Language selection


PWS Function Description

Table of Contents - this page

  1. Tools panel
  2. Data Format
  3. Conversion to CCRS image format + Load image
  4. Local Analysis Tools
  5. Image Synthesis Tools
  6. Full Resolution

2.1 Tools panel

When you execute PWS, the tools panel appears on the desktop. The tools panel contains:

  • Radarsat-2 to CCRS image conversion
  • Load image
  • Local Area Analysis
  • Image Synthesis
  • View data
  • Demo
  • Display PWS Output Images

2.2 Data Format

The PWS can read the following data:

Polarimetric Radarsat-2:
Single-Look Complex (SLC): 4 channels of complex values HH, VV, HV and VH

Polarimetric ALOS:
Single-Look Complex (SLC): 4 channels of complex values HH, VV, HV and VH

Single-Look Complex (SLC): 4 channels of complex values HH, VV, HV and VH.
Geocor product: Geocoded data under Kennaugh matrix format. The JPL compressed format was used to save data volume, under target and system reciprocity assumption (HV=VH).


SIR-C (quad-pol) SLC
SIR-C (quad-pol) MLC

2.3 Conversion to CCRS image format + Load image

2.3.a Radarsat-2 to CCRS image file format conversion.

Radarsat-2 image file scene is in TIFF format. Since PWS processes the SLC data under the to CCRS format, the conversion of RADARSAT2 Tiff file to CCRS format file is required.

  1. Click on “Radarsat-2-ALOS-PALSAR to CCRS Image Conversion”.
  2. Go to the directory which contains Rasarsat-2 scene.
  3. Select “product.xml” file
  4. After the file is selected you will be prompt to enter line and pass number (between 1 and 9) in order to build the output file names. (CCRS file format)
  5. When the file format conversion is completed Please load the Radarsat-2 CCRS format data as indicated in Section. (2.3d)

2.3.b ALOS-PALSAR to CCRS image file format conversion.

Since PWS processes the SLC data under the to CCRS format, the conversion of ALOS-PALSAR SLC files to CCRS format files is required.

  1. Click on “Radarsat-2-ALOS-PALSAR to CCRS Image Conversion”.
  2. Go to the directory which contains the ALOS scene.
  3. Select the LED file
  4. After the file is selected you will be prompt to enter line and pass number (between 1 and 9) in order to build the output file names. (CCRS file format)
  5. When the file format conversion is completed Please load the ALOS image. (2.3d)

2.3.c ALOS Faraday rotation correction

The Faraday rotation is measured using the Bickel and Bates method (Bickel and Bates 1965, Touzi and Shimada 2009). Farday rotation can be removed as follows:

  1. Load the ALOS image (2.3d) under the CCRS SLC format.
  2. Go to the Image Synthesis
  3. Select Faraday Rotation
  4. Select a sample (sub-image) for the computation of the Faraday rotation angle and Run
  5. The Faraday rotation angle is viewed. If the angle is larger than 3 degree, Please ask for Faraday removal.
  6. PWS generate the SLC files corrected for Faraday rotation.

2.3.d Load image

  1. Click on “Load image” on the top of the GUI.
  2. Select the header or image file that you want to display.
    • The CCRS-SLC files are generally named “l#p#??polgasp.hdr” for the
      header files and “l#p#??polgasp.img” for the image files. All four
      complex channel images HH, VV, HV, and VH are needed.
  3. After selecting your file, specify which type of data it is: SLC or Geocor product – (not yet implemented)
  4. An image size limit has been set to 50,000 columns or 2 Gb (decompressed GEOCOR product). If the image is larger than the limit, you will be prompted to crop the image. If desired, a cropping function will start automatically.
  5. The selected image will be displayed in a new window. The other polarizations can be visualized by clicking on the polarization menu at the top of that window. If this is the first time that the selected image/polarization has been loaded, a magnitude conversion process that converts the 32-bits Floats Complex data to 8-bits unsigned integers, is automatically launched.

2.4 Local Analysis Tools

Local analysis tools provide polarization information of a given sample selected with the Polygon function.

2.4.1 Define polygons

To use one of the local analysis tools, first you need to define a polygon by clicking on the “define polygon” button. You won’t be able to zoom in on the image during the polygon selection, so zoom in your area, if needed, before clicking on define polygon.

  1. To create your polygon, use the left mouse button to select corners of your area.
  2. To delete the last point, use “backpace” on your keyboard. Clicking the right button of your mouse sets the last point of your polygon.
  3. If you accept this polygon, you will be prompt to provide a name for that area.
  4. If some polygons have already been created, you can select one of them or create a new one. Created polygons are permanently saved.

