Radar data processing: Difference between revisions

From GRASS-Wiki
Jump to navigation Jump to search
(→‎Sample data: add new data)
m (URLs updated)
 
(5 intermediate revisions by 3 users not shown)
Line 4: Line 4:
* '''SAR backscatter analysis''' ([http://www.palsar.ersdac.or.jp/e/data/analysis.html about]): enables to estimate volume of water contained in soil, volume of biomass in forests, conditions of waves in the ocean etc.
* '''SAR backscatter analysis''' ([http://www.palsar.ersdac.or.jp/e/data/analysis.html about]): enables to estimate volume of water contained in soil, volume of biomass in forests, conditions of waves in the ocean etc.
** Requirements: ...tbd.
** Requirements: ...tbd.
* '''SAR interferometry processing''' ([http://www.palsar.ersdac.or.jp/e/data/interf.html about]): enables to produce elevation data and to measure topographical change (two SAR images are used - called Master Image and Slave image, respectively)
* '''SAR interferometry processing''' ([http://gds.palsar.ersdac.jspacesystems.or.jp/e/about/ about]): enables to produce elevation data and to measure topographical change (two SAR images are used - called Master Image and Slave image, respectively)
** Requirements: complex SAR data - Single Look Complex as parallax is not removed
** Requirements: complex SAR data - Single Look Complex as parallax is not removed
** General steps (cited from [http://www.nrcan.gc.ca/earth-sciences/geography-boundary/remote-sensing/radar-remote/2122 CCRS Radar tutorial]):
** General steps (cited from [https://www.nrcan.gc.ca/earth-sciences/geomatics/satellite-imagery-air-photos/satellite-imagery-products/educational-resources/9579 CCRS Radar tutorial]):
*** Process data to SLC images
*** Process data to SLC images
*** Register the two images to 1/10 pixel
*** Register the two images to 1/10 pixel
Line 43: Line 43:
|-
|-
|}
|}


== PALSAR analysis in GRASS GIS ==
== PALSAR analysis in GRASS GIS ==
Line 70: Line 69:
Optimize the color table to better see the pixels:
Optimize the color table to better see the pixels:
# Press the right mouse button on the PALSAR map in the layer manager and select "Select color table".
# Press the right mouse button on the PALSAR map in the layer manager and select "Select color table".
# In the "Required" tab the map name has to be selected; in the "Optional" tab select "Histogram equalization" and the "bcyr" color table. Click "Run" to generate the new color table. Then use the "Layer redraw" icon [[ファイル:GRASS-Layer-raster-add.png|GRASS-Layer-redraw.png|Layer redraw]] to see the updated map in the map display window.
# In the "Required" tab the map name has to be selected; in the "Optional" tab select "Histogram equalization" and the "bcyr" color table. Click "Run" to generate the new color table. Then use the "Layer redraw" icon [[Image:GRASS-Layer-raster-add.png|GRASS-Layer-redraw.png|Layer redraw]] to see the updated map in the map display window.




Line 78: Line 77:
# To calculate the image statistics ({{cmd|r.univar}}): in the menu, Raster -> Reports and Statistics -> Univariate raster statistics. Select the map (should be already there) and tick the checkbox "Calculate extended statistics" in the "optional" tab, then "Run". This takes a moment... Find the result in the "output command" tab. Check "mean" and "3rd quartile" and note the values down somewhere. In this example: mean: 352.068, 3rd quartile: 514 etc. Then close the dialog window.
# To calculate the image statistics ({{cmd|r.univar}}): in the menu, Raster -> Reports and Statistics -> Univariate raster statistics. Select the map (should be already there) and tick the checkbox "Calculate extended statistics" in the "optional" tab, then "Run". This takes a moment... Find the result in the "output command" tab. Check "mean" and "3rd quartile" and note the values down somewhere. In this example: mean: 352.068, 3rd quartile: 514 etc. Then close the dialog window.
# These values could be used to define a threshold filter in {{cmd|r.mapcalc}}.
# These values could be used to define a threshold filter in {{cmd|r.mapcalc}}.


