Jane Meiforth
New Member
Posts:6  
| 31 Jan 2021 03:37 PM |
|
|
Dear team at Harris Geospatial,
I saw in the Envi 5.6 release notes under "Fixed issues" that the "Radiometric calibration of WorldView data was incorrect." (ENVI-72153). Does this mean that the calibration with earlier versions was not correct? What are the new parameters? How big is the difference?
From my understanding Envi was calculating the Gain from the WorldView metadata file (gain = 0.1* AbsCalF actor / FWHM) and the offset was set to 0. However, Digital Globe published adjusting Gain and Offset values for the different years e.g. for 2016: https://dg-cms-uploads-production.s3.amazonaws.com/uploads/document/file/209/ABSRADCAL_FLEET_2016v0_Rel20170606.pdf
Are you now including the Digital Globe adjusting parameters for gain and offset?
I was using ENVI 5.3 with FLAASH for the radiometric and atmospheric correction of a WorldView-2 image for a publication. I noticed that the correction in the coastal blue band often resulted in negative values even in vegetated areas outside shadow and water bodies. Do I need to re-calculate the image with the new parameters?
With kind regards,
Jane Meiforth
|
|
|
|
MariM
Veteran Member
Posts:2396 
| 01 Feb 2021 10:02 AM |
|
|
Hello Jane,
We did update the 'adjust' values in ENVI 5.6 using the document you link to. You can actually view and/or edit these values should they need be updated before a new release. They are stored in the following location and file: C:\Program Files\Harris\ENVI56\resource\filterfuncs\sensor_attributes.json
The previous version had different values set here from a previous document. The differences were fairly small but we do try to update the calibration coefficients when we find they have been updated by the data providers.
|
|
|
|
Jane Meiforth
New Member
Posts:6
| 01 Feb 2021 08:06 PM |
|
|
Dear team at Harris Geospatial,
thank you for the reply! Do you have access to the values for each year? They change over time. It would be good to have the correct values added automatically for the radiometric calibration according to the date of the image. I only found the gain and offset values for 2016 and 2018 for the WorldView-2 satellite images.
I could significantly improve the spectra by adding additional Gain and Offset values provided by Digital Globe, compared to only the values provided in the image metadata that were implemented in Envi 5.3. I have ASD field spectrometer data from the same date as the WorldView-2 image as a reference for comparison. I am happy to send you a screenshot. This seems to solve some of the issues with the high NIR1 values in band 7 and the low values in the coastal band 1.
With kind regards,
Jane Meiforth
|
|
|
|
MariM
Veteran Member
Posts:2396
| 02 Feb 2021 09:56 AM |
|
|
I don't think they were produced for each year. We used the main reports from the data providers with the latest being the 'ABSRADCAL_FLEET_2016v0_Rel20170606.pdf ' which we confirmed with the author were the most up to date.
|
|
|
|
Jane Meiforth
New Member
Posts:6
| 02 Feb 2021 02:22 PM |
|
|
I found this document from Maxar with new calibration data from 2018: https://cdn1-originals.webdamdb.com/13264_102978826?cache=1580226296&response-content-disposition=inline%3Bfilename%3D102953_arc_202001.pdf&Policy=eyJTdGF0ZW1lbnQiOlt7IlJlc291cmNlIjoiaHR0cCo6Ly9jZG4xLW9yaWdpbmFscy53ZWJkYW1kYi5jb20vMTMyNjRfMTAyOTc4ODI2P2NhY2hlPTE1ODAyMjYyOTYmcmVzcG9uc2UtY29udGVudC1kaXNwb3NpdGlvbj1pbmxpbmUlM0JmaWxlbmFtZSUzRDEwMjk1M19hcmNfMjAyMDAxLnBkZiIsIkNvbmRpdGlvbiI6eyJEYXRlTGVzc1RoYW4iOnsiQVdTOkVwb2NoVGltZSI6MjE0NzQxNDQwMH19fV19&Signature=NN5xE4UwQhnbM6FgvAtV8nCprpCbrvSehq1SNsGvdkgJ27NWj1dTlYwubYpROAAe32RBM1ok2rbwgTjetr3rn2Q9NXfpxAk5kNeZiHGSf2mk2Cq0ff94tNu5gMSuyOxpBGmscN1fEwZhhLKxQs2sM4EVHg1I4gcIuwZYMyp3Hmw9bB4JQZAsHovZ9xn9PN8w51QsganDEWkCVPOXu~dO3fgF4L3Vt1ntHLoMW~2JoRV2MOVELv0ef6oZzJldSWEqflx92sgOIujiPkURj4qOVFP60gXumE8S45FEXhivKd6L7q-Mec83I1jhJYorongzepxt0ZM2sGUNpTVD~AgcyQ__&Key-Pair-Id=APKAI2ASI2IOLRFF2RHA
Sorry for the long link... So there seem to be more regular updates on the gain and offset values. Could you check with Maxar about newer calibration results? I asked, but did not get a response. Kind regards, Jane
|
|
|
|
Jane Meiforth
New Member
Posts:6
| 02 Mar 2021 03:11 PM |
|
|
Dear team at Harris Geospatial,
Maxar just confirmed that the 2018 calibration (see my last post) is the most accurate for WorldView-2 acquisitions and should be used for all years - see email below. They also sent the correct wavelengths. Since the 2016 calibration is currently implemented in Envi, it would be more accurate to change to the 2018 values.
With kind regards,
Jane
Customer Support
Tue 3/2/2021 2:47 PM
To: Ted Taylor
Cc: amelia.syafitri@digitalglobe.com; rafael.kargren@digitalglobe.com
Hi Ted,
We do not have a way of monitoring any drift in center wavelength from on-orbit. The Maxar sensors are very stable over their lifetime (<+/-3% so within the noise of the specification for the absolute radiometric calibration). That stability also allows us to assume that a single set of on-orbit calibration will work for the entire archive for each sensor. The 2018v0 calibration is the best calibration we have. It is different from 2016 in that there are improvements in how we perform the calibration (not differences in the sensor response). Please let the customer know that they should just use the 2018v0 GAINs and OFFSETs.
Here are the WorldView-2 convolved center wavelengths:
Pan 651.048
B1 428.584
B2 479.349
B3 548.072
B4 607.778
B5 658.922
B6 723.268
B7 825.064
B8 914.547
Hope the clarifications help. Thank you.
Cheers,
Alvin
|
|
|
|
MariM
Veteran Member
Posts:2396
| 10 Mar 2021 06:00 AM |
|
|
Thank you for the follow-up Jane! We have this on our backlog to update for ENVI.
|
|
|
|