‹ LibraryBasicChapter 4/18
Chapter 04

Introduction to Remote Sensing

Remote sensing enables observation of Earth from space and air, providing repetitive, consistent data across vast areas.

Google Earth EngineSNAP (ESA)ENVIERDAS IMAGINEOrfeo ToolBox
50+ YearsLandsat Archive↗ Since 1972
5 DaysSentinel-2 Revisit↗ 10m, 13 bands
70+ PBGEE Catalog↗ Free data

Theory & Foundations

Remote sensing exploits the electromagnetic spectrum. Different materials interact uniquely with EM energy, creating spectral signatures sensors detect from orbit.

The EM spectrum ranges from gamma rays to radio waves. Remote sensing uses visible (0.4–0.7μm), NIR (0.7–1.3μm), SWIR (1.3–3.0μm), thermal IR (3–14μm), and microwave (1mm–1m) regions.

Passive sensors detect reflected sunlight or emitted thermal energy — they need illumination and clear skies. Active sensors (radar, LiDAR) generate their own energy and work through clouds, day or night.

Four resolution types define sensor capability: Spatial (pixel size), Spectral (number of bands), Temporal (revisit frequency), Radiometric (sensitivity in bits). Higher spatial resolution usually means lower temporal coverage.

When EM energy hits a surface: absorption + reflection + transmission = 100%. Water absorbs NIR (appears dark), healthy vegetation reflects NIR strongly (appears bright). These patterns create unique spectral signatures.

In-Depth Coverage

The Electromagnetic Spectrum

Remote sensing exploits different wavelength regions for specific information.

  • Visible (0.4–0.7μm): Blue, green, red
  • Near-IR (0.7–1.3μm): Vegetation health indicator
  • SWIR (1.3–3.0μm): Moisture, mineral mapping
  • Thermal IR (3–14μm): Surface temperature
  • Microwave (1mm–1m): Penetrates clouds, soil moisture
💡 Pro Tip: Healthy vegetation reflects ~50% NIR but only ~10% red — this drives NDVI calculation.

Resolution Types

Four dimensions define sensor capability.

  • Spatial: Sentinel-2=10m, Landsat=30m, MODIS=250m
  • Spectral: Landsat=11 bands, Sentinel-2=13, Hyperion=242
  • Temporal: MODIS=daily, Sentinel-2=5 days, Landsat=16 days
  • Radiometric: 8-bit (256 values) to 16-bit (65,536)

Active vs Passive Sensors

The fundamental distinction in remote sensing technology.

  • Passive optical: Needs sunlight + clear sky (Landsat, Sentinel-2)
  • Passive thermal: Day/night but needs clear sky (MODIS LST)
  • Active SAR: Day/night, through clouds (Sentinel-1, ALOS-2)
  • Active LiDAR: Day/night, highly accurate 3D (ICESat-2)

Key Techniques

Atmospheric correctionBand compositingImage enhancementSpectral signature analysisVisual interpretation

Data Visualization

Satellite Sensor Band Usage in Research (%)
Visible15
Near-IR25
SWIR20
Thermal18
Microwave22
Remote Sensing Workflow
  1. 📡AcquireDownload from USGS, Copernicus, GEE
  2. ⚙️Pre-ProcessAtmospheric correction, georeferencing
  3. 🔍EnhanceBand composites, pan-sharpening
  4. 📊AnalyzeClassification, indices, change detection
  5. ValidateGround truth, confusion matrix

Practical Workflow

  1. Define monitoring objective
  2. Select sensor based on resolution needs
  3. Search data archives
  4. Pre-process: calibration, atmospheric correction
  5. Geometric correction if needed
  6. Enhancement: band composites, filtering
  7. Analysis: classification, indices, change detection
  8. Validate with ground truth data

Software & Tools

SNAPFreeDesktop Software — ESA Sentinel data processing
ENVICommercialDesktop Software — Advanced image analysis

Video Tutorials

Real-World Application

Landsat has provided continuous Earth observation since 1972 — 50+ years tracking Amazon deforestation (17% lost), Aral Sea shrinkage (90% loss), and global urban expansion.

Real-World Projects

Sundarbans Mangrove Monitoring📍 Bangladesh

Satellite NDVI time series tracks mangrove health in world's largest mangrove forest.

Impact: Early detection of 15% canopy decline triggered conservation action.

Case Study

ICIMOD uses Sentinel-2 to monitor 47 dangerous glacial lakes in the Hindu Kush Himalaya, protecting 5+ million downstream residents from outburst flooding.

Problem-Based Learning

P1 Monitor tropical deforestation through persistent cloud cover

Amazon 80%+ cloud cover during wet season. Optical satellites cannot see ground.

Solution: Use Sentinel-1 SAR radar which penetrates clouds. Process backscatter time series — deforested areas show change.

Sentinel-1 GRDGoogle Earth EngineSNAPRandom Forest
✅ Detected 2,340 km² deforestation during wet season that optical sensors missed entirely.

Knowledge Check

1. Which sensor images through clouds?
2. What does NDVI measure?

Further Reading