== Other PALSAR in GRASS GIS example ==
== Other PALSAR in GRASS GIS example ==
Line 84: Line 82:
This example is based on [http://www.palsar.ersdac.or.jp/e/product/sample.html sample data] on the PALSAR web site (L1.5 GC product, L150FBS_GC):
This example is based on [http://www.palsar.ersdac.or.jp/e/product/sample.html sample data] on the PALSAR web site (L1.5 GC product, L150FBS_GC):


   [[Image:PASL1500701140134440805020299B.jpg|600px|center|thumb|PALSAR L1.5 GC product sample scene]]
   [[Image:PASL1500701140134440805020299B.jpg|500px|center|thumb|PALSAR L1.5 GC product sample scene]]


Import:
Import:
Line 98: Line 96:
   d.vect osm_roads
   d.vect osm_roads


   [[Image:Palsar_osm_japan2008.png|600px|center|thumb|PALSAR L1.5 GC product sample scene imported in GRASS GIS with OpenStreetMap overlay (subscene, center_easting=133.838, center_northing=34.296)]]
   [[Image:Palsar_osm_japan2008.png|500px|center|thumb|PALSAR L1.5 GC product sample scene imported in GRASS GIS with OpenStreetMap overlay (subscene, center_easting=133.838, center_northing=34.296)]]


The map shows nicely the fit of the circa 6m resolution PALSAR data and the OpenStreetMap vector lines (compare to [http://maps.google.com/maps?ie=UTF8&oe=utf-8&q=34.296+133.838&ll=34.295799,133.838196&spn=0.177837,0.328217&t=h&z=12 Google Maps optical coverage]).
The map shows nicely the fit of the circa 6m resolution PALSAR data and the OpenStreetMap vector lines (compare to [http://maps.google.com/maps?ie=UTF8&oe=utf-8&q=34.296+133.838&ll=34.295799,133.838196&spn=0.177837,0.328217&t=h&z=12 Google Maps optical coverage]).
Line 121: Line 119:
== Sample data ==
== Sample data ==


* [http://www.palsar.ersdac.or.jp/e/product/sample.html ERSDAC data download] (various)
* [http://www.alos-restec.jp/en/staticpages/index.php/service-sampledata Palsar data download] (various)
* [http://radartools.berlios.de/data.php RAT data download] (various)
* [https://sourceforge.net/projects/radartools.berlios/ RAT data download] (various)
* [http://earth.esa.int/ew/earthquakes/Italy_April09/ ESA dataset package] (Italy Earthquake, April 2009)
* [http://earth.esa.int/ew/earthquakes/Italy_April09/ ESA dataset package] (Italy Earthquake, April 2009)
* [http://www.eorc.jaxa.jp/ALOS/en/kc_mosaic/kc_mosaic.htm PALSAR K & C Mosaic]
* [http://www.eorc.jaxa.jp/ALOS/en/kc_mosaic/kc_mosaic.htm PALSAR K & C Mosaic]
Line 137: Line 135:


[[Category: Documentation]]
[[Category: Documentation]]
[[Category:Image processing]]
[[Category: Image processing]]
[[Category: Import]]

Latest revision as of 11:52, 28 December 2016

Introduction to Radar data processing

SAR analysis methods

  • SAR backscatter analysis (about): enables to estimate volume of water contained in soil, volume of biomass in forests, conditions of waves in the ocean etc.
    • Requirements: ...tbd.
  • SAR interferometry processing (about): enables to produce elevation data and to measure topographical change (two SAR images are used - called Master Image and Slave image, respectively)
    • Requirements: complex SAR data - Single Look Complex as parallax is not removed
    • General steps (cited from CCRS Radar tutorial):
      • Process data to SLC images
      • Register the two images to 1/10 pixel
      • Over-sample by a factor of 2 in both dimensions
      • Filter common bands in spectrum
      • Conjugate multiply to form interferogram
      • Smooth the interferogram
      • Measure coherence
      • Unwrap phase
      • Estimate geometry parameters (especially baseline)
      • Remove flat-earth fringes
      • Convert unwrapped phase to height and/or motion
  • Polarimetric SAR processing (about): forest fire monitoring, classification of vegetation (height), water content in vegetation, monitoring of snow cover, condition of ice, flood monitoring and soil moisture.
    • Requirements: ...tbd.

