Polarimetric Tutorial
Polarimetric Tutorial
Polarimetric Tutorial
Polarimetric Tutorial
Issued March 2015
Updated August 2016
Luis Veci
Polarimetric Tutorial
The goal of this tutorial is to provide novice and experienced remote sensing users with step-by-
step instructions on working with RADARSAT-2 data with the Sentinel-1 Toolbox. The tools have
to be developed with the goal of making polarimetric processing easy to use and intuitive.
In this tutorial you will process a Quad Pol RADARSAT-2 product and produce polarimetric
classifications of the data.
For an introduction to polarimetric concepts, please see the "Radar Polarimetry" chapter of the
"Fundamentals of Remote Sensing" tutorial from the Canadian Centre for Remote Sensing
(CCRS).
Sample Data
Sample data for RADARSAT-2 Fine Quad-Pol products supplied by MDA can be found at:
For this tutorial, we will use the Vancouver Fine Quad Frame 1 dataset. Vancouver in British
Columbia is the third largest metropolitan area in Canada located on the Pacific coast.
Supported Products
The toolbox can support Quad Pol SLC products from:
RADARSAT-2
TerraSAR-X
ALOS PALSAR 1 & 2
SENTINEL-1
ENVISAT ASAR
RADARSAT-2
2
Polarimetric Tutorial
TerraSAR-X
ALOS PALSAR 1 & 2
Step 1 - Open a product: Use the Open Product button in the top toolbar and browse for
the location of the Vancouver Fine Quad Frame 1 RADARSAT-2 product.
If your product is contained within a zip file, the Toolbox will also be able to open the product
simply by selecting the zip file.
Opening a Product
Step 2 - View the product: In the Products View you will see the opened product. Within the
product bands you will see four polarizations:
VH
VV
HH
HV
For each polarization, there will be the complex data i and q bands and two virtual bands for
intensity and phase.
3
Polarimetric Tutorial
Products View
Step 3 - View a band: To view the VH band, double-click on the Intensity_VH band. Zoom in
using the mouse wheel and pan by clicking and dragging the left mouse button.
4
Polarimetric Tutorial
Intensity_VH Band
Creating a Subset
To reduce the amount of processing needed, you may create a subset around the particular area
in which you are interested.
Step 4 - Create a subset from the view: Once you have zoomed and panned to your area of
interest, right click on the image view and select Spatial Subset from View in the Context
menu.
5
Polarimetric Tutorial
Context Menu
The subset dialog will automatically select the area you were viewing.
6
Polarimetric Tutorial
By default, all bands will be included in the subset. You will need all the bands to do the
polarimetric processing. Press OK to create the subset.
Calibration radiometrically corrects a SAR image so that the pixel values truly represent the radar
backscatter of the reflecting surface.
The corrections that get applied during calibration are mission-specific, therefore the software will
automatically determine what kind of input product you have and what corrections need to be
applied based on the product’s metadata. Calibration is essential for quantitative use of SAR
data.
Step 5 - Calibrate the product: From the Radar menu, go to the Radiometric menu and select
Calibrate.
Radiometric Menu
The source product should be your newly created subset. The target product will be the new file
you will create. Also select the directory in which the target product will be saved to.
7
Polarimetric Tutorial
Calibration Dialog
If you don’t select any source bands, then the calibration operator will automatically select all real
and imaginary (i, q) bands.
NOTE
For polarimetric processing the data must be complex. By default, the calibration
operator will produce real sigma0 bands. To produce complex output, check
mark the Save in complex parameter.
Covariance matrix C2
Covariance matrix C3
Covariance matrix C4
Coherency matrix T3
Coherency matrix T4
The Coherency matrix T3 is sometimes preferred because its elements have a physical
interpretation (odd-bounce, even-bounce, diffuse, etc.).
8
Polarimetric Tutorial
Use the Matrix Generation operator when you would like to explicitly select which matrix to use.
For simplicity, a Quad Pol SLC product can be used directly by any polarimetric operator. In such
a case, the input Quad Pol will automatically be converted to a T3 matrix.
Step 6 - Generate a T3 matrix: Select Polarimetric Matrix Generation from the Polarimetric
menu.
Polarimetric Menu
The source product should be your newly created calibrated subset. The target product will be the
new file you will create. Also select the directory in which the target product will be saved to.
9
Polarimetric Tutorial
In the Processing Parameters tab, select a T3 matrix to convert the Quad Pol product into a
Coherency matrix T3. Press Run to begin processing.
When the processing completes, a new product will be added to the Products View. You will
notice the new bands produced correspond to the elements of the T3 matrix.
