14.1 The Electromagnetic Spectrum for Remote Sensing
Why satellites see what they see — the physics behind multispectral imagery.
Key takeaways
- Remote sensing exploits specific wavelengths of the electromagnetic spectrum.
- Atmospheric absorption limits which wavelengths reach sensors; "windows" exist.
- Different materials reflect different wavelengths differently — the basis for classification.
Introduction
Satellites don't "see" the Earth the way humans do. They measure electromagnetic radiation at specific wavelengths — visible, near-infrared, short-wave infrared, thermal, and microwave. Understanding the physics is essential for interpreting imagery correctly.
The spectrum
Electromagnetic radiation spans a huge range of wavelengths. Relevant regions for remote sensing:
| Region | Wavelength | Uses |
|---|---|---|
| Ultraviolet | 10–400 nm | Atmospheric ozone, SO₂ |
| Visible | 400–700 nm | Colour imagery; blue (450), green (550), red (650) |
| Near-infrared (NIR) | 0.7–1.3 μm | Vegetation health (NDVI) |
| Short-wave infrared (SWIR) | 1.3–3 μm | Water content, geology, burn scars |
| Thermal infrared | 3–14 μm | Surface temperature |
| Microwave (radar) | 1 mm – 1 m | Cloud-piercing imaging (SAR) |
Reflected vs emitted
Visible, NIR, and SWIR are dominated by reflected solar radiation (sunlight bouncing off Earth's surface).
Thermal infrared is dominated by emitted radiation — materials warm themselves and radiate heat.
Microwave (radar) is emitted by the satellite itself and received as a backscatter (active sensing).
Atmospheric windows
Earth's atmosphere absorbs some wavelengths strongly (water vapour in many IR bands; CO₂ and O₃ elsewhere). Sensors target atmospheric windows — wavelengths that pass through without major loss.
Main windows:
- 0.4–2.5 μm (visible + NIR + most SWIR).
- 3–5 μm (mid-wave thermal).
- 8–14 μm (long-wave thermal).
- Microwave (mostly transparent; few water lines).
Designers pick bands within these windows to maximise ground signal.
Spectral signatures
Each material has a characteristic reflectance curve — its spectral signature:
- Green vegetation: low reflectance in visible (chlorophyll absorbs red and blue), high in NIR (leaf cell structure).
- Water: moderate in blue/green, absorbing toward NIR and SWIR (nearly black).
- Snow: very high in visible, low in SWIR.
- Bare soil: moderate across spectrum, increases toward SWIR.
- Urban (concrete): relatively flat, high in SWIR.
Sensors with enough bands can distinguish these — the basis of classification.
Spatial, spectral, temporal resolution
Four resolutions matter:
- Spatial — pixel size.
- Spectral — number and width of bands.
- Temporal — revisit time.
- Radiometric — number of bits per pixel (sensitivity).
Trade-offs: more spectral bands often mean coarser spatial. Higher spatial often means narrower coverage (less frequent revisit).
Example: Sentinel-2 bands
| Band | Name | Wavelength | Resolution |
|---|---|---|---|
| B01 | Coastal aerosol | 443 nm | 60 m |
| B02 | Blue | 490 nm | 10 m |
| B03 | Green | 560 nm | 10 m |
| B04 | Red | 665 nm | 10 m |
| B05 | Red-edge 1 | 705 nm | 20 m |
| B06 | Red-edge 2 | 740 nm | 20 m |
| B07 | Red-edge 3 | 783 nm | 20 m |
| B08 | NIR | 842 nm | 10 m |
| B8A | Narrow NIR | 865 nm | 20 m |
| B09 | Water vapour | 945 nm | 60 m |
| B10 | Cirrus | 1375 nm | 60 m |
| B11 | SWIR 1 | 1610 nm | 20 m |
| B12 | SWIR 2 | 2190 nm | 20 m |
Each band is designed for specific applications — B11 and B12 excel at vegetation water content and burn-scar detection; the red-edge bands capture vegetation stress before visible chlorophyll change.
Active vs passive sensors
- Passive: measure natural radiation (reflected sunlight, emitted heat). Most optical / thermal satellites.
- Active: emit their own signal and measure the return. Radar (SAR), lidar, sonar.
Active sensors work at night and through cloud cover; passive sensors typically don't.
Self-check exercises
1. Why does vegetation look dark in the red band but bright in the near-infrared band?
Chlorophyll in leaves absorbs red light for photosynthesis — so reflectance in the red band is low. Leaf cell walls strongly scatter near-infrared wavelengths, producing high reflectance. This red↔NIR contrast is the basis for NDVI and dozens of other vegetation indices.
2. When would you prefer SWIR over visible imagery?
SWIR penetrates haze and smoke better than visible, making it useful for wildfire monitoring and post-fire burn mapping. SWIR also distinguishes water content (dry vs wet soil, senescing vs healthy vegetation), helping with drought and agricultural stress detection. It's invisible to the human eye, so natural-colour displays use visible, but analytical work benefits from SWIR.
3. Why can radar (Sentinel-1) see through clouds when optical sensors can't?
Radar wavelengths (1–30 cm) are much longer than cloud droplets (~10 μm), so the microwaves pass through clouds with minimal scattering. Visible and NIR wavelengths, being comparable to or smaller than droplet size, are strongly scattered and absorbed. This is why radar is essential in the tropics or polar regions with persistent cloud cover.
Summary
- The EM spectrum has many regions; sensors target specific atmospheric windows.
- Spectral signatures of materials allow classification from imagery.
- Four resolutions (spatial, spectral, temporal, radiometric) trade off.
- Sentinel-2 and Landsat sample the spectrum at bands chosen for specific applications.
Further reading
- Lillesand, Kiefer, Chipman — Remote Sensing and Image Interpretation (textbook).
- Jensen, J. R. — Remote Sensing of the Environment.
- NASA Earth Observatory — Spectral Signatures explainer.
- ESA Sentinel-2 mission documentation.