2.4.2 Local area tools

a) Polarimetric signature

Generate the co-polarized and cross-polarized polarimetric signatures of the selected polygon. Select “Save” in the File menu if you wish to save the results. The pedestal of the polarimetric signature is provided as well as the antenna illumination (incidence) angle. A minimum of 400 independent samples per polygon is required for accurate estimation of the polarimetric parameters.

Area 1: Ice
RADARSAT-2 Acquisition Date: 2008-03-07

Larger image


b) Statistics

Provide estimates of the polarimetric parameters that characterize the scattered and the received waves. The parameters computed for each selected polygon are saved in a “Comma Separated Values” .csv ASCII file. The following parameters are estimated for the polygon under study:

  1. so - mean and standard deviation in dB for the linear polarizations HH, VV, and HV.
  2. so mean and standard deviation for the right (R) and left (L) circular polarizations RR, LL, and RL.
  3. Span, which corresponds to the total scattered wave intensity.
  4. Extrema of the scattered wave intensity Romax and Romin.
  5. Extema of the degree of polarization pmax and pmin. The combination of these two parameters with Romax and Romin permits an excellent characterization of the type of scattering mechanism and its heterogeneity (Touzi et al. 1992).
  6. Extrema of the completely polarized wave intensity CPMAX and CPMIN.
  7. Extrema of the completely unpolarized wave intensity CUPMAX and CUPMIN.
  8. Extrema of the received power PMAX and PMIN.
  9. Extrema of the cross-matched power PXMAX and PXMIN; the scattered wave is maximized and the received antenna polarization is cross-matched to the scattered wave.
  10. Van Zyl coefficient of variation PMIN /PMAX.
  11. Polarization signature pedestals.
  12. Channel coherence amplitude ? and phase F (in degree) for HH-VV, HHHV, and VV-HV.
  13. The number of pixels per polygon, as well as the incidence angle (?) for Radarsat-2 SLC products, are also provided.

c) Histogram

Generate the amplitude histogram of the 4 polarizations. Select “Save” in the File menu if you wish to save the results.

Area 1: Ice
RADARSAT-2 Acquisition Date: 2008-03-07

Larger image


d) Poincaré Sphere

Map (as a red dot) the polarization state of the scattered wave on the Poincaré sphere. Multi-look data Polarization state of a partially polarized wave is mapped within the Poincaré sphere. The user can select the transmitted signal configuration and the number of looks (per pixel). To show the position of the polarization state within the sphere, a longitudinal section of the Poincaré sphere is presented. The longitude, as well as the thickness of the section can be selected using the slider control. Blue circles represent the longitudinal section intersection with the surface of the sphere. Red dots indicate the polarization states and their distance to the blue circles provides an indication of the degree of polarization. Select “Save” in the File menu if you wish to save the results.

Polarization State
Sphere surface

Larger image


e) Channel Coherence (magnitude and Phase difference)

The polar representation is used to present the magnitude and phase of the complex coherence.

Area 1: Ice
Correlation between polarization window: 6 x 3

Larger image


f) Cloude’s decomposition

Derive Cloude’s parameters (H), anisotropy (A), a and ß, derived from Cloude’s decomposition algorithm (Cloude and Pottier 1996). Two options are offered:

  1. Estimate of the means of Cloude’s parameters H, A, a and ß. To obtain unbaised estimates, the averaged coherency matrix is calculated over the polygon area under study, and used to derive the parameters.
  2. Display the histogram of a as a function of entropy H within Cloude’s parabola. a and H are derived using coherency matrix derived from multi-look samples. The number of looks (i.e. number of independent samples per pixel) can be fixed in terms of pixel window size.

Area 37: Line 1 Pass 9 9-Jul-1998
Cloude Alpha (a) vs Entropy (H) - Density Histogram

Larger image


g) Batch-Tool

Execute above functions in a batch processing over selected polygons. By selecting this function and clicking on ‘APPLY’, you are prompt to select the polygon(s) to process. You can select all the polygons by clicking on “Select all” button. Holding <CTRL> button and clicking on desired polygons can do specific selection. By clicking on “ok” button, you are prompt to select function(s) to process. Resulting figures are saved with a name based on the general header file, the polygon number, the function used and graphic format (.tiff, .bmp or .ps).


