Subscribe to this thread
Home - General / All posts - Create Multi-Channel Image from Sentinel 2 JP2 Files
joebocop
389 post(s)
#31-Oct-19 17:49

I believe this has been answered, I just can't seem to search my way to it.

I have imported Sentinel 2 jp2 files into a Release 9 .map file. Each is a single channel image component. I would like to create a single 3-channel image component from some of these images.

Is the workflow, generally, to create a new empty table having a <tile> column, and then INSERT INTO that table from each of the source image components?

If I'm only interested in viewing (and not performing analysis on) these data, and instead of importing the JP2 files I had linked them into a map file, is there a non-import/non-insert way to visualize the linked JP2 files as a single multi-channel image within Release 9? Within Viewer 9 even?

Thanks for pointing me in the correct direction on this one.

Dimitri


5,598 post(s)
#01-Nov-19 02:53

If I understand the requirement (could be wrong), I'm guessing that the Example: Import BIL and Combine 3 Bands topic shows one way to do it. That's a somewhat manual, albeit perfectly workable, way to do it for now. I expect in the not too distant future that raster features will expand to provide more point-and-click assembly of multichannel images from single-band images.

is there a non-import/non-insert way to visualize the linked JP2 files as a single multi-channel image within Release 9? Within Viewer 9 even?

Yes, using the approach in that topic you'd just write a query that assembled, on the fly, a multiband image from single band images that were linked. Could do that in Viewer. Save the query into a project using 9 and then users with Viewer could link images and run the query, making adjustments to the image names as appropriate.

Dimitri


5,598 post(s)
#01-Nov-19 11:37

To follow up: I re-read the topic recommended and the general idea for constructing a three-band image from three one-band images is good, but that particular topic uses a case where each of the three images is a different tile field in the same table. If you import three separate images so their tiles are in three separate tables you'll need to make some adjustments. I have to admit to being curious enough to want to try it. :-)

Do you have a link to a source for the Sentinel 2 jp2 files?

joebocop
389 post(s)
#01-Nov-19 19:09

Thank you!

The technique using TileChannelsConcat() demonstrated in Example: Import BIL and Combine 3 Bands must be what I'm after.

We are investigating workflows to assist in detection of various forest pest species, and as a first effort are trying to correlate this past season's field observations with Manifold-processed Sentinel L1C data captured on the same date.

Here is an example visualization of the NDVI combination of Sentinel bands on the date (and in the location) in question, using the EO Browser application:

https://apps.sentinel-hub.com/eo-browser/?lat=64.32173&lng=-140.12581&zoom=13&time=2019-07-18&preset=3_NDVI&datasource=Sentinel-2%20L1C

I have downloaded those L1C data (~700mb zip from https://scihub.copernicus.eu/), and have shared them on my Google Drive here.

You will find each band's lossless JP2 compressed file within the European standard "SAFE" data format container as depicted here (the TCI file is a true-colour 3-channel assembly of bands 4,3,2)

For interest's sake, a concise description of the Sentinel missions and their data is at https://eox.at/2015/12/understanding-sentinel-2-satellite-data/

Beyond pest detection, I am separately interested in automating forest fire growth detection. Sentinel revisit frequency is quite good at our higher latitude, around 3 days or so, and could be useful in automatically (and quite accurately) calculating burn extents, smoke prevalence, etc. That could be an interesting script to ship in Viewer where a paid SentinelHub API key could provide the most current imagery.

Attachments:
arc_dir_bands.PNG

Manifold User Community Use Agreement Copyright (C) 2007-2019 Manifold Software Limited. All rights reserved.