Run Coherent Change Detection (CCD)
Lesson 1 of 1
Run Coherent Change Detection (CCD)
In this quick guide, you will:
- •
Learn how CCD is used to monitor subtle changes to features on the Earth's surface. * •
Open and display two overlapping Umbra Single Look Complex (SLC) images from different dates. * •
Create a CCD classification image that reveals possible construction activity in the area.
Sample Data
The exercises in this quick guide use two Umbra SLC images for demonstration. Download the ZIP file below and extract its contents to a directory on your computer.
[SAREssentials_CCD.zip
884.9 MB
DownloadArrow down with horizontal line beneath it](assets/SAREssentials_CCD.zip)
- •
File names: Umbra_SaghandIran_2023-08-17_slc and Umbra_SaghandIran_2023-08-25_slc. ENVI header files (.hdr) and ENVI SARscape files (.sml) are also included. * •
Acquisition dates: 17 August and 25 August, 2023 * •
Processing notes: The source datasets were Sensor Independent Complex Data (SICD) files in National Transmission Imagery Format (NITF), which require the ENVI NITF/NSIF Module to read. Since not all users have the ENVI NITF/NSIF Module, we used the "Imported data" option in the SAR Basic Data Processing tool to create SLC images that are in SARscape format instead of SICD. * •
Source: Umbra Open Data Program(opens in a new tab), Attribution 4.0 International (CC BY 4.0)(opens in a new tab) license. The source IDs are 2023-08-17-05-53-58_UMBRA-05_SICD and 2023-08-25-18-47-23_UMBRA-06_SICD.
Background
CCD uses the phase information in backscattered SAR signals to detect subtle changes to features on the Earth’s surface. Radar wavelengths range from about 3 to 70 cm, depending on the mission. The CCD process computes a variable called coherence. This characterizes the sensitivity to changes in the relative position of reflecting objects whose dimensions are on the order of the transmitted wavelength (in centimeters) within a resolution cell. In other words, coherence can detect small variations that are not detectable in the amplitude component of a SAR signal.
CCD can detect and identify subtle changes such as:
- •
Placement of hidden mines and improvised explosive devices (IEDs) * •
Recent movement from vehicle or foot traffic * •
Recent explosive detonations * •
Natural changes such as vegetation growth * •
Effects of rain and wind on the surface, which can often mask man-made activity
Applications of CCD include monitoring illegal construction, border security, threat assessment, and more. CCD is especially useful in rocky and arid areas, since dry soils provide more accurate coherence measurements. Coherence indicates how well the phase signals correlate between two SAR images from different dates.
Data Requirements
Images used for CCD must meet strict requirements. They must be SLC images with the same:
- •
Frequency * •
Polarization: HH, VV, VH, or HV * •
Orbital direction: ascending or descending * •
Acquisition mode: i.e., spotlight, stripmap, etc. * •
Incidence angle: less than one degree of difference
In addition, the baseline separation must be smaller than 60% of the critical baseline. This is an advanced concept that is beyond the scope of this quick guide.
Tip: To check if your images are suitable for CCD, run the SAR Compatibility Check tool in the SAR Essentials folder of the Toolbox. See the SAR Essentials: Check Data Compatibility quick guide for more information.
Open and Display Umbra SLC Images
- 1
Start ENVI. - 2
Select File > Open from the ENVI Menu bar. An Open dialog appears. - 3
Go to the location where you saved the sample data for this quick guide. - 4
Use the Ctrl key to multi-select the following files, then click Open:
Umbra_SaghandIran_2023-08-17_slc
Umbra_SaghandIran_2023-08-25_slc
- 5
Wait for the raster pyramids to build in the Status bar, then click the Zoom to Full Extent button in the Toolbar. Both images are displayed at their full extent.
The village of Saghand is located near the center of each image. Uranium ore is extracted at the nearby Saghand mine, so any new development in this region is closely monitored. There were reports of construction activity in this area between 17 and 25 August, 2023. For this exercise, you will create a CCD image to look for evidence of this.
The Umbra images have an incidence angle difference of only 0.12 degrees and an azimuth angle difference of 0.15 degrees, so they are suitable for CCD analysis.
Run the SAR Coherent Change Detection Tool
- 1
Go to the Toolbox and expand the SAR Essentials > Change Detection folder. - 2
Double-click SAR Coherent Change Detection. The SAR Coherent Change Detection tool appears. The Input tab is active.
Select Input Files
- 1
Click the Browse button next to Before Image. A file selection dialog appears. - 2
Select the file Umbra_SaghandIran_2023-08-17_slc.sml and click Open. - 3
Click the Browse button next to After Image. - 4
Select the file Umbra_SaghandIran_2023_08-25_slc.sml and click Open. - 5
Leave the Spatial Subset field blank. This is for selecting a pre-defined shapefile of the area of interest. You will process the entire images instead.