Header: l2p3.hdr
Function: polarimetric signature
Polygon #: 3
Format: PS

Mask option: A bitmap file, generated by a classification/segmentation software, might be used as an input to the PWS to compute the statistics of the different classes/segments. All the pixels of each segment should have the same value in the bitmap file. The segment number is an integer number between 1 to 255, and pixels that are assigned the zero value are not considered. The bitmap file name is based on the general header file name, ex: 1) SLC case: “l1p8polgasp.hdr”, the corresponding bitmap name is “l1p8polgaspbmp.img”; 2) GEOCOR case: “imageken.hdr”, the corresponding bitmap name is “imagekenbmp.img”. The bitmap should have the same dimension as the SAR image. The user will not be prompt to select bitmap file, if the default file name exist, it uses it.

h) Class Stats

Evaluates the Polarimetric Signature (a) and the Statistics (b) on a selected class name. Two options:

  • 3. Polygons: Regroups all polygons with the same name; evaluates the average Kennaugh matrix; and computes the average statistic for that class name. User select polygons name on the pop-up list. Statistics results are saved in the default output.csv file. The Polarimetric signature figure is not automatically saved, user need to select “file/save” if desired.
  • 4. Bitmap: Regroups all image samples corresponding to a specific class defined by a bitmap image (mask) and evaluates the statistics. User needs to select a bitmap file, previously created from a classification/segmentation software. The bitmap is a 8 bits file, where samples value correspond to a specific class. “0” is reserved for unclassified samples. The bitmap image name need to have the extension “.img” and its corresponding header file “.hdr” should respect the structure of the following example:

    number_lines 300
    number_samples 3000
    line_up 101
    line_down 400
    sample_left 1001
    sample_right 4000
    1 Wheat
    2 Deciduous
    5 Grass
    10 Barley

number_lines and number_samples are the number of lines and samples per line in the bitmap image; line_up, line_down, sample left and sample_right correspond to location of the bitmap in the full SAR image (optional if the bitmap image is the same size as the SAR image); Bitmap_begin and Bitmap_end are the markers defining the beginning and the end of the class list. Each class in the class list is composed by a number (1-255, corresponding to value in the bitmap) and a class name.

User can select more than one class to run under batch mode.

i) Saving and viewing the results

Once the local area analysis tool is selected, select the “Apply” button to execute the function. Numerical output values are saved in a file such as “image_output.csv” (the csv extension is for Comma Separated Values format). View the results by clicking on the “View data” button on the bottom-left of the control tools panel. Microsoft Excel, Corel QuattroPro, and Lotus123 will automatically import the “image_output.csv” file.

2.5 Image Synthesis Tools

Image synthesis tools permits the synthesis of the received radar reflectivity images for any combination of transmitting-receiving antenna polarizations. The synthesis is performed on the portion of the image selected by the user. Define the coordinates of the sub-image before running the synthesis tool by selecting the “Define sub-image” button.

2.5.1 Define sub-image

You can define the sub-image in 3 ways:

  1. Full image: The synthesis will be performed on the entire image.
  2. Interactive: With your left mouse button select a corner of your subimage, hold and drag your cursor to generate the area. Release the button when the rectangle covers the desired area.
  3. Coordinates: The user is prompted to enter the top-left and bottom-right coordinates of the desired sub-image.

The user can specify the output sample spacing:

Full resolution: 1-look image.

Square samples: set automatically the size of the multi-look pixel to
synthesize an image with square pixel. For example, a sample of 9x1
pixels is used with the Convair-580 1-look SLC data.

User defined: The user specifies the multi-look pixel size (number of pixels
in azimuth and range per multi-look sample).

2.5.2 Tools

a) Sigma naught (ß°=s° sin ?inc)

Converts the selected sub-image, to a power image of the transmitting and receiving antenna polarizations defined by the user. The channel generated is saved using the 32 bits floating point format and a separate header file.
NOTE: Intensity value are s° for Radarsat-2 data

b) Coherence

Generates HH-VV, HH-HV or VV-HV channel coherence map for a selected sub-image. The saved Outputs are two 32 bits Real float images
(amplitude and phase) with corresponding header files. If “full resolution” is selected, only a phase image will be created (coherence magnitude is 1
in this case).

c) Swartz-contrast

Optimization of two-target contrast: generate the image at the antenna polarization combination, which optimizes the contrast between two selected targets, using Swartz method (Swartz et al., 1988) developed for azimuthally symmetrical targets. At least two polygons need to be defined by the “define polygons” function before to be able to run this function.

