University of Alaska Fairbanks
Department of Geosciences

GEOS F493 / F693 - Geodetic Methods

Lab 6: Processing w/ gmtSAR

Note that you DO have to work on LuNGS today!

Introduction

Complete shift in gears now! We're moving from GPS processing into InSAR processing (for a bit, GPS processing will come back later). The goal for today is to get you comfortable finding you way around gmtSAR. We'll do data access next week.

A note of warning: there's a lot of vocabulary flying (ha!) around in InSAR processing. Most of these are standard satellite related terms. If there's something you don't understand, it's usually a good idea to search for the term, the satellite, and read up on it.

1. gmtSAR - Intro

gmtSAR is an open source InSAR processing system. It makes use of the open source Generic Mapping Tools (GMT) (a tool very worthwhile learning in its own right!! Albeit with a bit of a learning curve; check out their tutorial and cookbook).

You can guess that ease of use is one of the main reasons I decided to use this tool in the class; other InSAR processors exist.

Read through sections 1.1 and 1.2 (no need to get into 1.2.1 - proper focus, right now) of the gmtSAR documentation and answer the following questions (in well-formed full sentences in the English language):

• What is the ultimate limitation of gmtSAR?
• Which satellites cannot be easily processed?
• The phase difference between two images is impacted by which factors/processes?

2. gmtSAR on the command line

Log into LuNGS (via badger) (please use ssh -XY to enable graphics forwarding!!). Please spread out across the machines a bit. First some preparation. Copy this file into your $HOME directory on LuNGS and make sure it's named .gmtversions. It's probably best if you go to your $HOME directory and download therE:

	$>cd $HOME
	$>wget http://www.grapenthin.org/teaching/geodesy/download/.gmtversions
	

Then add the following to your .bashrc file:

	export GMTSAR=/usr/local/GMTSAR
	export PATH=$HOME/this_gmt/bin:$GMTSAR/bin:"$PATH"		
	

Now run and pick option 2:

	$> gmtswitch
	1. Select /usr/local/gmt_all/gmt-4.5.3
	2. Select /usr/local/gmt_all/gmt-5.4.4

	Please choose a GMT version (1-2) [1]: 2
	

When you ask for the GMT version on the commandline you should get 5.4.4:

	$> gmt --version
	5.4.4 
	

Now go into your geodesy directory and run:

    $> gmtsar.csh
	

If the command is not found, log out and back in again and run the command again. This lists all the shell scripts (recipes, in a sense) that are currently available with GMT5SAR. Note that for most of these scripts you will get a hint on how to use them if you just type their name on the command line and hit enter (just like you did with gmtsar.csh above. Here are some questions:

• For which sensors/satellites does gmtSAR implement end-to-end processing?
• Execute one of these end-to-end scripts on the command line. How would you call it to process data? In which directory would you store the SAR data? In which directory would you store your DEM? How do you have to name the DEM file? Name the sensor in your answer!
• There are 2 ways to make or get a DEM. One works on the command line, one works on the gmtsar website. Explain both! (read the manual, check the website, the command line programs etc. to find them)

3. Processing some test data

Create a directory ~/geodesy/lab07 and place a gmtSAR test data set from /gps/InSAR_test/ into it.

Refer to the gmtSAR documentation, section 2.1, to understand what all the files are that gmtSAR produces during the processing. In fact, many things that youmay not understand are likely explained somewhere in there. Feel free to get well acquainted with this document.

After changing into the lab07 directory, copy and unpack the data:

	$> cd ~/geodesy/lab07
	$> cp /gps/InSAR_test/YOURDATASET.tar.gz ./
    $> tar xfz *tar.gz
	

List the directory contents (ls) and copy the output into your answer document.

You should have some form of README document. Read it! And follow the instructions that are in there. For some data sets you may need to create files with different names; it should all be spelled out in there. Do it! Check that the raw directory has the files listed in the README file.

The README should also list a command to be executed on the command line. It should start with p2p_ for process 2-pass followed by the sensor name. The next two arguments are Master and Slave scene and the last argument is a parameter file that controls gmtSAR's processing.

Open the config file config.* (the rest of the name depends on the sensor). You will find sections for configuration. If you don't have any SLC (single look complex) images in your raw directory, make sure to set proc_stage=1 otherwise proc_stage=2 is OK. Note that this is only important for the first run. If you decide to change parameters later to improve the processing (say the phase unwrapping, you can skip all the prior stages by setting this parameter to the processing step you want to play with!)

If a parameter doesn't have a value, gmtSAR uses a default. Leave that for now. Make sure the topo_phase is being subtracted from the the phase! Obviously, that's when you need your DEM! Make sure it is in ./topo/. Leave all the other parameters the way they are, but read through the file!

• What's snaphu? (you may want to run snaphu.csh on the command line if you can't figure it out from the config file.)

Now you're ready to run the processing - get the command line from the README file and execute it! This may take a while! (10s of minutes)

Once it's done, check which new directories were made (ls). Compare to your previous listing Copy into your answer file!

Your lab07 directory now should contain a intf directory. In there is another directory which notes which SAR scenes were interfered. Change into that directory and list its contents!

WHOA! There's a lot going on! Copy the output into your answer file!

Here are some viewers for the various files:

• .png: eog
• .ps: evince
• .kml: google-earth (though currently not running on LuNGS!)

Remember to refer to the gmtSAR documentation, section 2.1, to understand what all the files are that gmtSAR produces during the processing.

Note that the (mask)_ll files are projected into geographic coordinates.

• Investigate the phase and (if they exist) unwrap results. Is there any deformation in your interferogram? What might be the cause?
• What other processes could cause fringes in the interferogram / apparent deformation in the unwrapped data?
• Investigate the corr file. Does your interferogram have good correlation everywhere? How does correlation compare to information in the interferogram (phase/unwrap)?
• Note that there are amplitude images; look at them and see if you recognize anything.

4. Play with the processing

Repeat the following twice! Once with a very low correlation threshold for the phase unwrapping in snaphu, another time with a very high correlation threshold.

Save the old result directory to preserve your previous run (run1, run2...):

    $>cp -r intf intf_run1
	

Edit the config.* file to change the snaphu parameter, also set the proc_stage parameter to a value that starts from phase unwrapping. Rerun your p2p_.... Repeat the above for another snaphu parameter.

Copy the 3 unwrapped interferograms in the answer file and comment on the impact your changes had on the results.

Deliverables: (submit via blackboard!)

• an answer file (PDF) with all answers to the questions above.