Creating Radar Vegetation Index


#1

Hi,

I would like to calculate Radar Vegetation Index using Sentinel-1 data. The formula is: 8HV/(HH+VV+2HV) (all of them are sigma).
Sentinel-1 doesn’t have HH polarization. So, I have a question - how to calculate such index using Sentinel-1? Is there some modified formula may me or something?


For preprocessing the SENTINEL 1 data, is the products should be subset
Layer/Band Stack
#2

the RVI needs quad-polarized data so it is not possible.

This publication argues that one option could be having HH/HV data alone as HH and VV are related:


Source: https://www.researchgate.net/publication/267020154_Radar_Vegetation_Index_as_an_Alternative_to_NDVI_for_Monitoring_of_Soyabean_and_Cotton

But as most of the Sentinel-1 data is VV/VH this isn’t an option either.


Agricultural crops height variation monitoring using SENTINEL 1 DATA
Help documentation for operator Polarimetric Parameters
#3

Thank you!
I actually did VV/VH and it works very well for me for an Ascending orbit and not so well for Descending. So I fought, to try something else may be.


#4

Doe anyone have an idea how this formula should look like if i have already db units? I first thought replacing the division by a subtraction would be correct according to the log laws but the result was strange.


#5

You can use formula 1/db. It would be a conversion from db to linear units


#6

you mean like that?


#7

Sorry, I’ve little bit misinformed you. I used that formula to convert from linear units to ratio (how many times backscatter is weaker than initial signal). I did 1/linear units.
For db I used a little bit more complicated formula to convert it to the ratio mentioned above
=10^(ABS(Value in db)/10)
I believe that this formula is good, but the results are slightly different from 1/linear units.
In any case I would like to use linear units which SNAP provides. It looks easier than tricky works wiht dBs.


#8

alright, thanks for clarification!

So it would be that?


#9

Yes, it would be like you’ve written.
If Sentinel-1 had HH polarization, I would use RVI formula too.
May be it would be useful for you, we used VV/VH polarization ratio formula for our researches. And after that, we found a good relationship with vegetation indices.


#10

My goal is to see if there are any differences between the RVI formula based on

  • HH/HV
  • VV/VH
  • Full Polarization

#11

It’s very interesting! Would it be possible to read about your results?


#12

I’ll keep you informed as soon as I have something to show.


#13

Thank you!:slight_smile:


#14

Hi,

Would you like to share, do you have any results obtained?


#15

didn’t work on it since then due to other priorities. But it is planned for the upcoming months.


#16

Thank for an answer! Please let me know when you will recieve any results.


#17

Hi,
I would like to know that still, it’s not possible to calculate Radar Vegetation Index using Sentinel-1 data?


#18

a dual-pol alternative is suggested here: 267020154_Radar_Vegetation_Index_as_an_Alternative_to_NDVI_for_Monitoring_of_Soyabean_and_Cotton


#19

Thank you for your prompt reply.
According to the above-mentioned paper where it assumed that ϭHH ≈ ϭVV and since Sentinel-1 has VH and VV polarizations so can we modify the equation as follows?

RVIᵥᵥ=(4σ⁰VH)/(σ⁰VV+σ⁰VH)


#20

Yes, this is an option. Different scatterers might be targeted by VV but in general, the index is about the contribution of volume scattering which is indicated by cross-polarized response.