Output header file contains information on the transmitting-receiving configuration, the two polygons used and the optimum contrast obtained

d) Channel Phase

Generate Single-Look phase and phase differences between polarizations without reciprocal assumption.

e) Touzi SSCM: Symmetric Scattering Characterization Method

The symmetric scattering characterization method, known as the Touzi SSCM, has been recently introduced by Touzi and Charbonneau (2002) for high-resolution characterization of target symmetric scattering, under coherent conditions. The Touzi SSCM is based on the target Poincaré sphere representation, which was also introduced by Touzi and Charbonneau (2002) for a high resolution mapping of the target symmetric scattering, as well as assessment and validation of the coherence nature of target scattering prior to the scattering decomposition. The method is applied on coherent scattering of significant symmetric scattering component; and the ratio of the maximized symmetric scattering component to the total scattering, named dsym, is provided to assess the results. Each coherent symmetric scatterer is uniquely mapped as a point of latitude 2?c and longitude 2?c on the surface of the target Poincaré sphere presented below, where the scatter angles ?c and ?c are derived as a function of the symmetric scattering parameters. A partially coherent symmetric scatterer is represented as a point inside the sphere at a distance from the sphere center determined by the degree of coherence psym introduced also in Touzi and Charbonneau (2002). Psym is used to assess the coherence of distributed targets; a 9x9 square pixels is coherent if the degree of coherence Psym is higher than 0.86. The Rician test is used to validate the coherent nature of a point target, and a Rician threshold of 18 dB is taken. This threshold is applied on the maximumpower return |m|2.

To remove the rotation phase ambiguity, only half of the sphere is used with ?c varying within the interval [0,p/2].

Target Poincaré Sphere

Larger image


dsym: degree of symmetry - test_dysm.img
psym: degree of coherence - test_psym_w9.img
Huynen Max. Power |m|2 - test_Hm2.img
Coherent Target Bitmap - test_SSCMthr18g0.86w9.img
?c: target poincaré sphere ½ latitude - test_psic18g0.86w9.img
?c: target poincaré sphere ½ longtitude - test_chic18g0.86w9.img
?a : target polarization orientation - test_psia18g0.86w9.img

Generated outputs are labeled, based on the root name and threshold values.

Example: psym: 0.86 ; avg.window: 9 ; Rician: 18 dB ; and root: test


By default, Huynen maximum power |m|2, ?c and ?c are displayed on screen when processing is done. The coherence tests are run, and the SSCM is only applied within coherent regions. Non-coherent targets are assigned zero values for ?c, ?a and ?c. User can zoom on target, by right/left click on the images. User can also load the images: Degree of coherence, Degree of symmetry and Target Orientation from the "Load Results" option in the menu bar of the "Huynen Intensity" figure.

SSCM is a full resolution decomposition technique, which implies that the generated images are in full resolution sample size format, even if "square samples" or "user define" option was selected during the "Define sub-area" step.

Reprocessing with different thresholds

When the SSCM option is launch, user is prompt to generate the SSCM decomposition on the selected area (option: YES) or to use previous decomposition results (option: Use Existing). By selecting "Use Existing", the user can generate new output images from different analysis thresholds. User will be prompt to selected a previously process SSCM log file (*_sscm.log) as input. The corresponding area of the selected file doesn't need to be the same as the defined area on the "Image display" window.

WARNING: This method is defined only for SLC data products.

f) Touzi Decomposition

Generates the Touzi decomposition (Touzi et al. 2007, 2009) parameters for the dominant, medium or low scattering. All the decomposition parameters are provided in floating 32-bit format. Besides, a circular coding is used to represent the Touzi phase on a tif 8-bit file. This leads to a coarse phase classification that has been shown very useful for wetland scattering classification (Touzi, Deschamps et al. 2007, Touzi et al. 2009).

g) Save Kennaugh

Generates a 10 channels image file corresponding to the Kennaugh matrix (reciprocal assumption) of the selected sub-image. User can use the output file as an image file by loading it (“Load image” button) and selecting “GEOCOR (Kennaugh)” option on the “Data Format” window. Output file names (image and header) are build from the root name provide by user ex: root = test ? output: testken.img and testken.hdr. This function is useful when user want to keep a smaller image for further

h) Batch-synth

Executes above points a, b, d and f, as well as the Cloude’s parameters, in a batch process for the defined sub-image. User is prompted to select the options and output root name. Output filename are composed by the root name and option selected.

Root = test
Option: HH.    output: test_HH.img

Note: If large area is defined and multiple options selected, user should disable “Display” option to avoid running out of memory.

