Gamma MSP: Reference Manual

az_proc

ANSI-C programs: az_proc.c

NAME
az_proc - SAR Range/Doppler azimuth compression processor with along-track Doppler centroid update

SYNOPSIS
az_proc <SAR_par> <PROC_par> <rc_data> <SLC> [az_patch] [SLC_format] [cal_fact] [SLC_type] [kaiser] [npatch]

<SAR_par>	SAR sensor parameter file
<PROC_par>	MSP Processing parameter file
<rc_data>	input range compressed data file
<SLC>	output single look complex image
[az_patch]	along-track azimuth patch size (range lines): (2**N, 2048, 4096, 8192...) NOTE: See Examples in User's Guide for the specific sensor
[SLC_format]	SLC output format: 0: FCOMPLEX, pairs of 4-byte float values 1: SCOMPLEX, pairs of 2-byte short integers NOTE: default in MSP processing parameter file; if not given FCOMPLEX will be used
[cal_fact]	processor radiometric calibration factor (dB), (enter - for default, from PROC_par) NOTE: absolute calibration factors (dB) are listed in sensor_cal_MSP.dat add 60.0 dB to the value from sensor_cal_MSP.dat if SLC_format is SCOMPLEX
[SLC_type]	output SLC data type: 0: sigma0 = (SQR(re) + SQR(im) (default) 1: gamma0 = (SQR(re) + SQR(im)/cos(inc)
[kaiser]	Kaiser window parameter for azimuth reference function weighting (enter - for default: 2.120)
[npatch]	number of along-track patches to process (default: sufficent for entire file)

EXAMPLE
az_proc ERS2_ESA.par p19960421.slc.par 19960421.rc 19960421.cslc 2048 0 -2.8 0
Generates radiometrically calibrated SLC in FCOMPLEX format (pairs of 4-byte-float). SQR(re)+SQR(im) corresponds to the backscattering coefficient sigma0.

az_proc ERS2_ESA.par p19960421.slc.par 19960421.rc 19960421.slc 2048 0 -2.8 1
Generates radiometrically calibrated SLC in FCOMPLEX format (pairs of 4-byte-float). SQR(re)+SQR(im) corresponds to the backscattering coefficient gamma0 (gamma0=sigma0/cos(inc_ang)).

az_proc ERS2_ESA.par p19960421.slc.par 19960421.rc 19960421.slc 2048 1 57.2 0
Generates radiometrically calibrated SLC in SCOMPLEX format (pairs of 2-byte-short-integer). 1.e06*(SQR(re)+SQR(im)) corresponds to the backscattering coefficient sigma0.

DESCRIPTION
az_proc reads the MSP SAR sensor parameter file, MSP SAR processing parameter file, and the SAR range compressed data. It performs azimuth compression using the range/doppler algorithm. Range migration is performed using sinc interpolator optimized for minimum error. The azimuth reference function is calculated from the along-track velocity and slant range distance. Output is in the form of complex floating point numbers (8 bytes per pixel) or complex short integer (4 bytes per pixel). The data is optionally deskewed and phase corrected to zero-Doppler geometry depending if the azimuth_deskew parameter in the MSP processing parameter file is ON or OFF. Radiometric compensation and calibration is applied. Corrections are performed for the slant range, antenna pattern, and receiver gain.

The SAR sensor parameter File contains information on the SAR sensor system parameters and RAW data format. It also contains the SAR look angle at center swath and the name of the file containing the antenna diagram. The antenna diagram contains the 1-way antenna gain pattern. Antenna gain patterns for the current spaceborne sensors are found in the MSP/sensors/ directory. If the specific antenna diagram is not available a constant antenna gain (file MSP/sensors/constant_antenna.gain) can be used. The antenna gain pattern file shall be copied to the working directory.

The two main functions of the MSP processing parameter File are the specification and documentation of SAR processing parameters. Certain processing parameters were extracted from the CEOS leader file provided with the SAR RAW data. Others were directly estimated from the SAR RAW data. Parameters specifying the image section to process, are defined by the user.

For the output SLC the MSP processing parameter file documents the processing done as well as the image geometry and radiometry.
az_proc has the added capability to track the azimuth Doppler centroid variations when a 2D Doppler polynomial is stored in the processing parameter file. This 2D Doppler polynomial is generated using the program doppler_2d and is most useful for multi-frame Radarsat or JERS-1 data sets.

