Preprocessing S1A and S1B

Hello everyone :slight_smile:

I downloaded the scenes of Sentinel -1 for the entire 2017 year with the following characteristics:
Satellite platform: S1A and S1B
Product type: GRD
Sensor mode: IW
Polarisation: VV+VH
Order by: Descending
Relative orbit number: 153

I have 49 scenes with these characteristic. I need to do the preprocessing of these scenes and eliminate the speckle by averaging.

I created Graph with the next steps:

1.Apply-Orbit file(Add/Radar/Apply-Orbit file)
2.Thermal Noise Removal (Add/Radar/Radiometric/Thermal Noise Removal)
3.Calibration (Add/Radar/Radiometric/Calibration)
4.Terrain Correction (Add/Radar/Geometric/Terrain Correction/ Terran-Correction
5.LineartoFromDB (Add/Raster/Data Conversion/LineartoFromdB)

After Batch processing i did these steps:

1.Create Stack
2.Stack Averaging

Also, i used Slice Assembly and Subset before batch processing. I used Slice Assembly, because the area of interest was not always covered by one scene.

Is this all right? Do I need to take another step to get better results?
Can I use the same preprocessing for S1A and S1B?
Is there a presentation where preprocessing is explained.

You should do LineartoDB after stack averaging to get correct results. Otherwise your processing-chain looks good. How does the average scene look like?

Did I understand you well?
You think that LineartoFromDB work 2x: in the processing chain and after stack averaging. Do you think it’s just done after stack averaging?

I work average of scenes for a period of 3 month, 1 month and 15 days. I’ll do average of all 49 scenes for 2017… When i’m done, i’ll present result.

My goal is to show how speckle is reduce by averaging.

I think that the point is that averaging values in log form (db) is mathematically not the best idea. Therefore, you first create the average based on the original data, and then only convert the average image to db.

1 Like

Exactly, averaging in dB gives incorrect results.

1 Like