Optical Analysis with Sentinel-2

Cloud masking, spectral indices, and time series — turning raw scenes into monitoring signals

Optical Analysis with Sentinel-2
Intermediate Remote Sensing 5 lessons Samuel Appiah Kubi

About this course

Raw Sentinel-2 scenes are not analysis-ready on their own — they need atmospheric correction, cloud masking, and often compositing across several dates before the numbers mean anything. This course builds that full preprocessing pipeline and applies it to two real Ghanaian problems: monthly vegetation monitoring over the White Volta catchment and cocoa farm detection across the Ashanti/Brong-Ahafo belt.

What you'll learn

  • Apply the Scene Classification Layer to correctly mask clouds, shadow and snow
  • Explain why L2A (surface reflectance) is used instead of L1C (top-of-atmosphere) for quantitative work
  • Compute NDVI, NDWI and NDBI and know exactly when to use each
  • Build a 12-month NDVI time series over a real catchment
  • Build a median composite from multiple dates and explain why it beats a single scene

Requirements

  • Python for Geospatial Analysis (or equivalent)
  • pygeofetch installed and working

Course content

The Scene Classification Layer (SCL): Your First and Most Important Filter Preview 22 min
L1C vs L2A: Why the Processing Level You Choose Matters 16 min

NDVI, NDWI, NDBI: the Three Indices That Cover Most Use Cases 24 min
Building a 12-Month NDVI Time Series over the White Volta 28 min
Case Study: Detecting Cocoa Farms in the Ashanti/Brong-Ahafo Belt 32 min
Free
Login to Enroll
  • LevelIntermediate
  • Lessons5
  • CertificateYes
  • AccessLifetime
Samuel Appiah Kubi Instructor