Finding Earthquakes Using InSAR

When I first met with Gareth, I felt my brain slowly overloading. I took my notes and tried to process the material presented to me the best I could.

I will lay it out. Essentially, I am an earthquake “finder” which means I am helping Gareth locate as many earthquakes as we can worldwide. We are doing this through InSAR which stands for Interferometric Synthetic Aperature Radar (I’ll explain later). First, we find potential candidate earthquakes, then we download data from this NASA website called EarthData under data discovery Vertex ASF which is the Alaska Satellite Facility.

Here’s the link to the website (you must have an account to be able to download anything though!) and this is what it looks like:

I mentioned InSAR;

Now the data being downloaded is collected by the European Space Agency’s Sentinel-1 satellites. Essentially, the ESA describes it as, “..a constellation of two satellites orbiting 180° apart, the mission images the entire Earth every six days.  As well as transmitting data to a number of ground stations around the world for rapid dissemination, Sentinel-1 also carries a laser to transmit data to the geostationary European Data Relay System for continual data delivery.” I retrieved that bit of information from their website, found here:

That’s how I get the data I need to be able to process the interferograms we use to find target earthquakes.

The first step, is to find a good place to start the search.

After I first met with Gareth, my job was to compile a spreadsheet with possible candidates for earthquakes. I was to find earthquakes of magnitude 5.5 and above with depths below 20 kilometers (anything more is undetectable by the satellites).

He provided me with two websites: and

I use the latter. I customize my search and when I hit the search button, this is what I find:

During this stage, I need to make sure the earthquakes I search for are onshore. Earth is primarily water so the majority of earthquake activity occurs offshore. As you can see on this screenshot, there are only a few potential candidates here.

After this part of the search I keep a detailed record on this spreadsheet.

That is a snippet of what my spreadsheet looks like. I put the location of the earthquake, latitude and longitude, magnitude, depth, date and time. On the far left, I have an interferogram ran column and on the far right columns I have information about the frames and tracks I have downloaded from Vertex. I will show you that as well:

After I have my list of potential earthquakes, I download data on these earthquakes like I have already mentioned. I then match up the location from the USGS site with the map on Vertex and I search a time frame including before and after the earthquake date.

Now comes the more complicated part. I use what is called a terminal, and I take the data I download, organize it, and then I unzip the files. WIth the terminal, I had to learn various commands I need to use to navigate all of the directories on the computer in Gareth’s office. This is what the terminal looks like:

There is the terminal I had for an earthquake in Tajikistan that I was investigating. What you are seeing is the listing of what is in the swath 2 directory of the Tajikistan directory I created.

The terminal consists of directories. We use those directories to organize files and run programs such as the python which is what’s used to run our topsApp.xml files.

At first, this process was confusing and frustrating to me. Now, it’s definitely simpler to use (after understanding the directory structure). This / means root. The best analogy Gareth used to describe it to me was a tree. Pretty much, if you use / then you are going into the root directory. For instance, if I want to access the scratch directory, then I type cd /scratch. In the scratch, that’s where I can access the files we have for various earthquakes. The cd command stands for change directory, so if I type cd /data then I am accessing the files that are placed in data; I needed to use / to denote that I am going into another directory. When I want to see what is in the directory, I use ls which means list.

 I download the SLC files from Vertex into the data directory. SLC stands for Single Look Complex, and it is a file type which is processed for resolution. As I mentioned earlier, with my spreadsheet I have to make a guess as to which swath the earthquake is located in. I also record what is the path which is the track number (the orbit that the satellite takes) and the frame number (current location on the track) of the data I download from Vertex. I can either download ascending and descending data.

Astrid blog 1 img 6.jpg

Here is a quick illustration I made. On the right is ascending (swaths 1, 2, 3) and on the left is descending (swaths 3, 2, 1). 

Each of those strips is a swath. Swaths are divided up into bursts. The satellites have a beam that swings in different directions which results in the bursts (little pieces of data).

After I input the swath, track, and frame numbers for my spreadsheet, I am ready to run the interferogram and hopefully find some earthquakes.

Here are the commands I type into my terminal:

ls / data -sf (data file name)

cd / scratch/ (data file name)


cd swath (pick)


gedit topsApp.xml topsApp.xml --steps

cd merged /

ls filt_topophase.flat.geo&

That last command runs the interferogram. These steps are much more simplified thanks to a graduate student who had also been running interferograms.

Now I bet you are wondering, what do these interferograms look like?

Like this!

This is an interferogram I ran for an earthquake in Sary-Tash, Kyrgyzstan. The area where you see the color fringes is the earthquake. With the many interferograms that we have ran, the amount of earthquakes we have found is relatively low. Now the reason may be because we are looking in the wrong location, vegetation, and water vapor in the atmosphere creates noise or static as you can see in certain patches here.

When Gareth first showed me an interferogram, I compared it to a tie-dye shirt. Wouldn’t you agree?

Any way, now with the results of the interferogram, I have to take note whether the interferogram is masked (which means the earthquake is overwhelmed or washed out by aftershocks), decorrelated (meaning there is too much atmospheric noise), or coherent. Interferometric correlation means you get a signal, and coherence is when the neighboring pixels are moving the same way which results in lots of color.

The next step, after obtaining these interferograms, is to model the earthquakes.

Gareth has listed me as an author for an abstract he put together about finding earthquakes using InSAR. The title is “A Systematic Study of Global Earthquake Detectability” we are going to model the twenty or so earthquakes he listed in this abstract.

I’ll post more details about the modeling process soon!