van de Hulst, H. C., Light Scattering by Small Particles, pp 28-59, Dover, New York, 1981.
VanZyl, J. et. al., Imaging Radar Polarization Signatures: Theory and Observation, Radio Science, Vol. 22, No. 4, pp 529-543, July/August 1987.
Zebker, H., et. al., "Calibrated Imaging Radar Polarimetry: Technique, Examples, and Applications, IEEE Transactions on Geoscience and Remote Sensing, Vol. 29, No. 6, November 1991.