Radar Vegetation Index

siddhartha · December 9, 2020, 10:46am

I am trying to calculate radar vegetation index with dual polarization data (VV-VH). So at first I am calculated DOP or Degree of polarization (VV/VV+VH), and after that value of DOP is multiply with RVI formula. Before multiplying DOP value, I am using square root of DOP. The formula is
sqrt(dop)((4VH)/(VV+VH)). The RVI range should be 0 to 1.

But my final image highest value is 1.5 and lowest value is -439666.75. And my final image content are only agriculture area.

So my concern is, is it possible the lowest value come in minus and highest value >1? Also am I doing correct calculation of RVI?

Please guide me on this topic.

I am referring the following pdf https://custom-scripts.sentinel-hub.com/custom-scripts/sentinel-1/radar_vegetation_index/

ABraun · December 9, 2020, 11:39am

are you using data in linear or log scale? This makes a difference for the formula. Also in the source to the script it is not specified if linear or log data was used.

siddhartha · December 9, 2020, 11:45am

I am using data in linear not log scale. How should I use linear data for RVI?

ABraun · December 9, 2020, 12:06pm

Sigma0 is linear
Sigma0_dB is log scaled

probably the formula works with dB scaled data.

Right-click on the Sigma0 band and select “Linear to/from dB”

siddhartha · December 9, 2020, 12:21pm

Sorry for wrong information, after geometric correction I have done the “Linear to/from db” conversion and then I used “Convert Band”. Then I Export it in Geotiff/Bigtiff.

ABraun · December 9, 2020, 12:23pm

Please calculate the RVI in SNAP before exporting. Chances are high that the data was rescaled during the export process, leading to different input values for the calculation.

siddhartha · December 9, 2020, 12:32pm

Ok. I am trying to do that. In SNAP how to calculate RVI?

ABraun · December 9, 2020, 1:31pm

you can access the raster values with the Band Maths: https://www.youtube.com/watch?v=D4lSjLfPVwY

siddhartha · December 11, 2020, 10:32am

I am calculate the RVI in SNAP but how to check RVI range in SNAP?

After calculating RVI with above described formula, I am exporting the RVI with Two Sigma0_db band and DOP. When image was open in arcgis then value range is showing 1.5 to -3651.56.

ABraun · December 11, 2020, 1:39pm

with the histogram or statistics tool

Don’t export to GeoTiff, simply open the img files inside the .data folder of the BEAM DIMAP product in ArcMap. Conversion of formats potentially rescales the data.

siddhartha · December 16, 2020, 7:41am

As your consulting, I am calculated the RVI without linear to db conversion. But RVI value range is come between 0 to 4 where its suppose to be 0 to 1.

So my consulting is why RVI value range is exceeds 1.

ABraun · December 16, 2020, 9:13am

is your data calibrated to Sigma0?

Please have a look at the histogram of the RVI to see the range of the data.

siddhartha · December 16, 2020, 9:30am

Yes my data is calibrated to Sigma0. I checked the histogram, max value showing 4

ABraun · December 16, 2020, 9:33am

I understand that, but maybe these are outliers caused by double bounce or volume scattering while the majority of the data lies between 0 and 1.

Please show the histograms of both input products (VV and VH) as well as of the RVI.

siddhartha · December 16, 2020, 10:53am

ABraun · December 16, 2020, 12:16pm

I tested with my data and also get some values above 1. The linear scaled RVI looks better than the dB scaled one.

ABraun · December 16, 2020, 1:55pm

this is how I calculated it

DOP = Sigma0_VV / (Sigma0_VV + Sigma0_VH)
RVI  = sqrt(DOP) *((4*(Sigma0_VH))/(Sigma0_VV + Sigma0_VH))

This is the value range
grafik

The thing is, the link you provided does not allow to check if this index really ranges between 0 and 1.
I would just see it as a relative measure, regardless of the scaling.

Maybe you can get in touch with the author.

siddhartha · December 17, 2020, 7:00am

Can you tell me what are the pre-processing steps you followed before calculating RVI? Because this time I am calculating RVI in linear scale but this time I am getting Lowest value in minus and when I am open RVI.img file in arcgis its completely white like this

ABraun · December 17, 2020, 12:37pm

Radiometric calibration to Sigma0 and Terrain Correction.

You have to apply a correct color stretch in ArcGIS, the nodata value outside is not set correctly. Go to the symbology and stretch the colors between 0 and 2 to have better visual contrasts.

siddhartha · December 18, 2020, 9:28am

Now its work correctly. Thank you for your guidance.