Flood Detection with Sentinel SAR 1 or related satellite data

Hello. It’s my first time here. I would like to ask whether anyone has tried to detect flooding in dense urban settings such as a city like Metro Manila? I have tried using subtraction of before and after images of flood events with GEE; however, it has yielded inaccurate results. I have also used a machine learning software to identify flooded areas, but it had similar results. Additionally, data from GFM Glofas are similar as well. Are there any other satellite data and better algorithms or models that I could for my purpose?

1 Like

Welcome @Nick_Anthony_Miras,

I just saw a talk by Eduard Heijkoop at NASA JPL on high-res flood maps in the pacific. They tool he talks about is here:
Pacific Flooding Analysis Tool, but Manilla isn’t listed.

Probably not exactly what you’re looking for, but maybe some of the data and techniques might be useful.

2 Likes

Welcome!

I remember seeing stuff about the ETCI NASA-Impact competition, like this solution: GitHub - sidgan/ETCI-2021-Competition-on-Flood-Detection: Experiments on Flood Segmentation on Sentinel-1 SAR Imagery with Cyclical Pseudo Labeling and Noisy Student Training. Perhaps you can find some cool ideas there, although I don’t think any of the data is in very dense urban areas!

What data sources have you considered so far? My feeling is that in very dense urban settings, detecting floods directly from satellite imagery will be challenging (for instance with SAR you are looking at an angle, so you will have a lot of geometric effects caused by tall buildings, and for optical data, well… clouds)

I always find it useful to extensively inspect the imagery sources before / during the event I’m trying to detect - can you spot the flooded areas by eye in some band combinations?

Thank you so much, glad to be here! Regarding the talk, are there any recordings of it?

Thank you! So far, I’ve looked into ALOS-2 PALSAR-2 and Sentinel-1. Unfortunately, a resolution of 25 and 10 meters, respectively, is not enough to identify flood inundation in Metro Manila through ‘eyeballing’ it. Do you have other recommendations for alternative SAR data sources?

Maybe have a look at the papers published by LIST for urban flood mapping: Google Scholar