2016 Kumamoto Earthquakes: DInSAR Surface Deformation Mapping
- Jedidiah Chibinga
- Dec 23, 2025
- 9 min read
Updated: Jan 2
Between 14 and 16 April 2016, the Kumamoto region of Kyushu Island in Japan was struck by two powerful earthquakes, measuring 6.2 and 7.0 magnitude, accompanied by several smaller earthquakes. It was the deadliest to strike the Japanese Island since the 2011 Tōhoku earthquake, and until the 2024 Noto earthquake.
To give more context on the magnitude and widespread damage of the earthquake (according to this study):
Approximately 183,882 people had to evacuate as a result of the damage to about 194,888 buildings, 42,786 of which partially and totally collapsed.
The death toll was 273, and 2809 people were injured. Many of the injured required immediate healthcare, which could not be immediately provided as personnel in some hospitals had to be evacuated, causing "severe structural and utility lifeline shortages".
The main shock also caused a power shortage to 476,600 customers.
DInSAR in Surface Deformation Mapping
To map and visualise the surface deformation caused by the earthquake, I carried out a Differential Interferometric analysis using Sentinel-1 SAR data. DInSAR uses radar signals to capture changes in the Earth’s surface by comparing two or more radar images taken at different times.
This technique is useful for detecting small ground movements with millimetre precision. By analysing the phase difference between radar images, DInSAR can detect displacements caused by events such as earthquakes, volcanoes, and landslides.
Visualising the Displacement Map
Before generating the final displacement map, we get a wrapped differential interferogram, which represents the differences between the phases of the images before and after the earthquake.
With this interferogram, we can see the fringes which depict the disturbance caused by the movement of the ground after the earthquake.
To obtain the displacement map, the phase values are "unwrapped" to obtain a more continuous spread of the displacement from the approximated fault line. This makes it easier to obtain more accurate displacement values.
The map shows how the ground shifted during the 2016 Kumamoto earthquake, with different colours revealing areas that moved in opposite directions along the fault, highlighting the scale of the event.
The displacement map can also be visualised here.
Interpretation of the displacement map
The wrapped differential interferogram and the displacement map show how the ground moved during the earthquake.
In the differential interferogram:
The intervals between the successive same colour fringes represent an approximately equal displacement from the approximated fault line, with more fringes close together showing more displacement.
In this interferogram, the values also range from π to −π. In the interferogram, we can also see fringes in a blue-yellow-red and red-yellow-blue pattern.
The former pattern (blue-yellow-red) shows areas where the ground may have moved increasingly towards the satellite, while the latter (red-yellow-blue) may show movement in the opposite direction.
This is visualised more clearly in the displacement map.
In the displacement map:
We can see the movement of the ground relative to and from the satellites' Line of Sight (LOS) along the fault line (which was approximated by drawing a line between the areas where the displacement direction changed steeply).
The negative phase values visualised in dark blue depict movement away from the satellite, while the positive phase values in dark red show movement towards the satellite.
We can also see the magnitude of the movement or displacement along the earthquake's fault line.
In the direction away from the satellite, the ground moved by as much as 1 meter. Towards the satellite, the ground movement was more than half a meter, with displacement reaching 0.6 meters (60 centimetres).
With movements of 1 and 0.6 meters in opposite directions, the maximum total displacement was roughly 1.6 meters.
Together, the map reveals how the Earth’s surface shifted when the fault suddenly slipped.
Relevance of displacement mapping
Displacement maps are a crucial tool for understanding how the ground moved. They help:
Identify which faults ruptured,
Estimate the strength and extent of the earthquake, and
Assess areas that may be more vulnerable to future seismic activity.
This type of satellite analysis is especially valuable in regions where ground sensors are sparse or damaged after disasters.
The process used to acquire the displacement map is shown in the steps below.
Workflow
Identifying the Area of Interest
I used the USGS Earthquake Catalog to determine the earthquake's location. The catalogue provides information on selected recent or historically significant earthquakes.
This was done by filtering for the date around the time the earthquake occurred. In this case, I filtered for earthquakes around the world with a magnitude of more than 2.5 and between 2023-01-28 00:00:00 and 2023-02-10 00:00:00.
