For my PhD research I have four Worldview-2 (Standard2A) multispectral (PAN + 8 MUL bands) satellite images, that were collected between 28th and 31st of October, 2016. Since my research area spreads over all four images, I have to mosaic images in order to produce single image that covers whole area. However, four WV-2 images have significant spectral differences (see picture in the attachment) that I couldn't correct using the standard color correction methods. So I am hoping to solve this problem by applying radiometric correction. My questions in regard to radiometric correction of WV-2 Standard2A satellite imagery are:
1.) What is the best radiometric correction method for WV-2 imagery ?
- I have been using ATCOR radiometric correction in Geomatica 2016 softvare, but I have seen that there are many other radiometric correction methods (e.g. FLAASH in ENVI).
2.) What are the correct gain and offset values for panchromatic and 8 multispectral bands of Worldview-2 imagery ?
- gain and offset values read automatically from WV-2 images (.TIL format) are different then official values given in DigitalGlobe's whitepaper: Absolute Radiometric Calibration: 2016v0.
Thank you for your help!
Fran Domazetović think in three processes separately.The first is the atmospheric correction of the bands; this will remove the effect of the atmosphere on reflectance. In this process the digital numbers are converted into surface reflectance. ATCOR, FLAASH, and DOS, are methods for atmospheric correction and can be applied to images from different sensors, acquisition time or locations. The coefficients gain and offset are given in the metadata.
When all scenes have been corrected, you can then apply the histogram matching technique (second process) to get all images look the same (the same distribution of the pixels) this will not alter the reflectance values .
Finally you can create the mosaic (third), merging all scenes and adjust the visualization.
...What is the best radiometric correction method for WV-2 imagery ?...
I think that you need to carefully check the input parameters to get an accurately atmospheric correction. See https://www.harrisgeospatial.com/portals/0/pdfs/envi/Flaash_Module.pdf#page=31
Cheers
Lizardo Reyna
You can see this:
http://www.pancroma.com/downloads/Radiometric_Use_of_WorldView-2_Imagery.pdf
ATCOR and FLAASH are easy to apply. They are more completed in comparison with DOS. However, the most complete method is 6S, considering the metheorological variables and others that you need. But, if you do not have experience on correction ATCOR and FLAASH are a good alternative. You have some papers doing the comparison between corrections.
Can any of you speak to the appropriate range/datatype of WorldView2 rasters if employing any of these algorithms?
I am using the 6S algorithm on WV2-2A imagery via the i.atcorr module in GRASS GIS. I.atcorr requires the input raster be in at-sensor-radiance or at-sensor reflectance first, which I have done according to DigitalGlobe documention (link below). DG suggests all transformations be done in FLOAT32 datatype; however, i.atcorr requires a specified input/output range that is positive.
Now, I could re-scale to positive values (?), but I am not sure if this would leave me with the WorldView2 range that i.atcorr is seeking... For example, Sentinel 2 L1C are also encoded as UINT16 with 15 significant bits. However, documentation states that the correct i.atcorr Sentinel range is 1-10000; 10000 being a specified QUANTIFICATION_VALUE by Sentinel.
Is there a WorldView 2 equivalent to this quantification value? Even when using the new images converted to radiance? Is there an alternative range specific to WV2? And either way, what would the appropriate output/rescale value be for using i.atcorr with WV2?
Any insight would be greatly appreciated!
*WV2 is encoded in UINT16 with an 11-bit dynamic range
https://dg-cms-uploads-production.s3.amazonaws.com/uploads/document/file/209/ABSRADCAL_FLEET_2016v0_Rel20170606.pdf
https://dg-cms-uploads-production.s3.amazonaws.com/uploads/document/file/209/ABSRADCAL_FLEET_2016v0_Rel20170606.pdf
Fran Domazetović, how did you resolve this issue?
I am thinking of writing a script to read the metadata of the WV2 then use the equations provided by MAXAR in the attached pdf to calculate TOA. Since most software I've tried are not reading the DigitalGlobe product metadata.
I've tried using the Orfeo Toolbox the absCalFactor, and the effectivebandwidth cannot be read, the same problem for GeoEye. Pleiades read successfully in OTB.
Adebowale Daniel Adebayo at the end I haven` manage to solve this problem, regardless of any radiometric correction that I have used differences between WV-2 images were too obvious after mosaicking them.
But using the ATCOR in Geomatica and then manual mosaicking I have manage to merge smaller images with larger ones, and as a result from 4 separate images I got two larger WV-2 images of whole island (284 km2). At the end I have processed and classified every created WV-2 image separately, using the identical classification methodology.
And Geomatica is able to read the DigitalGlobe product metadata, and I find it really easy to use...
Fran Domazetović, thanks for your response.
I resorted to writing the script as mentioned earlier because the application is quite specific, and all I needed was just the WV-2 atmospheric reflectance.
- Badges
- Science method
- Similar topics
- Biological Science
- Ecology