Case study
SAR-Based NDVI Prediction and Crop Spectral Library
RMSE 0.07 across a cloud-cover prediction workflow
The problem
Optical satellite imagery can be unavailable under cloud cover, leaving crop-health monitoring without a useful observation.
Our approach
We built mean ± standard-deviation spectral profiles for 8 crops across 15 districts and trained a Random Forest model on Sentinel-1 backscatter to predict NDVI.
- Build spectral signatures from Sentinel-2 for 8 crops.
- Organize the library across 15 districts.
- Train the Random Forest model on Sentinel-1 backscatter.
- Evaluate predicted NDVI against the target signal.
Technology
Results
- RMSE 0.07 for NDVI prediction during cloud cover.
- Spectral library for 8 crops across 15 districts.
What this means for your project
This is a geospatial data-processing and automation pattern for teams working with satellite data and the reliability of downstream analytics. Explore the matching service
Related projects
A clear next step
Have something in mind? Let's talk.
Tell us what you're building and we'll help scope the right technical path.
Start a Project