In order to produce a correlation plot, you need to have two values (x and y) and plot them to see the level of correlation between these two values. For instance (lets use SENTNINEL-1 Wide Swath image mode as an example), you want to see the correlation between the incidence angle of the wide swath SENNTIENL-1 image and the backscatter value. So, you have two fields, the first one is you incident angle and the second one is your backscatter values. When you perform a linear regression analysis on these two fields, you will see a linear pattern. As the incidence angle increases, the backscatter values of the images decreases.
Hence, what this plot can tell us is that we need to perform incidence angle normalization so that to correct for this linear dependency.
This example is one of many examples you can use for producing correlation plots
I do not know if the example i gave you is understandable.
I hope it makes sense to you