Situation: I have merged six Landsat 8 image tiles to one big tile for the bands 4,3 and 2. These are Landsat level 2 products, so already corrected to surface reflectance and need no further processing. I have separately downloaded six band 8 image tiles for the same region. They are of a different date.merged those to one tile too. All the raster images are 16 bit. What options do I have to perform pansharpening on the 432 band combination? Grass Gis apparently only takes 8 bit. Semiautomatic classifications plugin in QGIS requires a metadata file. Of course no one single metadata file applies once the tiles are merged to one. Also, the band 8 is of a different date than the other bands. please advise.
It is not valid to pansharpen floating point (reflectance) using 16-bit data. It is also extremely difficult to correct the pan band to reflectance. The correct workflow would be to use a pansharpening statistic that, more or less, retains the spectral fidelity, operating on the 16-bid data, and then apply the reflectance (TOA) correction. Convolution methods are best for retaining spectral fidelity whereas, PCA methods introduce considerable bias.– Jeffrey EvansApr 1, 2019 at 19:27
Many thanks. Two additional questions in this context: 1.) Is it possible to pansharpen a big raster that has been stiched together from 6 individual landsat tiles? 2.) Must I pansharpen the three input bands individually before I make virtual raster catalogue or can I apply the pansharpen to the output of the virtual raster catalogue (RGB image)?– charlie mikeApr 3, 2019 at 15:01
Since you have QGIS you'll have access to Python and the various GDAL tools and scripts. As of GDAL 2.1 there is a pansharpening Python script available called
gdal_pansharpen.py. If you don't have it already it is avaliable here. You can view the arguments and their descriptions at the official documentation page along with several examples. Additionally, the script is used in this article from Planet.
Another option is Orfeo Toolbox which is an open-source remote sensing toolbox. It contains various tools that can be used from the command line and through QGIS. OTB has its own pansharpening tool which is documented in the OTB CookBook. OTB setup for QGIS is covered here.
Here is nice blog post on pan-sharpening Landsat 8 OLI satellite data in GRASS GIS a productive opensource GIS software.
Alternatively, you may use SAGA GIS wherein multiple options are included to pansharpen data with ease.
These tools can be accessed within QGIS facilitating Pansharpening Landsat data. It is highly recommended to use bands from same acquisition dates to get optimal results regardless of whatever techniques is applied. Otherwise, it is usually necessary to apply a sun elevation correction and an earth–sun distance correction i.e. radiometric correction to input data to compensate changes in scene illumination, atmospheric conditions, viewing geometry, and instrument response characteristics etc.