X
PrevPrev Go to previous topic
NextNext Go to next topic
Last Post 07 Mar 2015 11:48 AM by  anon
Processing order for ASTER L1A Imagery
 1 Replies
Sort:
You are not authorized to post a reply.
Author Messages

anon



New Member


Posts:
New Member


--
07 Mar 2015 11:48 AM
    Hi there, I have 14 ASTER LIA Images that I need to process before I can start analyzing them for mineral spectra. However, literature for processing LIA is rare and I find myself perplexed by the order of processing steps to achieve well corrected and calibrated imagery. There are some striping issues with some of the images - the TIR I have managed to destripe using ERDAS software, which was much more robust than what is offered in ENVI, from what I can see. Three SWIR images I will entirely discard. Otherwise striping is not an issue. Radiometrically correcting the VNIR and SWIR appears simple using the Radiometric Calibration tool. But what do I do for TIR? Do I need to do anything? Geometric correction is giving me a headache and I'm not sure if I'm supposed to do it before or after radiometric correction. My area is NE Banks Island in the Canadian arctic archipelago - so I have found an orthorectified Landsat7 mosaic and subsetted it using an ROI. This I have used in image to image registration, where I automatically generated tie-points with an RMS error of less than 2. However, registering my ASTER imagery warps it, and I can therefore not use it in any of the orthorectification processes. This really confuses me. I need GCPs to obtain a workably accurate orthorectified end image, but I can't get GCPs without warping the image first. What should my order be here?? I assume any atmospheric correction should be done last. I have read that for ASTER I should be using FLAASH, but that is an extension that my current computer doesn't have so will figure that out later on. Sorry for the long missive - I guess my real question is: What order to I process the image in? Destriping - Radiometric - Geometric - Atmospheric? Or Geometric - Radiometric - Destriping - Atmospheric? And what order do I do things in geometric correction? Registration - GCP creation - Orthorectification? And is this the same for all the instruments? Thanks in advance for any help. Anne

    MariM



    Veteran Member


    Posts:2396
    Veteran Member


    --
    09 Mar 2015 09:12 AM
    ASTER 1A data in HDF format comes with sensor information that allows ENVI to open the dataset with RPCs (rational polynomial coefficients) that can be used to orthorectify the data. I would say that typically a radiometric calibration is performed first and then orthorectification. To process 1A data, I would do the following: - open the 1A dataset as EOS->ASTER. The 4 separate datasets (VNIR + 3b, SWIR, TIR) should open and the map coordinates should show "RPC Geographic Lat/Lon' - first run Radiometric Calibration since you need to extract information from the HDF to perform the calibration. You can calibrate all datasets here. The TIR bands can be calibrated to TOA radiance. VNIR and SWIR can be calibrated to TOA radiance or reflectance. The processed datasets will retain the RPCs. - use RPC Orthorectification workflow to orthorectify the datasets to a standard map projection. You can include GCPs here to improve the RPC model. If some of the datasets do not align after orthorectification, you can use the Image registration workflow to perform an image-to-image registration. - Atmospheric correction for VNIR and SWIR can be accomplished through our ACM Module (FLAASH and Quac), or by included methods such as empirical line calibration, flat field, IARR, dark subtraction, log residuals. ENVI has the Thermal Atmospheric Correction tool for thermal datasets. Then the emissivity tools can be used to convert to temperature. These are all located under the Radiometric Calibration tools in ENVI 5.2. There are no tools specific to removing striping from ASTER data. If you are referring to the stagger issue in ASTER TIR bands, it looks to be compensated for with preprocessing performed as ERSDAC: http://www.science.aster.ersdac.jspac... This description states there is a correction applied and that it is good to +/-4 pixels, so you may still see offsets.
    You are not authorized to post a reply.