Jupyter Notebook script for extracting Sentinel-1 VV/VH Time Series from Google Earth Engine over burned area AOIs

Descriptions: 

First part - Sentinel-1 VV/VH Statistical Time Series Extraction Script (Median, 5th, and 95th Percentiles)

This repository contains a Python script for extracting percentile-based time series of Sentinel-1 VV and VH radar backscatter coefficients using the Google Earth Engine (GEE) platform.
The workflow was developed within the study “Response of the Sentinel-1 radar backscattering to an extreme wildfire event: surface soil moisture and vegetation cover implications” (Esposito et al., under revision) and is designed to support environmental monitoring of soil moisture and vegetation dynamics following wildfires or other surface disturbances.

The script automatically filters Sentinel-1 Ground Range Detected (GRD) data over a user-defined Area of Interest (AOI) (uploaded here as gpkg), processes ascending and descending orbits separately, and computes the 5th, 50th (median), and 95th percentiles of VV and VH backscatter for each acquisition date.
The resulting statistics are exported as CSV tables to Google Drive, providing ready-to-analyse datasets for temporal backscatter analysis, model calibration, and validation of optical–radar indices.

This workflow facilitates reproducible and scalable radar-based monitoring across different environmental contexts, and it is particularly suited for analysing the hydrological and ecological responses to wildfire events.

 Second part 3-Month Averaged Time Series (Post-Processing Script)

This Python script processes and visualises Sentinel-1 VV and VH percentile time series (5th, 50th, and 95th percentiles) previously exported from Google Earth Engine (GEE).
It aggregates the radar observations into custom 3-month intervals corresponding to the following fixed periods:

• Feb–Apr → 30 April

• May–Jul → 31 July

• Aug–Oct → 31 October

• Nov–Jan → 31 January (next year)

For each period, it computes the mean values of VV and VH percentiles and generates both:

• aggregated CSV tables, and

• interactive Plotly-based visualisations (HTML format) (see the attached example)

This workflow enables a compact representation of radar temporal dynamics suitable for long-term analyses of surface moisture, vegetation recovery, and post-fire backscatter trends.

The same scripts could be adapted to extract time series over other Study areas or using different satellite types available on GEE. 

Inputs

• CSV files exported from GEE containing the following columns:

   

  date, VV_p5, VV_median, VV_p95, VH_p5, VH_median, VH_p95

• Separate files for ascending and descending orbit passes are supported (e.g.
  VV_VH_ASC_Median_Perc_TS.csv and VV_VH_DESC_Median_Perc_TS.csv).

::::::::::::

Dependencies

• Python ≥ 3.8

• Google Earth Engine Python API (earthengine-api)

• Jupyter Notebook (recommended for interactive use)

Setup Instructions

1. Authenticate your GEE account in the terminal:

    

   earthengine authenticate

2. Open and execute the script in a Jupyter Notebook.

3. Edit the AOI asset path and orbit track numbers as needed.

4. Run the script to initiate Google Drive exports.

5. Monitor export progress at: https://code.earthengine.google.com/tasks

6. Open Google Drive and download the CSVs
7.  Run  the second part of the script 

AOI Sardinia  fire (July 2021) 

Moreover, the 18 AOIs selected for analysis are presented in the paper and are attached as a GPKG file. 

:::::::::

References to related work 

Giuseppe Esposito, Massimo Melillo, Davide Notti, Maria Teresa Brunetti, Silvia Peruccacci, Luca Pisano, Luca Brocca, Rosa Maria Cavalli,
Response of the Sentinel-1 radar backscattering to an extreme wildfire event: surface soil moisture and vegetation cover implications,
Science of Remote Sensing, 2025, 100339, ISSN 2666-0172, https://doi.org/10.1016/j.srs.2025.100339.