- 6
Click the Optional tab. - 7
Click the DEM Option drop-down list and select the Use Input DEM option. - 8
Click the Browse button next to Input DEM and select the file SRTM DEM Iran.dat. - 9
Click Open. This is a Shuttle Radar Topography Mission (SRTM) DEM with 1-arc-second resolution. A DEM is required in order to co-register the input images and to georeference the change detection image. - 10
Select the Yes option for Subtract Geoid. The CCD workflow requires ellipsoidal heights in the DEM. Since SRTM DEMs contain geoidal heights, you must subtract the EGM2008 geoid from them. - 11
Click the Geoid Type drop-down list and select EGM2008. - 12
Remove the default value of -32768 from the Data Ignore Vaue for DEM field. - 13
For the Output Coordinate System, keep the default value of WGS 1984.
Set Export Options
- 1
Click the Export tab. - 2
In the Grid Size field, enter a value of 3 meters. Although the Umbra images have a resolution of 0.372 meters, increasing the Grid Size value will speed up processing and result in a smoother CCD image. - 3
In the Output options field: select Filtered Classification, hold down the Ctrl key, and select Coherence. - 4
The filtered classification image and coherence image will be written to the directory specified in the ENVI Output Directory preference. To specify a different output folder, click the Browse button next to Output Folder and choose a different folder.
- 5
Click the Next button. Processing takes several seconds to complete. When it is finished, the coherence image is added to the Layer Manager and displayed in the Image window. The CCD tool advances to the Change Detection Threshold step. The CCD process is not yet complete; you will need to review the coherence image and determine suitable threshold settings next. - 6
In the Layer Manager, uncheck both Umbra images to hide them. Only the coherence image is displayed.
Review the Coherence Image and Threshold Setting
Coherence values range from 0 (low coherence, or unstable or non-reflecting areas) to 1 (high coherence, or stable areas). Thus, brighter pixels in the coherence image indicate stable areas, or those that did not change during the eight-day period. Black and dark-gray pixels indicate changes to the landscape that occurred during that time. The figure below shows some examples that are likely caused by excavation, soil disturbance, and vehicle movement. The black linear feature could be a trench. The coherence image definitely reveals construction activity in and around the village.
The Coherence Threshold value determines how much change the final classification image should show. Moving the slider toward the right (increasing the threshold) will classify more pixels as “Change.” Moving the slider toward the left (lowering the threshold) will classify fewer pixels as “Change.”
- 1
Keep the default value of 0.20 for Coherence Threshold.
- 2
Click the Next button. The tool proceeds to the Classification Smoothing step, and ENVI displays the initial classification result. - 3
Look at the Layer Manager and notice that the classification image has three classes: “Unclassified” (black), “No Change” (gray) and “Change” (green).
Smooth the Classification Image
This step is optional; You can clean up the classification image by filtering out small groups of pixels and aggregating the remaining pixels into clusters. This step removes most of the speckled noise that you see in the initial classification image.
- 1
The defaultKernel Size value is 3. Enable the Preview option to see what the final classification will look like with this value. The image reveals more groups of “Change” pixels, but many pixels are still scattered. - 2
Enter a Kernel Size value of 7 and press the Enter key. The Preview Portal updates accordingly. Increasing the value will cluster adjacent pixels into larger groups while filtering out smaller groups and individual pixels. The result is a cleaner classification image.
- 3
Disable the Preview option. - 4
Click the Next button to proceed to the Report panel. - 5
Click the Finish button. Three new layers are added to the Layer Manager and displayed in the Image window:
- •
coherence.dat: Coherence image * •
acd_ann_info.anz: Annotation layer showing the line-of-sight (range) and heading (azimuth) directions, along with basic metadata. You must zoom out of the view to see the annotation layer, which is displayed to the upper-right of the coherence image. * •
change_detection_map_fil.dat: Smoothed/filtered coherence classification image
- 6
In the Layer Manager, drag and drop the change_detection_map_fil.dat layer above the coherence.dat layer.
- 7
Uncheck the No Change and Unclassified classes in the Layer Manager. The green “Change” pixels overlay the coherence image.
Since the Umbra images have a resolution of 1 meter, they are limited in their ability to detect foot traffic. Higher-resolution SLC SAR data is necessary for that. Instead, areas with low coherence in this example most likely indicate vehicle traffic and excavation.
Another alternative is to use the SAR and Coherent Amplitude Change Detection tool with the same imagery. This tool provides a more thorough classification, as it considers both phase (for coherence) and amplitude (for backscatter ratio).
This concludes the quick guide.
Your input is important to us, please take a few moments to fill out ourQuick Guide Feedback(opens in a new tab)form.
© 2025 NV5 Geospatial Solutions, Inc. This information is not subject to the controls of the International Traffic in Arms Regulations (ITAR) or the Export Administration Regulations (EAR).