Multi polys, script download, clip details

EO Products / Scripts
1

I registered on the EO Browser and downloaded a S2 analytical image for a poly. I’ve few questions…

  1. Can I upload multiple disconnected polys to get their clipped images? So how will it play out if the polys and S2 images span multiple UTM zones?
  2. Can I customize with a script to download a time-series of different image types (L8, S2, MODIS)? Ideally, I want to use a script and upload multi-part polys to download multiple analytical image types over a time range (think like download L8, S2, MODIS images of NIR, NDVI, NDWI, true color, false color for US, Canada, India, UK, Australia for Apr-Oct 2014-2017).
  3. Is it possible to customize the names of downloaded clip images?
  4. I noticed that my downloaded S2 clip wasn’t exactly 10x10 m2, why is that?
  5. What do 3 bands of the analytical NDVI mean?
  6. If I download “analytical” true and false color images, are these composites made from 32-bit bands? As these composites are purely for visual purposes, how I can I specify just the true and false color image downloads as “visual”.
  7. Is there pansharp false color for Landsat-8?
  8. How to get MODIS NDVI?
2

  1. EO Browser does currently support one polygon only. However, the back-end, Sentinel Hub services, do support multi-polygon. I suggest you check Python libraries and the this FAQ for more information.
    For “multiple UTM zones” - this should not be a problem. You have to define, which coordinate system you want to get data in and the result will be reprojected to that CRS.

  2. I suggest you check our time-lapse python script, which should provide you relevant examples:
    https://github.com/sentinel-hub/time-lapse

  3. Add a parameter “NICENAME” to the call, e.g. “NICENAME=image-2018-04-11-MODIS.jpg”.

  4. You might not have set the proper resolution setting? Or the area that you wanted to download, was too large. Maximum download size is 5000x5000 px.
    Generally you can define resolution by using “RESX=10m&RESY=10m” parameters instead of WIDTH and HEIGHT.

2 Likes
3

Thanks Grega, I edited my questions. Please edit your answer.

I like the “nicename” but can I format it more unique by adding path-row ID (L8) or Tile ID (S2) as “sensor + ID + date”. Can I customize the date as 12Apr2018?

Can you provide code to create pansharp true and pansharp false color for Landsat-8? Also how does the pansharp script change if one is to use reflectance data?

4

Analytical export in EO Browser provides option to download in GeoTiff, different resolutions, etc. It does still export the data as it is configured.
3 values in NDVI refer to red, green and blue values of the image being generated. If you want to get actual NDVI data, you should configure your own instance to return actual values.
See:
https://www.sentinel-hub.com/faq/how-do-i-get-actual-ndvi-values
https://www.sentinel-hub.com/faq/how-create-eo-product

I am not sure I understand the question correctly. The source Sentinel-2 data (for example) are 15 bit. We use these original values when creating composites.
The output of the product is based on your selection - either 8bit, 16 bit or 32bit.
For visual inspection i recommend you choose 8bit.

Landsat-8 pansharpening band is applicable only to red, green and blue bands. So applying it to NIR, which is used in false color, would not produce scientifically sound results.
You can do it though if you want:
https://www.sentinel-hub.com/faq/how-do-i-manually-pansharpen-specific-band-eg-red

You need to go to Configuration utility, create a new instance, e.g. “MODIS” and set-up “actual NDVI value” layer as described in answer 5 above.

2 Likes
5

Adding to 7, there are some ideas however on using pan-sharpening with NIR. See here: http://www.mdpi.com/2220-9964/3/2/507

2 Likes
6

For L8 pansharp, I meant (see: https://goo.gl/dti0cO):
Pansharp true: B04 ®, B03 (G), B02 (B), B08 (Pan)
Pansharp false: B05 (NIR), B04 ®, B03 (G), B08 (Pan)

7

Sure, that is very easy. You simply create a product with the following Custom script:
return [B04,B03,B02,B08]
or
return [B05,B04,B03,B08];

1 Like