After filtering, I selected the earthquake location with the largest magnitude, copied the coordinates and used them in the next step to download the SAR dataset.
Downloading SAR Data
The SAR datasets were downloaded from the ASF Data Search. Sentinel-1 was selected as the dataset, and filtered by the date and location obtained from the USGS Earthquake Catalog.
Two datasets with the same footprint from about a month before and after the earthquake date were selected. The two datasets selected were Single Look Complex (SLC) images acquired using the Interferometric Wide Swath(IW) beam mode and with a Vertical-Vertical (VV) polarisation.
Extracting the Interferogram
The interferogram was extracted using Sentinel Application Platform (SNAP), an open-source tool developed by ESA. This was done using the steps outlined below, sourced from Practical: Generation of a Displacement Map by Dr Karima Hadj-Rabah. More detailed explanations for the reasons for each of the steps used, and what the different parameters do, can be found on the SNAP Online Help website.
Data Preparation
This (preprocessing) step ensures that the two images are aligned and ready for preprocessing. To do so:
Import the Sentinel-1 images into SNAP.
Either go to File → Open Product → Select the two images → Open
Or drag both images to the Product Explorer panel in SNAP.
Importing the two datasets.
Split the subswath to get the subswath and bursts encompassing the earthquake's region.
Go to Radar → Sentinel-1 TOPS → S-1 TOPS Split.
Select the first image. Specify the file path and rename the output file if needed.
Specifying the dataset for TOPS splitting and the output directory, Select the sub-swaths and bursts that contain the earthquake's region.
Selecting the subswath, polarisation, and bursts. In this case, the subswath IW2 and all 9 bursts contained the area of interest. VV polarisation was selected because*****.
Repeat the step for the other image.
Apply the orbit file to the 'split' images to correct for satellite positioning errors due to the slight movements of the satellite as it moves along its orbit.
Go to Radar → Apply-Orbit-File. Keep the default settings. Select Run.
Repeat the step for the other image.
Topographic Interferogram Generation
This is a key step to extracting the displacement caused by the earthquake. The topographic interferogram is generated to highlight phase differences caused by the elevation of the terrain. This can be done by:
Coregistration (Back-Geocoding), a mandatory step to align the images:
Go to Radar → Coregistration → S-1 TOPS Coregistration → S-1 Back-Geocoding.
Select 'Add opened' to add both images. The first image will be the 'master' and the second the 'slave'.
Configure the parameters in the Back-Geocoding tab. Select Run.
Selecting the Digital Elevation Model (DEM).
Enhance Spectral Diversity (optional step) to apply range and azimuth shift corrections.
Go to Radar → Coregistration → S-1 TOPS Coregistration → Enhanced Spectral Diversity. Keep the default settings. Select Run.
This step was not applied during this exercise.
Interferogram generation to compute the interferogram by comparing the phases of the slave and master images.
Go to Radar → Interferometric → Products → Interferogram Formation.
Select the coregistered (or spectrally enhanced) stack as input and alter the Processing Parameters if needed.
In this case, the Coherence Range Window Size was set to 20.
Intensity (top), phase (middle) and coherence (bottom) after interferogram generation.
TOPSAR Deburst to remove the burst boundaries (horizontal lines) in the image:
Go to Radar → Sentinel-1 TOPS → S-1 TOPS-Deburst. Select Run.
Phase (top), and coherence (bottom) after Deburst, and before removing the topographic phase.
Generation of Differential Interferogram
This step involves removing the topographic effect or extracting the phase changes caused by ground displacement by generating a differential interferogram. This is achieved through:
Interferogram formation, which first carries out topographic phase removal by estimating and subtracting the topographic phase from the previous topographic interferogram:
Go to Radar → Interferometric → Products → Topographic Phase Removal.
Input the topographic interferogram from the previous step.
Selecting the DEM. In this case, SRTM 1Sec HGT was specified as the Digital Elevation Model (DEM) for the topographic correction. We can also output the topographic phase band and the elevation band if desired. Select Run.
Intensity (top left), phase (top middle), coherence (top right), elevation (bottom left) and topo phase (bottom right) after interferogram formation.
