United
Nations
Step-by-Step: Flood Mapping with Sentinel-1 Interferometric Coherence
The Step-by-Step Explanation is also available as a PDF, which can be downloaded.
1. Downloading the Scenes from the Alaska Satellite Facility
The Alaska Satellite Facility (ASF) downlinks, processes, archives, and distributes remote-sensing data to scientific users worldwide. It is a convenient platform to download Sentinel-1 SLC data:
Each scene has 4 to 5 GB. Based on your AOI, one orbit should be selected, in which all scenes are located. Three S-1 Scenes are necessary: two pre-event scenes and one post-event scene. The scenes need to be of file type L1 Single Look Complex (SLC).
Drawing the AOI:
After loading the page, you can directly start drawing your AOI. Then use the “Filters” option to select the date range. Under “Additional Filters” select “L1 Single Look Complex (SLC)” as file type. If you already know the path, specify it in the section “Path and Frame Filters”.
How to use the Filters:
Select Start Date and End Date: they should be apart by at least one month, end date should be shortly after the disaster happened.
File Type: select L1 Single Look Complex
Path and Frame filters can be specified later
Click „Apply“
After clicking on apply, you will see the scenes (in blue) covering your AOI (yellow/orange). By hovering above the unique rectangles, you will see to what extent they cover your AOI (red). Decide for one rectangle and click on it. In the scene list, you will see it highlighted. You will also see the Path and Frame of the Scene. Use these numbers to go again into “Filters” and add them to path start/end and frame start/end. After clicking again on SEARCH, you will only see one rectangle left on the map.
All three scenes need to be in the same rectangle, i.e., they need to have the same path and frame!
Now, with the shopping cart icon you can add the three scenes to the download section.
In the download tab, you can start the download of the scenes, by clicking on the cloud icon. This will take some time as each scene has between 3 to 5 GB.
So please make also sure that there is enough disk space available on your computer.
2. Processing the Scenes with a pre-defined workflow in SNAP
Drag and Drop the three Scenes (zipped) into the Product Explorer Window in SNAP.
Click on the Graph Builder tool in the tool bar. On the bottom of the Graph Builder Window, click on „Load“. Browse to the directory, where you stored the Coherence.xml file. Double click to open it.
This is how the loaded Graph should look like. Essentially, this is a processing pipeline, for the three input SLC scenes.
The individual processing steps in the blue boxes can be also found in the normal task bar.
A few things need to be selected manually in the graph window:
For the three Read tabs, the Sentinel-1 scenes need to be selected. This has to be in the right order:
Read = Pre-Event 1
Read(2) = Post-Event
Read(3) = Pre-Event 2TOPSAR-Split Tab, select the Subswath and the bursts. The positioning of the subsets can be seen in the map on the bottom of the window. Apply the same selection to all the TOPSAR-Split tabs (2) and (3).
TOPSAR Split (In Detail)
https://step.esa.int/docs/tutorials/S1TBX%20TOPSAR%20Interferometry%20with%20Sentinel-1%20Tutorial_v2.pdf
Sentinel-1 scenes, that can be used for interferometry, are captured in three sub-swaths (IW1, IW2, IW3), using Terrain Observation with Progressive Scans SAR (TOPSAR). Each sub-swath image consits of a series of bursts, where each burst is processed as a separate image.
That is why we first split the scene, and later deburst it, i.e. put it back together into one image with the TOPSAR Deburst tool.
A few things need to be selected manually in the graph window:
In the Terrain-Correction Tab, you can control the output layers, which should be named:
coh_IWx_VV_pre-Event2_pre-Event1
coh_IWx_VV_pre-Event2_post-Event
If you also selected the VH polarization, two more layers will appear with VH instead of VV in their names.
If the dates are switched, go back to the Read Tabs, and control the selection order described in step 4.
A few things need to be selected manually in the graph window:
In the Write Tab, select the destination directory, to write the output to.
Click on Run.
The execution and the calculation of the coherence will take some time. Depending on your computation capacity, it can take up to a few hours.
If the graph does not work, you can also execute the steps individually in SNAP. You will need more storage capacity on your machine for that, as interim products need to be stored after each step.
Apply Orbit File (10 min)
S-1 TOPS Split (select only VV polarization, up to 3 bursts in one IW swath)
Coregistration: S-1 Back Geocoding (DEM: COP-30; uncheck “Mask out areas with no elevation”)
Coherence Estimation (square pixel size?)
Deburst
Speckle Filtering (median filter?)
Terrain Correction (DEM: COP-30, Map projection: Auto UTM, include layover shadow mask as output band)
3. Exporting the Coherence GeoTIFF
After the processing has finished you can close the graph builder window. In the product explorer window on the left, you will find a new product [4], called
S1A_IW_SLC__1SDV_xxxx_xxxx_xxxx_xxxx_xxxx_Orb_Stack_Coh_Deb_ML_TC
Click on the product, then go to File on the Task bar and select Export. Select GeoTIFF. Choose a directory for your export and click Export Product.
4. Importing the Coherence raster into Google Earth Engine
Open the following Google Earth Engine Script:
In the left panel, go to the Assets Tab, click on the red NEW button, under Image Upload select “GeoTIFF (.tif,.tiff) or TFRecord (.tfrecord + .json).”
Select the exported tif from the previous step, by clicking on SELECT and give a name to the file under Asset Name. Click on UPLOAD.
Under tasks, you will see the upload progressing. Afterwards click on “Run.”
Under the Assets tab on the left side, you will see your raster under CLOUD ASSETS. If it does not show up, click on the Refresh Icon.
By hovering above the file, you will see an arrow, by clicking on which, you can import the raster into the script.