Cloude decomposition outputs correspond to:

Entropy (H) - test_H.img
Anisotropy (A) - test_A.img
Alpha angle (a) - test_APH.img
Beta angle (ß) - test_B.img
Eigenvalues (?1,2,3) - test_EIG#.img

Outputs files generated in this section can be reopen with the “Raw Image” function (on the top of the “Polarimetric Workstation” window). Images are displayed in full resolution. If image is to large, corresponding 8 bits image is displayed. Note, if enhancement function is used, the right color bar disappears because it’s become useless. Any image processing software can also read these output images as RAW data format.

2.6 Full Resolution

On the Image Display window, there is an option Full Resolution. When clicking on this option, use the mouse to select the center of a sub-window to be displayed in full resolution.

Full resolution menu

- Position

Gives the coordinates and the power in dB of the pixel under the cursor in the Full resolution window. The coordinates are relative to the sub-window. By click-hold left mouse button and moving the mouse, you get the distance between two pixels (distance is in pixel units).

- New Area

This function allows you to relocate the Full resolution window on the Image Display. Click on main Image to select a new area.

- Target Analysis

2.6.1 Target analysis

This tool analyses and computes reference point target signals for scene calibration [Touzi et al. 1993]. Peak signal intensity location is given for the 4 channels to assess channel registration. Channels are oversampled (by 8) using the zero-padding method, to give the exact position of the peak signal intensity. The phase and intensity of the peak, as well as the peak intensity-to-clutter ratio are computed. The intensity integration method [Gray et al. 1990] and the complex integration method [Touzi, 1992] are implemented in order to provide an accurate estimate of target phase and magnitude at the presence of system misfocussing.

By clicking on this option, Target Analysis GUI will pop-up. You need to defined the following parameters:

Get Targets: Select point target(s) (1 or more) in the Full Resolution window with left mouse button. When done with selection, use right mouse button.

Deselect: Remove selected target(s) (latest add, first remove)

Get Clutter: Define the clutter area (target's background). Draw a polygon area close to the target. Be careful to not be too close to the target. Area should contain at least 400 independent samples. Use mouse to draw the polygon (left button for control points and right button to end polygon

Target Type: Select the corresponding type of your targets selected

  • Corner corresponds to calibration corner reflector or any point
  • Polarimetric active radar calibrator (PARC) corresponds to the response of a Portable Active Radar Calibrator. Active calibration target with a delay line loop that gives a series of repeated transmissions of attenuated signals at fixed time intervals. Target analysis provides the parameters needed for PARC signal assessment [touzi 93].

    For PARC, user should select (Get Target) only the first response
    (the brightest) as target, the others recirculations will be
    automatically detected.

Integration Window: The window used to integrate the information on targets. Its dimensions are function of the 3 dB width of the target peak (azimuth and range). User selects the number of side lobes to integrate.

Lobes = 0 :  Only main lobe is integrated

Lobes = 1, 2, 3 :  Integration window contains secondary lobes 1, 2, 3.

Peak intensity selection:

  • Auto: Will look inside the integration window to find the highest point target backscattering and centred the integration window analysis on
    this target. This procedure is repeated for each selected target defined by Get Targets.
  • Manual: Will centred the integration window analysis on the input pixel coordinates.

Filename: Output root filename of the numerical and graphical results.

Example:  Filename = test
Output files: (graphical file)
test_target.csv (ascii text file)

The postscript (*.ps) file contains all the graphical results figures displayed. The numerical values are also saved in the *.csv (can be read with any spreadsheet handler or text software). The postscript format is used because it permits to save multiple figures in the same file. Postscript files can be open with Ghostview freeware available a or other www sites.


Execute the target analysis task


  1. For a target, if the target-to-clutter for a given polarization is lower than clutter threshold, the analysis is not retained. The clutter threshold is given by the average clutter intensity plus one standard deviation.

  2. To avoid multiple graphical windows on desktop, if more than a point target is selected, the target analysis graphical windows will be close automatically. The graphic results can be view after task by ghostview.

2.6.2 Graphical results examples:

a) Passive Target:

Peak Analysis of line 2 pass: 2 l2p2hhpolgasp.img 01-Oct-2002 12:08:07

Larger image


b) PARC:

ARC Analysis of merbleu2001 line: 2 pass: 2 l2p2hvpolgasp.img 01-Oct-2002 12:08:18

Larger image

Report a problem on this page
Please select all that apply:

Thank you for your help!

You will not receive a reply. For enquiries, contact us.

Date modified: