🚀 New open-source code release!

We released a python-based demonstration pipeline comparing Sentinel-1 SAR and Sentinel-2 Optical satellite time series for crop monitoring over tropical agricultural landscapes.

Featured Field #584 (Soybean → Millet rotation, 782 ha). Left: Sentinel-2 NDVI curve breaks during wet-season cloud cover. Right: Sentinel-1 VV/VH backscatter (dB) tracks the full crop growth cycle without missing dates.

The demo focuses on the LEM+ ground-truth field dataset in Luís Eduardo Magalhães, western Bahia, Brazil — tracking fields across a full crop year (Oct 2019 – Sep 2020) with monthly crop label updates.

🔑 Key Features:

  • Cloud-resilient tracking: Demonstrates how Sentinel-1 VV/VH backscatter maintains continuous time series during wet-season cloud cover when Sentinel-2 optical imagery suffers data gaps.
  • Consistent geometry: Automatically pins the stack to a single Sentinel-2 MGRS tile (23LLG) and single Sentinel-1 orbit (DESCENDING 24).
  • Field-level phenology: Synchronizes zoomed field-level imagery with time-series curves for key crop rotations (e.g. Soybean → Millet and Soybean → Corn).
  • Lightweight export: Modular pipeline that streams data from keyless STAC catalogs (AWS Earth Search & Planetary Computer) and exports versioned animated GIFs and contact sheets.

💻 Source Code & Tutorial:

Check out the step-by-step tutorial and repository on GitHub:
👉 victorohden/agriculture-sar-optical-monitoring