Interferometry Process Descriptions
by Matt Patrick



MSP Processing: The Modular SAR Processor creates a Single Look Complex image (SLC) from the raw data.

ERS_fix: This process looks at the raw data for missing lines. When it detects one, it replaces the missing line with the previous line. On the command line, you can choose to have cross correlation on or off. For us, we turned it on the first time and it seemed to detect missing lines which didn¼t exist. The result was a mass replacement and the rest of the processing didn¼t work. We then turned cross correlation off and the processing went fine.

ERS_PROC: This program contains all the MSP processing programs together: it is the main MSP processor. Consult the MSP user guide for a description of each program. Some of these programs are turned off (by having a # on their line). You can turn them on by erasing the #.

par_MSP: This creates the SLC parameter file. The parameter file contains information on the day and time of the scenes, the size and orientation of the scene, as well as the orbital location of the satellite.


ISP Processing The Interferometric SAR Processor creates an interferogram from the two SLC's.

init_offset: This makes an initial estimate of the offset between the two scenes. It takes a small window in the middle of each scene and compares them. The offset estimate is given as one of the last output lines, and should not exceed 10 for the range. The window coordinates can be changed in the command line.

create_offset: This creates the interferometry parameter file. It is similar to the SLC parameter files because it contains information on the size, orientation, and offset of the two scenes.

offset_pwr: This makes a precision estimate of the offset between the two scenes. It determines offsets to sub-pixel accuracy by fitting polynomials to the offsets.

offset_fit: This determines the offset polynomials based on a least squares error method.

SLC_interp: This resamples the second SLC to the geometry of the first SLC.

SLC_intf: This process makes the interferogram.

multi_look: This makes a multi-look intensity image of both scenes.

base_orbit: This process uses the satellite¼s state vectors to compute the baseline. The baseline is needed to subtract the flat earth phase from the interferogram. Flat earth phase is due to the geometry of the satellites. That is, fringes are produced simply as a result of the baseline. Base_orbit calculates the baseline from the satellite¼s on board instruments. The result of the base_orbit calculations are used in ph_slope_base, which subtracts these baseline effects from the interferogram. Unfortunately, the satellite¼s instruments are often incorrect, and ph_slope_base will not have subtracted the correct amount of flat earth phase. The alternative is to use the interferogram to calculate the baseline. By measuring the fringe rate, base_est_fft can calculate the baseline.

base_est_fft: This calculates baseline from the interferogram, by measuring fringe rate.

ph_slope_base: This process subtracts the flat earth phase, using the baseline. It reads the initial baseline (TCN).

cc_wave: This maps the correlation of each spot across the interferogram. The color range is from blue to purple to yellow. Blue being lowest, purple medium, and yellow highest.


Unwrapping Phase unwrapping takes the phase measurements, which are in units of 2pi with no reference, and sums them up from the start to determine the cumulative phase change across the scene. Areas of low coherence, however, make unwrapping impossible, so these areas are marked (or "masked") and skipped temporarily while unwrapping takes place. These areas of low coherence are then unwrapped using bridges, which are not necessary for this interferogram.

adapt_filt: This smooths out the fringes of the interferogram. This reduces the liklihood of "hanging" firnges; that is, those fringes which aren't continuous throughout the interferogram.

corr_flag: Those areas of low correlation would screw up unwrapping, so these areas must be identified and „masked¾ during initial unwrapping. The unwrapping skips these areas. Later, these areas can be unwrapped using bridges. The correlation threshold defines what is low correlation. Anything below the threshold will be masked. Therefore, try a threshold and run the program. One of the last output lines is fraction LSNR points, which shows how much of the scene is below that threshold. If you get less than 10-15%, then you should be alright. If above 20%, then try a lower threshold. Obviously, don¼t have the LSNR too low, either. That would mean that areas which should be masked may not be, and unwrapping could crash.

neutron:

residue:

tree_cc: This connects residues through neutral trees.

grasses: This is the actual unwrapping of the phase. Unwrapping takes the phase measurements, which are in units of 2pi with no reference, and adds them up from the start to determine the cumulative phase change across the scene.

gcp_ras: This is the interactive display of the .mli scene in which you choose ground control points and input their exact elevation.

base_ls: This computes a precision baseline using the ground control points. The new data is writtent to the precision baseline line of the baseline file.


Making the Height Map Making the height map involves two steps. First, the interferogram is tilted to fit the surface defined by the ground control points. Second, the scene is converted to orthonormal coordinates.

hgt_map: This makes the height map. It uses the precision baseline in the baseline file. It tilts the interferogram (no warping) to fit the real surface, which is estimated by using the GCP¼s.

res_map: This converts the height map to orthonormal coordinates. The areas on the edges of the scene will be altered due to their angle and distance from the satellite. This step will in a sense warp the height map to represent the real surface.