Intensity Correlation Coefficient

Dear S1TBX people:

I have been looking trough the forum for question related to Intensity Correlation and I have only found threats on the topic posted by me. So I consider maybe important to open a Topic on the subject of Intensity Correlation Coefficient.

I have 3 S1A IW images and each image has the q,i, and Intensity band.

In the intensity band of my images, the values range from 0-thousands. I was told in previous answers that I need to first apply radiometric calibration in order to obtain the true backscattering coefficient which creates a sigma band with values that range between 0-1.

However, to calculate the intensity correlation coefficient, I need to work with the square of the amplitude, so I am assuming this is my original Intensity Band without radiometric calibration. If I want to apply a formula which is basically the same as the coherence estimator, what are the steps that I need to follow?.

In reality I dont think I need to work with the sigma band (or the radiometric calibration), but this is the concept that has me a little confused.

Thank you for any insights.


could you please provide a link to the calculation of the intensity correlation coefficient which you are referring to?

Dear @ABraun:

Here I would like to attach some parts of the text I have read in order to calculate the Intensity correlation coefficient. Taken from the work of Arcinegas, Gustavo in 2005

The formula I want to apply is the one at the buttom. I have found that this formula is similar to the one for the coherence estimator , differing that in the top is the complex coherence multiplied by the conjugate of the same image, and of course the values are complex data not intensity.

S0, as you can see… my question is. Should I use my intensity image? In the work of this author he says " I took the amplitude image and square it, and applied my formula".

in this case I would also use the Intensity but I’m also not sure if calibrating first would maybe improve the result.

Hello, I am trying to apply the aboved formula to my already co-registered images.
However as I need to apply different window sizes i need to export this images and use ERDAS. In SNAP i have the sigma0 for each image and shows values ranging from 0-1 however, when exporting to GeoTiff of BigTIFF the pixels valuey change to 0-255 simple gray scales. How can I export so that i preserve the original pixel values and can conitnue working?

you would need to export to a data format that handles float values correctly.
To my experience, GeoTiff can do this. Maybe you try another raster format or you use the “convert data type” module in advance.

1 Like

Thank you. I wasnt selecting float as output. It worked.
Kind Regards

can the values of sigma0 go from 0-more than 1?

yes. sigma0 can be > 1 (e.g. for strong scatterers in urban areas, windmills at sea, electricity pylons, etc.).