Frequency and Wavelength of the IEEE Radar Band designation

Frequency [GHz] Wavelength Band
1-2 30-15cm L Band
2-4 15-7.5cm S Band
4-8 7.5-3.75cm C Band
8-12 375-250mm X Band
12-18 250-167mm Ku Band
18-27 167-111mm K Band
27-40 111-0.75mm Ka Band

PALSAR analysis in GRASS GIS

First we need to rewarp the .dat format to GeoTIFF which also applies the included Ground Control Points (GCPs) to the image. The resulting GeoTIFF file is then geocoded in Lat-Long/WGS84:

  1. gdalinfo PASL1500911220116591001270049_ortho.dat
  2. gdalwarp PASL1500911220116591001270049_ortho.dat PASL1500911220116591001270049_ortho.tif

Now we can auto-generate a GRASS Location from the GeoTIFF:

  1. Start 'grass64 -wx' or from menu (graphical user interface)
  2. Select Location Wizard
  3. Define a good name for the Location, then Next
  4. Select the item "Read projection and datum terms from a georeferenced file", then select the newly created PASL1500911220116591001270049_ortho.tif file, Next.
  5. Now the location is generated and the region is set to the image dimensions (degree in this case)
  6. You will return to the GRASS entry screen and can open the new location. Still the data have to be imported.

When in GRASS, import the map and display it:

  1. import with File -> Import raster map -> Import raster data using GDAL (r.in.gdal) -> Select again PASL1500911220116591001270049_ortho.tif and define name for this map in GRASS, when finished close dialog window
  2. Add map to map tree in the layer manager, using the icon Add raster map layer. It should be automatically displayed
  3. Use the zoom tool Zoom in in the map display manager to zoom Mount Fuji
  4. You will not recognize much yet since a better color map has to be assigned

Optimize the color table to better see the pixels:

  1. Press the right mouse button on the PALSAR map in the layer manager and select "Select color table".
  2. In the "Required" tab the map name has to be selected; in the "Optional" tab select "Histogram equalization" and the "bcyr" color table. Click "Run" to generate the new color table. Then use the "Layer redraw" icon Layer redraw to see the updated map in the map display window.


PALSAR in GRASS 6.4

Simple PALSAR image statistics:

  1. To calculate the image statistics (r.univar): in the menu, Raster -> Reports and Statistics -> Univariate raster statistics. Select the map (should be already there) and tick the checkbox "Calculate extended statistics" in the "optional" tab, then "Run". This takes a moment... Find the result in the "output command" tab. Check "mean" and "3rd quartile" and note the values down somewhere. In this example: mean: 352.068, 3rd quartile: 514 etc. Then close the dialog window.
  2. These values could be used to define a threshold filter in r.mapcalc.

Other PALSAR in GRASS GIS example

This example is based on sample data on the PALSAR web site (L1.5 GC product, L150FBS_GC):

PALSAR L1.5 GC product sample scene

Import:

 r.in.gdal PASL1500701140134440805020299_ortho.tif out=PASL1500701140134440805020299_ortho
 g.region rast=PASL1500701140134440805020299_ortho

Outlier reduction (only for visibility purposes!)

 r.univar PASL1500701140134440805020299_ortho
 # .. using 90% percentile:
 r.mapcalc "PASL1500701140134440805020299_ortho_filt  < 416.0, PASL1500701140134440805020299_ortho, null() )"
 r.colors PASL1500701140134440805020299_ortho_filt color=grey
 d.rast PASL1500701140134440805020299_ortho_filt
 d.vect osm_roads
PALSAR L1.5 GC product sample scene imported in GRASS GIS with OpenStreetMap overlay (subscene, center_easting=133.838, center_northing=34.296)

The map shows nicely the fit of the circa 6m resolution PALSAR data and the OpenStreetMap vector lines (compare to Google Maps optical coverage).

Processing ALOS-PALSAR for Landcover Classification

Note: Original process was taken from an ASF MapReady and ENVI tutorial, the objective is to "port" the process using GRASS-GIS

  • Pre-processing and image interpretation
    • Conversion of raw ALOS-PALSAR mosaic image into GeoTIFF.
    • Spatial subset
    • DEM pre-processing
    • Extracting Slope and Aspect from DEM
    • Contrast stretching
    • Edge enhancement - median and laplacian filter
  • Speckle suppresion
    • Antenna pattern correction
    • Speckle filtering - lee, kuan, frost filters
  • Supervised classification
    • Conversion from aplitude to normalized radar data cross section
    • Image slope/indidence angle correction
    • Image classification

Sample data

You may want to get additionally some vector data, for example from Openstreetmap:

See also