Processing an entire GRD swath with gpt

Microwave Toolbox
1

Hello. Does anyone ever pre processed an entire GRD swath without subsetting it? Do you have any recommendation?
Thank you

2

Do you mean prcessing all the bursts of a SLC subswath ?

3

Hello,
I’m working with GRD, and I have significant difficulties, even with more powerfull PC in processing the entire image and not a smaller subset of it. Even if I split the graph in more parts.

I’m using the GPT tool from batch, in a Python script which also automatically download images and then, after the preprocessing, compute some final products.

I will have to operationally perform these procedure for a large area (the Italian Alps) and so I’m trying to figure out if there is a way to process large data with SNAP, and if possible which is the best way to do that.

Thank you

4

@jun_lu could you help?

5

Could you provide more details (e.g. the processing graph, products and parameters used etc.) regarding your processing?

6

Yes. I perform the following pre-procesing steps:

  • Orbit file application
  • Radiometric correction → beta_nought
  • Range-doppler terrain correction → gamma_nought
  • Geometric terrain correction
  • Speckle filtering
  • Convertion in DB
    For now I’m using all the dafault parameters of SNAP, and the DEM used is the Coperniscus30. It works fine for subset of the entire image, even of 40x40 km
7

Batch processing of a graph generally consumes a lot of memory especially for large number of products. One thing you can try is to separate the graph into two sub-graphs: one graph for all processing before terrain correction; one for processing start from terrain correction. Once the batch processing of the first graph is done, close SNAP and start it again. Hope it helps.

1 Like