Gamma MSP: Reference Manual

rc_real

ANSI-C program: rc_real.c

NAME
rc_real - Range compression of offset video (real) SAR raw data (SIR-C).

SYNOPSIS
rc_real <SAR_par> <PROC_par> <raw_data> <rc_data> [loff] [nl] [kaiser] [nr_ext] [fr_ext] [r_chirp] [rfi_flag] [rfi_thres]

EXAMPLE
rc_real SIRC.par p19951009.slc.par p19951009.raw 19951009.rc

DESCRIPTION
This program range compresses offset video SAR data, such as produced by SIR-C. A matched filter is calculated from the parameters stored in the Sensor Parameter file and applied using FFT convolution. The fully range compressed samples are stored on disk for autofocus and azimuth compression. The program also accepts a a text file containing the range chirp values rather than calculating a synthetic chirp. This feature can be useful if a non-chirp, or distorted chirp waveform is used for range compression. The format of the range chirp file consists of pairs of entries: sample number floating point chirp value, one pair per line, separated by spaces or a tab character (white space).

Input data are offset video rather than IQ complex format, therefore the data must be converted to IQ complex prior to azimuth filtering. A "quicklook" image can be created by first range compressing the data with rc_real followed by azimth filtering and decimating the result. The program prefilt performs this function on the range compressed data set.

rc_real has the capability to extend the processed image swath beyond the region of full resolution. The nominal number of samples in the range compressed data for offset video input data is:

(echo_samples_processed - range_chirp_samples)/2.

Outside of the nominal full resolution swath are two bands, one in the near range and the other in the far range. Both of these bands are as wide as the range chirp width in samples. Across these bands the resolution degrades linearly starting at the full resolution and going to zero. These regions also fade in intensity proportional to the loss in resolution due to the decreasing number of echo samples for these image pixels. The user can specify the swath extension for both the near and far range in the processing parameter file or via the command line parameters nr_ext and fr_ext.

Radio frequency interference (RFI) due to ground radars can be a problem with L-Band data collected by SIR-C. The interference will cause noisy regions to appear in the image, possibly completely obliterating the scene. rc_real has the option to apply an filter that adaptively suppresses RFI at the possible cost of higher range and azimuth sidelobes. The interfering signals are generaly narrow band and can be detected in the range spectra. An adaptive notch filter eliminates the affected spectral windows when the spectrum is greater than the expected value as specified by the rfi_thres command line parameter. Increasing rfi_thres decreases the amount of filtering while lower values of rfi_thres will increase the amount of rejected bandwidth.

The raw data is divided into blocks of 512 lines, Within each block the spectrum is evaluated for RFI and filtered. A plot of the spectrum and the notch filter is stored in the file range_spectra.dat for the first block. The program gnuplot can be applied to examine the spectrum and associated filter with the command:

To plot the spectrum with gnuplot on the screen:


echo "set xlabel 'FFT sample'; set ylabel 'Relative Power'; set title 'Power Spectrum'; plot 'range_spectra.dat' with lines lw .5 "|gnuplot

or plot to a png image file with gnuplot:

echo "set term png; set output 'range_spectra.png'; set xlabel 'FFT sample'; set ylabel 'Relative Power'; set title 'Power Spectrum'; plot 'range_spaectra.dat' with lines lw .5 "|gnuplot


SEE ALSO
Users Guide, typedef_MSP.h, Sensor parameter file, MSP Processing Parameter File.