Compare S1a and s1b images for change detection

Microwave Toolbox
#1

Hello,

I am trying to find the flooding and the change that is caused by cyclone in the area using sentinel-1.
However, for the event of cyclone two images are available pre-disaster( Sentinal 1 A ascending)
and post-disaster 22 May (S1-B descending).

Can we compare the two images they don’t have the same orbit or ascending and descending?
What operation do I need to do on it before proceeding to next step of flood mask detection?
I want to analyze the area in between the two rivers
Images attached:

image
image900×620 196 KB

#2

Images acquired from different directions (asc/dsc) do not always contain the same scattering mechanisms, because the signal interacts with different surfaces and their shape, orientation ect…

For flood masking, this is of minor importance because the flooded areas are mostly flat and you are searching for a strong decrease of backscatter intensity.

Are these two images before and during the flood? In such case, you can collocate the products to bring both bands into one stack and then use the mask manager to find pixels which have a strong decrease of pixels, for exmaple with this expression:

image1 +5 < image2

This returns true for all pixels which are at least 5 dB lower in image 2.

#3

How to collate them? when i am making a stack the slave image is getting black.
The master image is coming as correct.
image

#4

Please try Raster > Geometric Operations > Collocation

#5

I would say that is not correct. Terrain correction should be used before collocation or stacking.

#6

Based on the images in the first post, I assumed that the products are terrain corrected.

#7

Good point Andreas! :slight_smile:

1 Like
#8

Yes the images are terrain corrected. Is there any standard that we have taken 5 db in this analysis

#9

Could you please specify this question?