Threshold Standardization in S1 data?

Question

I am trying to extract waterbody for Vietnam for last 5 years using S1 data in SNAP tool, however the backscatter coefficient value keeps on changing for the years.

Is there any way to Standardize the value?

Answer 1

Within SNAP, the "Stack Averaging" tool should help, when dealing with Multitemporal analysis of data. Radar > Coregistration > Stack Tools > Stack Averaging.

The following document should help:

http://eoscience.esa.int/landtraining2017/files/materials/D2P1__I.pdf

Using Python, I believe Scikit-Learn (Sklearn) will be able to help here. There are a few options to rescale data. "Normalization", "Standardization" and "MinMaxScaler".

This, however, will require the data to be in tif. format. Though, this is easily done in SNAP. When saving the product out, change the format from "BEAM-DIMAP" to "GeoTIFF" or "GeoTIFF-BigTIFF"

https://machinelearningmastery.com/rescaling-data-for-machine-learning-in-python-with-scikit-learn/

Answer 2

After working on the waterbody extraction, I realized that it is better to go with Google Earth engine than SNAP, where we can run Otsu algorithm and extract the water body as per our need.

Here is a code to refer https://code.earthengine.google.com/499580510635aa5f421886d4e409af03 from Calculating water occurrence of Sentinel-1 images in Google Earth Engine