I would like to ask about the basic pre-processing steps applied on a S-1 SLC product. So far I have been working with GRDH images only. The objective is to see whether the image captures specific plastic targets in a coastal region (they have an area of 100 sq. meters each). So firstly, get as clear an image as possible in order to make a visual check and then run marine object detection algorithms.
The ROI is well captured in one subswath of the image which I have subset (TOP split), but which operations are appropriate form now on, apart from radiometric calibration? Most tutorials I found so far focus on interferometry so I cannot tell the steps that are appropriate for this application. Any help is welcome.
interferometry above water areas seems not logic to me because the surfaces change too rapidly.
Have you seen that there is a Ship detection module in SNAP? It works on GRD data but you can select the minimum/maximum size of the detection objects and a few background windows sizes based on the CFAR algorithm. https://en.wikipedia.org/wiki/Constant_false_alarm_rate
However, note that the resolution of Sentinel-1 might be too low for plastic in the ocean: https://sentinel.esa.int/web/sentinel/missions/sentinel-1/instrument-payload/resolution-swath
Maybe there are S1 Wave Mode data in your area?
Exactly, interferometry has nothing to do with this, thus I can’t get enough help from the tutorials The object detection will be run in SUMO which also ingests SLC products. I am aware that the area is small for S-1 but want to try it anyway. Thing is , what is the appropriate pre-processing? For visually inspecting the image.
(SUMO doesn’t require radiometric calibration, only geocoding which means Ellipsoid correction if I am not mistaken)
And what about Wave Mode? Can you please specify? There might be wave mode data, will have to check…
the same as suggested here: Radiometric & Geometric Correction Workflow
No Terrain Correction is required (or possible) over water so you use the Ellipsoid correction.
the Wave mode has a higher spatial resolution but is not available for most parts of the earth: https://sentinel.esa.int/web/sentinel/user-guides/sentinel-1-sar/acquisition-modes
Is there a reason why you want to stick to SLC data? Both SUMO and SNAP’s ship detection make use of GRD and you can skip many steps to create the ground range data.
I read that the spatial resolution is finer for the SLC product (don’t have the reference at hand but can look it up if needed). So I thought of trying SLC as well. At GRDH, targets are not visible or detected by either SUMO or SNAP.
the initial resolution is higher for SLC, that is true. But it is unlikely that information which is present in SLC is lost after conversion to GRD. What is basically done is converting the rectagular pixels to squared pixels.
Have a look at this post: GTC (Geocoded terrain corrected) data
If the plastic is the size of a single pixel you might get no result anyways because of effects of speckle and waves in SAR data. A mean filter which enhances bright pixels might be suitable but that depends on the smoothness of the water surface.
I want to pre-process WV Mode acquisitions, however, it seems not possible to use the chain described due to several errors:
- Error: [NodeId: Calibration] WV is not a valid acquisition mode from: IW, EW, SM
- Error: [NodeId: TOPSAR-Deburst] WV is not a valid acquisition mode from: IW, EW
Can someone give me some advice how to process S1- SLC - WV mode data?
I am working on a similar issue, Ship Detection in the high seas using Sentinel 1. For this only WV mode acquisitions are available. The Pre-processing I have done so far is Multilook (4 times in Range) to reduce speckle.
We have to use SLC data since GRD data is not available over the sea.