ALOS PRISM CEOS Terrain Correction problems

Hi,

The first thing to mention is that you cannot do terrain correction (a SAR process) on ALOS Prism data because Prism is optical data, not SAR. We do not yet support orthorectification of optical images. That said, I think our software should exit more gracefully with a more explanatory error message! I'll fix that!

The next error, the inability to export the DEM to a GeoTIFF has me wondering though ...Are you starting with the DEM in ASF Internal format (.img and .meta files)? I tried importing and exporting an SRTM DEM here and wasn't able to reproduce the error, so I suppose I would need a bit more information from you on how to reproduce your steps. It looks like you are getting the equivalent of a segmentation fault (the code trying to read or write to memory locations that it's not supposed to), so I'd like to look into that. If nothing else, maybe you can provide us with your DEM file that you tried to export?

Finally, the third issue is for you to produce 1B2G level data. The only difference between 1B2R and 1B2G is that the version ending in 'G' is geocoded. You should be able to geocode your Prism data with no problems by just browsing to your input file (pick the LED- file) and entering your desired settings on the Geocode and Export tabs. If you just skip the Radiometry and DEM tabs, then the Convert tool will not try to terrain correct your data and all should work fine.

Brian

Hi,

I'll answer your questions in the order asked:

- Orthorectification of optical images?

It has been discussed now and then, but historically the radar world has kept us pretty busy. Now that we support ALOS Avnir and Prism optical data, the topic may come up again but I can't say for sure.

- Import/Export?

I misunderstood.

- Importing Seamless Server SRTM DEMs?

Apparently my SRTM DEM had some differences than yours. When importing yours, I experienced a segmentation fault as well. We used to use separate importers (code) for various flavors of GeoTIFFs but now we are migrating everything into a single 'generic' GeoTIFF importer ...a part of the newer code (somewhat obviously) had a bug in it. It has now been fixed and will be available in the next beta or release version of the software, whichever occurs first. Thanks for pointing out the bug to us!

=> On our FTP server, you will find the imported version of your DEM. I placed it into the incoming folder under the following name:

N53E82-84g_imported.zip

When you do terrain correction and wish to use the N53E82-84g DEM, just browse to the .img version of the file instead of the TIFF and it should work fine. NOTE: Many of the Seamless Server DEMs have a lot of negative values in them ...and yours is no exception. These negative values are typically less than -900.0 and indicate "no data" for the pixels which have that value. Your file uses -32767.0 as the "no data" value. As a result, viewers may have trouble with viewing this file in a way that lets you see the topology nicely, but the ASF terrain correction already knows all about Seamless Server DEMs and automatically screens out the negative "no data" values so the terrain correction can occur properly.

=> If you have other DEMs that may be subject to the same bug, note that you can either import them with the current stable release (v3.1.16) or put them on the ASF ftp site for us to process, i.e. as I did for the N53E82-84g DEM. You can have both the stable release and the beta release installed at the same time by the way.

Regards,
Brian

Cihan,

uysalc wrote:I am a master student at Istanbul Tech. Univ. in Turkey. I have a thesis about generating DEM from ALOS Prism triplet mode. I have a ALOS PRISM prism triplet(ob1) dataset. Its level is 1B2R. Also I have PCI Geomatica 10.1 software.
I want to extract DEM from this dataset.Also I have GCP file(shape file format received from IKONOS at UTM projection and WGS84 datum).

PCI Orthoengine has the ability to generate a Digital Surface Model (DSM) from PRISM data. It can be a bit tricky to use, but does generate a reasonably good DSM. PCI recommends use of the L1B1 product as this is what their software is written for. A colleague of mine has been able to generate DSMs from the L1B2R products also, however.

Quote:PROBLEMS:
1)
I have triplet images (backward,nadir,foreword) but software doesnt support this three images at same time.I want to use all (backward,nadir,foreward) to generate one stereo image. what can ı do?

I am not aware of any software that deals with the three data sets of a triplet at the same time. Most packages are written to deal with stereo pairs. So I believe you need to work with your data set in pairs and combine the results later.

Quote:2) how many gcp is necessary for it? and extra reference DEM is necessary for it?

The exact number of your GCPs is not as relevant as the quality of them. You can make it work with very few.

Quote:3)Do you have any material about generating DEM from PRISM at PCI Geomatica 10.1
note:3 images have img and vol. files and they are ceos format.

I don't have any material. I did not get the time to have a closer look at the various software packages available. I have a background in photogrammetry, so I am certainly familiar with the concepts. I hope to have a chance to look into this some more.

I hope this answers your questions.

Cheers,
Rudi

Hi
I am also given a new task to extract DEM form prism tripplet using Geomatica 10.0 or using LPS. I have viewed all discussions in this forum and feel it better to use geomatica. Now I have a confusion to purcahase ALOS PRISM data. It is clear I must go to level IB1 but the problem is that in alos description guide I have find two terms i.e Scene ID and Uncorrected Scene ID. I could not distinguish between these two, please tell me the difference between these two and also tell me what I will need for DEM extraction simple or uncorrected and also please inform me does geomatica 10.0 support CEOS format or I have to purchase data in different format e.g geotiff etc

thanks
Hamdam