Poor sentinel -1 processing results or lack of knowledge

Dear sentinel-1 SAR users. I have tested Sentinel-1 data (SLC - interferometry, GDHR - polarimetry) for couple of weeks now, using various approcahes and processing parametres for the purpose of agriculture in Slovenia, but receiving very poor results.
With SLC I have tested phase to displacement, where I was hoping to detect minimum agricultural activity (mowing, plant growing, ploughing), but no relevant info collected from interferometry on the parcel level, only some clustered data.
Than I have tried to use GDHR products to create initial crop map based on S-1 data. Although I receive some promissing results for small number of parcels, I have struggled to improve results using different parametres, but no success.
Has someone share same experiences? I usually started with standard procedure provided on youtube:
SLC interferometry: RUS Webinar: Land Subsidence mapping with Sentinel-1 - HAZA03
GRDH polarimetry: ESA Echoes in Space - Land: Crop type mapping with Sentinel-1
I can provide some of my results if requested.

I’m not sure you’re using the right derived products for the job. For activities like mowing, look at the coherence from two coregistered SLCs at 6 or 12 days apart before and after the event.

You should look into 6- and 12-day coherence and see where that gets you. I can recommend this thesis :wink:, it uses ERS-1/2 Tandem data but the principles apply to Sentinel-1 coherence exploitation as well:


1 Like

Dear all, many thanks for your posts. In fact I have used for phase to displacement 12 days apart S1 SLC products and the same now for coherence within time period where mowing took place.
For crop map I have used 24 GDHR products 12 days apart, from January 2019 to October 2019.
However, due to the average size of AG parcel in my conutry some more sofisticated approach may be needed. ERS-1/2 has much better spatial resolution than S-1, which may have significant impact on the results, where I try to analyse AOI of 0,5 ha to 2 ha.

I am tnow trying to get as much as possible of S-1 products in terms of phase to displacement products, such as early warning land slides, open mining (sand harvest), of landfil (volume of waste) as precise and realiable as possible. Any hint what parametres shall be used for the following procedure I use, would be very welcome:
0: Top split
1: apply orbit file
2: Back geocoding
3: ESD
4: TOPSAR deburst
5: Topophase removal
6: Multilook (try to keep pixel size as small as possible, although it is a trade-off)
7: Goldstein filtering

SNAPHU export
SNAPHU unwrapping
SNAPHU import
terrain correction
raster projection
export as GeoTIFF

S-1 has much better range-resolution, which really makes a positive difference for interferometry. If your area is vegetated, as most areas outside cities are, you are much better off with using 6-day than 12-day data.

Maybe you could try to use this tutorials to find that agricultural activity.

I’ve used it to detect changes in agricultural lands, but I don’t know exactly to what extent it works.