OpenStreetMap

Copernicus Sentinel's TCI

Posted by AkuAnakTimur on 3 March 2017 in English. Last updated on 4 March 2017.

… which stands for True Colour Imagery.

It’s really wonderful! I used to download all separate three visible bands and took significant time to make them into RGB ones. Because I lack the hardware and skills to georeference them properly using suitable applications, I had to rely on Mapwarper(‘s precious bandwidth and storage). Very pleased with that: now I could use it, like, almost in an instant.

Now I can’t wait for the launch of the second Sentinel 2 satellite l̶a̶u̶n̶c̶h̶, and I’m hoping it will be a̶ successful l̶a̶u̶n̶c̶h̶.

EDIT: OMG, my English; or what Malaysians would say, “oh my English”

Discussion

Comment from BushmanK on 3 March 2017 at 17:57

Could you share a link to it? There are so many places where Sentinel 2 data is available for download, so it is hard to keep track on them all.

Comment from ff5722 on 3 March 2017 at 20:46

Using ESA’s software and uploading the exported geotiff to Mapbox studio, it’s a fairly easy process now. Open zip, generate image, adjust colours, export, upload and use. The most annoying part is finding the right tile (location, cloud coverage and tile coverage).

https://wiki.openstreetmap.org/wiki/User:Ff5722/Using_Sentinel-2_imagery

Comment from BushmanK on 3 March 2017 at 21:56

This interface might be easier to use for downloading. https://landsatlook.usgs.gov/sentinel2/viewer.html Basically, it’s an easier to use front-end to USGS EarthExplorer.

SNAP is supposed to be easy to use, however, there are numerous reports of an abnormal behavior of this software (I mean, it just doesn’t work, lacks some components and so on).

Another alternative to Mapbox Studio is Nextgis.com that allows you to create web maps and data sources accessible via WMS.

Comment from AkuAnakTimur on 4 March 2017 at 03:29

I happen to stick with this one: http://sentinel-pds.s3-website.eu-central-1.amazonaws.com/image-browser

Best bit of this web app is that you could look for Landsat 8 imageries as well, if that is your cup of tea.

Comment from AkuAnakTimur on 4 March 2017 at 03:34

Oh, I forgot one more thing: you could even visualise (preview) your chosen images (Sentinel-2 only) up to its native resolution.

Comment from BushmanK on 4 March 2017 at 03:43

Thank you. So, now we have infrared composite from USGS and real color from ESA. That’s pretty cool.

Comment from BushmanK on 4 March 2017 at 03:45

They, indeed, have a WMS service with some sophisticated custom filtering parameters, but that’s nearly unusable for the purpose of mapping.

Comment from AkuAnakTimur on 4 March 2017 at 04:15

WMS service

I thought it’s a paid service provided by Sinergise. But, anyway, I’m happy with the provided TCI files. Download a copy, then trim them in QGIS and save the desired chunk as geoTIFF files. Finally, load it through JOSM using the ImportImagePlugin.

Comment from AkuAnakTimur on 4 March 2017 at 04:19

However, I still need some guide on how to create an RGB Landsat 8 imagery from the three raw files. Tried to Google some guides and carefully follow them step by step, but no fruitful results.

Comment from BushmanK on 4 March 2017 at 04:38

There are several WMS services, but messing with parameters makes it unusable anyway.

For Landsat, you could try this https://fromgistors.blogspot.com/p/semi-automatic-classification-plugin.html Common problem with Landsat and Sentinel raw data is that it has 16 bit per sample. So, there is no universal way of dealing with histogram stretch to turn 16 bit to 8. You can always use a very rough approach of clipping about 2% of usable data on both ends, but certain scenes will become completely defective after that (blown highlights, lost bright details). To fiddle with black/white point of each channel you have to actually understand the theory behind it. These two links telling some stuff about that (ImageMagick is for regular pictures, but if you know how to use that, you can deal with Landsat too) http://www.imagemagick.org/Usage/color_mods/#levels http://www.imagemagick.org/Usage/color_mods/#histogram

Comment from AkuAnakTimur on 4 March 2017 at 04:49

I have worked with ImageMagick before, usually fiddling the images (levels, curves, etc) with Photoshop after that. As a consequence, the georeference elements is gone, going that way.

So, after creating a composite, I would need to georeference it again. Before I upgraded my machine, I don’t have any other options, so I have to rely on Mapwarper and for example, the provided metadata for Sentinel-2 imageries.

And so I managed to load it through JOSM but with a major caveat: I have to check its offset, at almost every places.

By the way, thanks for your link. I’ll definitely check it later.

Comment from BushmanK on 4 March 2017 at 05:05

Using listgeo and geotifcp utilities from LibGeoTIFF you can dump all metadata from any GeoTIFF file and bring in back from dump after editing. I suppose, these two utilities should be a part of FWTools package http://fwtools.maptools.org/ and OSGeo package https://live.osgeo.org/en/index.html https://trac.osgeo.org/osgeo4w/

Log in to leave a comment