The ouput SLC is either in FCOMPLEX (pairs of float) or SCOMPLEX (pairs of short integer) format, as specified in the slc_format parameter on the command line (0:FCOMPLEX, 1:SCOMPLEX). The advantage of the SCOMPLEX format is the smaller (half) size of the output file. Nevertheless, care has to be taken that the real and imaginary parts of the output values are in the short integers range. The values can be scaled with the calibration factor cal_fact. Assuming absolute radiometric calibration and a scale factor of 1.0e06 (i.e. 1'000'000) results in a maximum backscattering coefficient of +30.3 dB.

Continuous patch processing of the range compressed data is implemented. The range compressed data are processed in patches with FFT sizes of N*1024 (2048, 4096, 6144, 8192, 10240, 12240, 16384 ...) lines/block. The size of the block is optimally 2 times the length of the azimuth reference function. For ERS/ENVISAT this is 4096 samples.

After the processing of each patch the result is written out to the output SLC. The number of lines written out is the patch size reduced by the length of the azimuth reference function and some additional lines to account for the image skew. To optimize the efficiency of the processing large patches are recommended, but the entire data for one patch should be in the computer memory without using the much slower swap space. This way multiple frames of ERS data with over 150000 lines have been processed to full resolution.

Radiometric Calibration
The az_proc supports the radiometric calibration of the processed data. The output data is calibrated for the:
- range spreading loss
- antenna gain pattern
- ground-range projection factor

The radiometric calibration factors are stored in a text format file (sensor_cal_MSP.dat) that can be found in the MSP/sensors/ directory. It can be examined with a text editor.

Calibration experiments were conducted to determine absolute calibration factors for the processing of ERS1, ERS2, and JERS. In the case of ERS1 and ERS2 SAR RAW data over the Flevoland calibration site was used to determine the calibration factor based on active transponders with known radar cross sections. The MSP ERS1 and ERS2 calibration factors were validated by comparison of the MSP processed data with ERS PAF processed SLC data which were calibrated following the methodology described by H. Laur et al., ERS-1 SAR Radiometric Calibration, ESA-ESRIN, Frascati, Italy. The offset between MSP and ESA PAF processing was below 0.5 dB for both ERS-1 and ERS-2. The often almost identical values observed in ERS1 / ERS2 Tandem pairs further confirms the very good ERS calibration accuracy.

Similarly, JERS RAW data over the Japanese Calibration site was used to determine the calibration factor based on active transponders with known radar cross sections. The MSP JERS calibration factor was validated by comparison of MSP processed data with a NASDA processed PRI product over a tropical forest site. For the validation example the NASDA processed values are 0.77 dB lower than the MSP ones.

Values for ENVISAT ASAR are currently available for the main mode, i.e. IS2 VV-polarization. The calibration constant was determined by comparing the MSP processed data with SLC data from ESA and calibrated with the GAMMA software. Several ROIs were selected and intensity values were compared, showing a constant offset.

Similarly the calibration constants for a number of ALOS PALSAR modes were determined by comparing the MSP processed data with SLC data from JAXA EORC (calibrated). Several ROIs were selected and intensity values were compared. Typically the difference showed a constant offset which was set to correspond to GAMMA's MSP calibration constant. For more details it is referred to the file sensor_cal_MSP.dat.

The calibration factors in sensor_cal_MSP.dat are for float value SLC data. If the data are in SCOMPLEX format, this scale factor must be augmented by 60 dB greater than the value in sensor_cal_MSP.dat . Currently we are determining the absolution radiometric calibration constants for Radarsat-1.

Supported SLC output types are sigma0 and gamma0. In the case of sigma0 the SLC image intensity (SQR(re) + SQR(im)) corresponds to the backscattering coefficient sigma0 (normalized to the horizontal ground surface). In the case of gamma0 the SLC image intensity (SQR(re) + SQR(im)) corresponds to the backscattering coefficient gamma0. gamma0 and sigma0 are related by gamma0 = sigma0/cos(inc_ang) with inc_ang the incidence angle of the horizontal surface.

As default Kaiser window parameter beta for the azimuth reference function 2.120 is used. A lower value such as 1.0 leads to somewhat sharper point targets (such as calibration targets), nevertheless at the price of a higher stopband ripple. The radiometric calibration is not affected by the selected kaiser window parameter beta.

The resulting complex images in FCOMPLEX format can be displayed using disSLC. The resulting complex images in SCOMPLEX format can be displayed using the ISP program disERS_SLC. Notice that the zero byte line header size needs to be indicated when using disERS_SLC. Multi-looking of the complex output image in FCOMPLEX format can be done with multi_SLC. Multi-looking of the complex output image in SCOMPLEX format can be done with the ISP program multi_look_ESA after creation of the ISP SLC parameter file using the ISP program par_MSP.