1 Remote Sensing Basics
- Definition: Acquiring information about Earth's surface without physical contact, using EM energy.
- Active RS: Own energy source (RADAR, LiDAR). Day/night, penetrates clouds.
- Passive RS: Depends on sunlight (Landsat, SPOT). Affected by clouds/darkness.
2 EM Spectrum Bands
- Visible: Blue (0.45–0.52), Green (0.52–0.60), Red (0.63–0.69 μm).
- NIR: 0.76–0.90 μm — vegetation health.
- SWIR: 1.55–2.35 μm — soil moisture, minerals.
- Thermal IR: 10.4–12.5 μm — surface temperature.
- Microwave: >1 mm — RADAR, penetrates clouds.
3 Spectral Signatures (MOST ASKED)
- Vegetation: Low visible (photosynthesis absorption), sharp rise at Red Edge (~0.7 μm), very high NIR, dips at water bands (1.4, 1.9 μm).
- Water: High blue-green, near-zero after red. Clear water absorbs NIR.
- Soil: Gradual increase visible to SWIR. Wet soil = lower reflectance.
4 Four Resolutions
| Resolution | Definition | Example |
|---|---|---|
| Spatial | Pixel size on ground | Landsat: 30 m, WorldView: 0.31 m |
| Spectral | Number/width of bands | Multi: 4–10, Hyper: 200+ bands |
| Radiometric | Bit depth per pixel | 8-bit: 256, 16-bit: 65,536 levels |
| Temporal | Revisit time | Landsat: 16 days, MODIS: daily |
5 Landsat 8 Band Combinations
| