Hi all! 🙂 I’m working on a project in ArcGIS Pro 3.6 using Sentinel-2 data to identify localized vegetation stress (specifically mountain pine beetle damage). I'm trying to replicate the exact look of the Copernicus Browser NDVI visualization, where subtle yellow stress spots really show up against the forest green.
I’ve hit a wall getting the contrast right. Here is what I’ve tried so far:
- using a classify approach with 21 intervals and the exact hex codes and breaks used by Sentinel Hub.
- set to bilinear to smooth the transitions.
- verified I'm using 2A data
Has anyone had success with this before? Are there specific stretch type or statistics settings that help highlight these subtle shifts in healthy vs stressed canopy?
Thank you for any insight!
Copernicus Sentinel-2A
ArcGIS Pro 3.6 NDVI
Having the Copernicus NDVI ramp is a great start.
I believe that Copernicus is working with Sentinel 2, NDVI is a float.
In Pro I would compute NDVI as a float raster, use raster functions > NDVI (B8-B4)/(B8+B4). Make sure the output is 32 bit float not 8 bit
Then check the raster properties Right click NDVI layer > Properties > Source > Statistics, check the min/max, if it is something like 0.02-0.85 then this can "break" the Copernicus look
Turn off the DRA and auto statistics
In the symbology pane set Primary symbology to Stretch or classified, Turn off DRA
In Symbology > Statistics, choose from Custom values
Set the min = -0.5 and max = 1.0
Don't recompute statistics from the dataset
You have two options depending on how you applied the symbology
Classified
Method: Defined Interval
Enter breaks manually from the ramp
Each class assign the hex colour
In the colour scheme set interpolation to discrete (not continuous)
Make sure Gamma is 1.0 (no gamma correction
This should give a 1:1 mapping NDVI bin> Copernicus colour
Stretch
Click the colour scheme > Format colour scheme
Add the colour stops at the NDVI values from the ramp and assign the hex code
Colour scheme properties
Interpolation: Continuous
Gamma: 1.0
Symbology > Statistics
From custom values: Min = -0.5 and Max = 1.0
DRA = off
To help pull out the stressed tree data you could take a copy of the NDVI layer
On the copy set the min/max to 0.3 - 0.7
Set a blend mode (multiply or overlay) over the base NDVI or true colour layer.
I hope the above helps or at leasts points to the solution you are aiming for.