Multilooking to improve the visual interpretability of the differential interferogram by reducing speckle noise and obtaining square pixels.
Go to Radar → SAR Utilities → Multilook.
Specify the number of looks for range and azimuth.
Specifying the number of range looks. In this case, 8 was selected for the Number of Range Looks.
Phase (left), and coherence (right) after applying multilooking. In the output, we now see that the images are more portrait-oriented, as expected.
Goldstein Phase Filtering to improve the phase quality, ensure accurate unwrapping and calculate displacement by reducing noise or residues.
Go to Radar → Interferometric → Filtering → Goldstein Phase Filtering.
Specifying the coherence threshold. Input the differential interferogram and configure the processing parameters. The coherence threshold was set to 0.8.
Phase (left), and coherence (right) after Goldstein phase filtering, and removing the topographic phase.
Generation of displacement map
Before generating the displacement map, the wrapped phase values need to be unwrapped to obtain continuous displacement measurements. To do so, carry out:
SNAPHU export. This step formats the data in a way that is compatible with SNAPHU processing and builds a SNAPHU configuration file (snaphu.conf), where processing parameters for SNAPHU are stored. To carry out this step:
Go to Radar → Interferometric → Unwrapping → SNAPHU Export.
Specify the target folder for the output.
Specifying the parameters for Snaphu export. Configure the export settings, as needed. In this case, I used the default Number of Tiles and Rows, and increased the number of processors to 10 (to match the number of processors on my computer).
Note: If you reduce the number of tile rows and columns, the system won't make use of the multicore processing, resulting in a longer execution of Snaphu and possibly different results (more details here).
Install the SNAPHU plugin.
Ensure that the SNAPHU plugin is installed. It can be downloaded from the ESA website.
Follow the Installation and usage instructions 1-3 under Manual SNAPHU installation to install it correctly.
Execute SNAPHU to unwrap the phase by using the exported files.
Open the snaphu.conf using Notepad.
Replace .snaphu.img with the name of the .img file that was generated in the Snaphu Export folder. In this case, we replace it with coh_IW1_VV_08Apr2016_20Apr2016.snaphu.img.
Comment out (put '#' in front of) the LOGFILE.
Copy the command to call snaphu, which in this project was snaphu snaphu -f snaphu.conf Phase_ifg_srd_VV_08Apr2016_20Apr2016.snaphu.img 3444
The .conf should look something like this.
Go to the Snaphu Export folder → Right mouse click → Open in Terminal → Paste the command to call snaphu that was copied earlier. Press Enter to run the command.
Import SNAPHU processing results to construct an interferogram that will contain the unwrapped phase band.
Go to Radar → Interferometric → Phase Unwrapping → SNAPHU Import.
In the 2-Read-Unwrapped_Phase tab, select the Unwrapped HDR File as shown in the image below.
Selecting the unwrapped phase. In the 4-Write tab, add _unwrapped to the name of the output file. Select Run.
Specifying the suffix and path for the unwrapped phase. The result should have an unwrapped phase as shown below.
Output of phase unwrapping.
Generation of a Displacement Map to convert the unwrapped phase to displacement, for visualising and analysing the ground movement.
Go to Radar → Interferometric → Products → Phase to Displacement.
Input the unwrapped phase product. Select Run.
Generated displacement map from the phase image.
Lastly, we apply Terrain correction to georeference the displacement map to a coordinate system and remove distortions caused by topography, ensuring accurate spatial analysis.
Go to Radar → Geometric → Terrain Correction → Range-Doppler Terrain Correction.
Input the displacement map.
In the Processing Parameters, select a DEM and configure the Map Projection. In this case, SRTM 1Sec HGT was selected as the DEM WGS84 as the projection.
Selecting the Digital Elevation Model. This is the output after the terrain correction.
Output of Range-Doppler Terrain Correction.
Visualising and Exporting the Interferogram
After generating the displacement map, we can now visualise it in different ways to understand what is in the image.
We can change the colour scheme under Color Manipulation, in the bottom left panel, by changing the Scheme. By selecting From Data under Range, we can stretch the Palette according to the displacement values in the image.
To export the displacement map, right-click on the image and export it in the desired format.
Or go to File → Export.
