Create RGB

How to create a rgb from sigma0_VH and sigma0_VV?

I made average scenes for a period of 3 months. Before that, for all the scenes I did:
2.Speckle filter
3.Terrain correction
4.Linear to DB
Also, I used Slice Assembly for two scenes, when there was a need (right after download scenes).

If I use the following formula, I do not get a rgb image.
Red: Sigma0_VV
Green: Sigma0_VH
Blue: Sigma0_VH-Sigma0_VV

I use Sentinel-1 GRD, VV+VH, IW…

what happens? An error message, a black result, invalid expression?
You can use the […] button behind each color to define the term in a band maths expression.


As VH is usually smaller than VV I also recommend VV-VH instead.

I’m sorry, I did not explain well.
I get the result, but that’s not the natural look of that area.

How to get a picture of a natural look as under picture.

I see that both are different, but I don’t understand what you mean by “natural look”. Can you please post a link to the source of the second image?
Maybe you can adjust the contrasts in the color manipulation tab to get closer to the image shown below.

When I say natural look, I mean the look of this space in real time (the river is blue, meadows are green …) - as if it was painted with a regular camera. Can I get rgb that will be similar to scenes at Google Earth or at least approximately similar?
The picture below is on the following link:

I don’t think this was generated by an RGB composite. The red component comes from the SAR information, green vegetation and blue water bodies were classified in this case.

As microwaves are not related to color it is unlikely that you find an RGB configuration which looks like an optical image. You can, however, play around with the contrast of the three channels to create balanced colors.

Nice example (VV - VH - VV/VH)


Thank you very much!

What is the best way to remove the speckle?
I used the Speckle filter and the average of scenes for a period of 15 days, a 1 month and 3 months. Based on this, I see that the average of scenes removing speckle, but not completely. Is there another way to eliminate speckle?

temporal means are very effective, maybe in combination with Speckle Filters!

If you need them for simple visualization purposes, I can also recommend the Google Earth Engine. You can quickly generate averages over multiple S1 scenes within seconds:

Average of 94 scenes


Nice demonstration:

1 Like

What is the channel schedule for this RGB (R :, G :, B :)?
How did you get this rgb?

all of them are VV polarization

R: 03.11.14
G: 15.11.14
B: 27.11.14

1 Like

I’m new to this :grinning:

Can I do a preprocessing of scenes for a period of 3 months (3.01.2017. - 28.3.2017.) and get a scene(average) so that when I zoom in at the level of the field, I clearly see the boundaries (approximately google earth). Is this possible with these footage?

Step of preprocessing:
2.Apply-Orbit file
3.Thermal Noise Removal
5.Speckle Filter (Filter: Lee Sigma, Windows size:7x7, Sigma:0.9, Target: 3x3)
6.Terrain Correction

Is my result Ok or can better? :thinking:

zoom1 zoom

1 Like

I’d say this is quite good for a three month average. The only thing to test would be the effect of the different multi-temporal filter types. Some enhance finer linear structures while others are more aggressive but create a very smooth image.
The IDAN filter is based on region growing so if you are interested in the forest area you might check it as well.

Okay, I’m testing the IDAN filter. :slight_smile:

You might like to know that the quick-look images included in the S1 products (file quick-look.png) is generated using the DN amplitude values in dual-pol products to give blue water and green land for most products:

  1. Assign RGB channels. Red = co-polarisation amplitude, Green = cross-polarisation amplitude and Blue = Green/Red
  2. Multiply the amplitude of the cross-pol image by 2.3
  3. Multiply the amplitude of the RGB channels: Red: 0.891, Green: 1.209, Blue: 66.300

Since you are using sigma0 rather than DN you will need to adjust the values in steps 2 and 3.


You are degrading the resolution of your result by doing speckle-filtering before averaging in time. Since averaging many scenes gets rid of speckle anyway you should drop the speckle-filter, or filter the average-image afterwards (but that will again degrade the resolution)

1 Like

You need to make a Staker