Boxcar
Improved Lee Sigma
Refined Lee
Intensity Driven Adaptive Neighbourhood (IDAN)
Step 7 - Apply a Speckle Filter: Select Polarimetric Speckle Filter from the Polarimetric
menu.
10
Polarimetric Tutorial
In the Processing Parameters tab, select the Refined Lee speckle filter. Press the Help button
to call up the online help for further information.
When the processing completes, your new speckle filtered product should have the same bands
as your T3 product, however the data will have been filtered.
11
Polarimetric Tutorial
Open the T11 band in both the T3 product and in new speckle filtered T3 product to compare
before and after images. The resulting image will have less speckle but also appear more blurred.
Before After
To clean up your Products View, you may now right-click on the speckle filtered T3 product and
from the popup menu select Close All Others to close all other products and leave only the
speckle filtered T3 opened.
12
Polarimetric Tutorial
Polarimetric Decompositions
Polarimetric decompositions allow the separation of different scattering contributions and can be
used to extract information about the scattering process.
Sinclair
Pauli
Freeman-Durden
Yamaguchi
Van Zyl
Cloude
H-a Alpha
Touzi
13
Polarimetric Tutorial
The window size parameter corresponds to the amount of averaging applied to each pixel.
When the processing completes, your new Freeman-Durden decomposition product will have
three bands corresponding to double bounce, volume scattering and surface scattering.
14
Polarimetric Tutorial
Step 9 - View in RGB: You can view all three bands in an RGB colour view by right-clicking on
the product name and selecting Open RGB Image View from the popup menu.
Within the RGB channel selection dialog, select the Red, Green, Blue components for the
respective bands Freeman_dbl_r, Freeman_vol_g, Freeman_surf_b. Press OK to create the
RGB view.
15
Polarimetric Tutorial
The resulting RGB image shows surface scattering in blue from the airport runways, roads and
water bodies. Buildings produce double bounce and are shown in red. Vegetation produces
volume scattering and is therefore shown in green.
Step 10 - Export the RGB as an image: You may now wish to export the colour image to an
image file format such as JPEG or PNG to use in a report or presentation.
The processed data will usually be saved as Float64 data type. Although it is possible to save
JPG, PNG and GeoTIFF files as Float64, the typical image viewing software may expect data to
be in UINT8 or UINT16.
To convert the data type, select Convert Data Type from the Utilities Data Conversion menu.
16
Polarimetric Tutorial
In the Convert Data Type dialog, choose to output UINT8 using linear scaling clipping to 95% of
the histogram.
17
Polarimetric Tutorial
Now you may export the converted product to a common image file format.
From the File menu, go to the Product Writers submenu and select Export View (bmp, jpg,
png). In the Save dialog, enter the name, location and type of image to export.
Step 11 - Produce all other Decompositions: Using the speckle filtered T3 as input, repeat the
decomposition processing for all other decompositions and compare the results.
18
Polarimetric Tutorial
Sinclair Decomposition
19
Polarimetric Tutorial
Pauli Decomposition
20
Polarimetric Tutorial
Yamaguchi Decomposition
21
Polarimetric Tutorial
22
Polarimetric Tutorial
Cloude Decomposition
23
Polarimetric Tutorial
H-a-Alpha Decomposition
To clean up your Products View, you may now right-click on the speckle filtered T3 product and
from the pop-up menu select Close All Others to close all other products and leave only the
speckle filtered T3 opened.
24
Polarimetric Tutorial
Unsupervised classification does not rely on user specified classes to be matched. Instead, it
automatically determines what classes exist in the data and how best each pixel can be grouped.
Step 12 - Apply unsupervised classification: Using the T3 speckle filtered product as input,
select Unsupervised Polarimetric Classification from the Polarimetric menu.
Press the Help button to call up the online help for further information.
25
Polarimetric Tutorial
When the processing completes, your new classification results product will have a single band
with several regions each belonging to one of nine classes.
26
Polarimetric Tutorial
Both the Cloude-Pottier and the Wishart classifiers are based on the use of the Entropy (H) /
Alpha (α) plane. The Wishart classifier will continue to compute the centres of the nine clusters,
and then reclassify the pixels based on their Wishart distances to cluster centres.
27
Polarimetric Tutorial
You may change the default colours for each class within the colour manipulation tool window.
Use the pull-down control on the colour to select a new colour. Use the export and import buttons
on the right-hand side to save and load colour look-up tables.
28
Polarimetric Tutorial
http://step.esa.int/main/doc/tutorials/
http://forum.step.esa.int/
29