Canopy/vegetation height

s1tbx
#1

Dear all,

I am rather new in SAR topic but I interesting in obtaining canopy/vegetation height using Sentinel 1 and SNAP.
Could someone provide me with help/steps how to do that?
I would be very thankful.
Best,
Stan

Predict above ground biomass from Sentinel 1
How to estimate Aboveground biomass using Sentinel-1 Data?
#2

You would need to look into PolInSAR methods using RVoG or coherence models.
These aren’t available as tools in SNAP but they could be developed.

Look for papers by Konstantinos Papathanassiou and/or Shane Cloude.

#3

There might be some tools in PolSARPro.

#4

Thanks for reply.
I was thinking about a way how to take the absolute height and subtract the DEM (difference would be the canopy height).
As I understood from the tutorials this is feasible in SNAP, right?

Best,
Stan

#5

This is not easy because of two reasons:

  • Most DEMs you get (such as SRTM) already include the top of the canopy. They are rather surface models than terrain models.
  • To derive elevations from SAR interferometry you need high coherence values (as a measure of quality). Unfortunately, coherence is worst within vegetated areas.
#6

Thanks a lot for the reply.
But lets assume I have an external,highly precise DEM derived from Lidar or contour lines.
Is it then possible to extract somehow the canopy tops?

#7

If you would have LIDAR data you could directly extract the canopies directly by comparing first and last pulse.

If you had a digital topographic model which fully excludes forest cover you could theoretically use InSAR for the estimation of heights. But as already mentioned, your result won’t be very good above forest areas.

It also depends on the wavelength of your sensor: While X-band signals backscatter from the crowns and leafs of the canopy L-band waves, for example, penetrate these structures and interact with the stems and larger branches.

#8

Thanks for reply.
My attempt was here is to compare the LIDAR canopies (DSM) with what I can get from SAR when Iam using the LIDAR DEM (to subtract the SAR).
So, assuming to use the X-band from Sentinel 1 can you briefly provide me some step-by examples how to get the DSM?

#9

Sentinel-1 is C-band with a wavelength of about 5 cm.

Extracting elevations from SAR is not trivial but this forum is full of examples and similar challenges.
It is basically having two images, coregistering (=stacking) them, calculating the phase difference (interferogram), unwrapping it and convert the unwrapped interferogram to heights.

These two may help you for the first steps:
http://sentinel1.s3.amazonaws.com/docs/S1TBX%20Stripmap%20Interferometry%20with%20Sentinel-1%20Tutorial.pdf
http://sentinel1.s3.amazonaws.com/docs/S1TBX%20TOPSAR%20Interferometry%20with%20Sentinel-1%20Tutorial.pdf

But most importantly: Don’t just follow steps, try to understand the reasons behind it. You can find a nice overview about interferometry here:

#10

You can try it but it’s unlikely to work (it was found to be unreliable with ERS-1/2 tandem data with 1-day repeat):

http://ieeexplore.ieee.org/document/1386491/?arnumber=1386491&tag=1

1 Like
#11

good reference regarding the potential.
For those without IEEE license:
http://earth.esa.int/fringe03/proceedings/papers/47_santoro.pdf (mabye slightly different)

#12

If using non-Sentinel-1 images is an option then a recent method that has shown good results is http://www.mdpi.com/2072-4292/8/9/700. They use the interferometric coherence of TerraSAR-X – TanDEM-X tandem interferograms to estimate canopy height.

#13

Hi,
I would like to know how can i obtain canopy height with ALOS2 and every software?

#16

This is a very challenging task because it requires utilization of both interferometric and polarimetric information of the SAR image. You would probably need several quad-pol datasets to achieve this.

Maybe these tutorials give you a first insight on how it works: