ALOS PALSAR Level 1.1 Processing

Dear Colleagues

I have problem when I process ALOS PALSAR SLC Level1.1. My data got from ESA website. with 2. gigabyte. the problem is I can’t processed my data with same the tutorials parameters available in ESA STEP tutorials pdfs titled (ALOS PALSAR Orthorectification Tutorial) especially in that terrain correction ((SAR Simulation Terrain Correction)) in GCP selection Coarse Registration Window Width/Height the 1024, and RMS Threshold 0.5. Some times the notepad residual GCP results in the end of process with no image in color manipulation window . The another Terrain Correction method (RD Terrain Correction) has been done without problems.

The question is how it these parameters effect on the quality and quantity of images where I’m adopted method for detection the archaeological remains (surface and subsurface).

Note: I’m following the workflow
1- Calibration
2- Deskewing before multilooking based on the advice of Mr. @lveci in the one of the posts in forum.
3- Multilooking
4- Speckle filtering (Refined Lee)
5- SAR Simulation Terrain Correction with same parameters mentioned in Tutorial above)

I waiting for any Advice.

SAR Simulation Terrain Correction can be more precise, but it fails if topography is not very pronounced in your image or if the underlying DEM is not of sufficient quality. But as the initial distribution of GCPs is random, some processes can be successful where others fail. A good way to overcome this is to increase the number of initially created GCPs.
I comparison, Range Doppler Terrain Correction is more robust.

The parameters you list are referring to the selection of GCPs between the simulated SAR image and the ALOS PALSAR image for proper coregistration.

I think you should invest some time in the selection of a suitable filter which prepares the data best for your needs. This is more important than the parameters of terrain correction I’d say.

Thank you so much Mr. ABraun (@ABraun)
Firstly, my area is classified as lowland Area, where the topography between 5-30m. I haven’t GCPs, just entered the default parameter like available tutorial.
Did yo mean that I should try more speckle filtering options for my needs or computing filtered band from Raster menu (Detect Line_Detect Gradient_Smooth_Sharpen_Morphological) filtering?.
Thank you for helpful reply.

Then Range Doppler Terrain Correction is absolutely sufficient for your case. SAR Simulation works best in mountainous environments.

Thanks a lot Mr. ABraun. What about the second part of my ask above. I adopted procedure for detect Subsurface Archaeological remains using ALOSPALSAR 1.1 and ALOSAVNIR2… So do you think that the speckle filtering options and filter band play roles for detect these target.

Real Thanks

I don’t know how these targets look like but it is worth to try different filters and compare their outputs to see which ones fit the best in your case.

That it really what I do, especially with the frequent failed SAR simulation TC process.

Real Thanks

Good, I’ll try it. and inform you about updates.


I want to ask a question about PALSAR processing/rectification, so I think this tittle is suitable… I was reading data descriptions and I am a bit confused.
I have ALOS2 -PALSAR level 1.1 ceos image HH and HV. I have rectified it with similar methodology above: specifically:
1- Calibration
2- Deskewing : not needed
3- Multilooking: not preferred
4- Speckle filtering (Refined Lee)
5- Range Dopler: used 1 sec SRTM (is this SRTM converted to ellipsoit automatically?)
in the end, it is rectified but when I overlap with 2.1 orthorectied PALSAR and other optic images it does not fit perfectly, like shifted a little bit. How can I correct this geometric shift and overlay with other dataset. Any suggestion is welcome…Regards

One of the two products either yours or the Level 2.1 product was probably rectified by a DEM of lower quality which could explain the slight shift.
Do you have a basemap to check which of them is less correct?