First of all, thank you very much to ESA for all it has done with this Forum.
Im a student of Telecommunications at Canary Islands and Im just finishing my studies, making a Final Project studyng the capabilites of SAR Sentinel-1 for land classification.
Last day (23 October) a village of Canary Islands suffered the worst rainning event of the last years…110 litres/square metre (2 hours). The rain did a lot of damage to a little part of the coast of the Island of Gran Canaria. The main thing here is that, Sentinel-1 just made one SAR capture at the moment of the event (Coincidence!). An IW product.
So I want to know if it is possible to determine the areas with water with this data. I have read that it is better to have a quad pol product, but these ones are dual polarisation only. So, before starting making tests…do you recommend me some steps for water/soil detection on land with this data?
Im going to try with Polarimetric Decomposition, as it says on tutorials…but they are thought for Quad Polarisations…
The exact product I have got is this: S1A_IW_GRDH_1SDV_20151023T190457_20151023T190526_008284_00BACF_DFB1 (available also on RAW or SLC)
Thank you very much in advance, it would be great to have results for this weather event in which goverment has declared a local emegency situation.
Sentinel-1 has only dual polarization capabilities, and over-land, usually, you will find products in DV (VV/VH) as is the case with your product or in vertical polarization (SV).
The flood (or water) tend to appear black on SAR images, if the area is rural (this could be the case for a village) and is totally submerged. In this case, you could already locate the flood visually and try to find a pre-flood image (acquired in non-flooded situation) and perform a change detection between the two (the difference between the two images for example). If you cannot find a pre-flood image, you could choose a manual threshold (just to have an idea…).
Obviously, before applying the flood detection, your image need to be calibrated (see Radiometric Calibration), and potentially speckle-filtered (see Adaptive filters : Lee, Frost, Gamma)/ terrain corrected (using SRTM).
Flooding can be detected when the flooded scene is compared with a pre-flood scene or scenes. Try creating ratio-images between a pre-flood and flooded scene and see if the flooded area stands out.
Hi hakim, I have a quastion about flood monitoring by sar images, I want to know why we use GRD format instead of SLC one? you know better that we have both phase and amplitude information in slc products, (amplitude=square(i^2+q^2). Am i right? so I am confused know and pls explain the reason to me. thanks
Hi @zahra0729, as far as I know, the only difference between a GRD and a SLC product is that the GRD one has squared resolution cells and has been multilooked to reduce the speckle. Check pages 8 and 9 in the following document about TerraSAR-X products.
A GRD product looses the phase information as well, and therefore there’s no way to apply InSAR techniques to it.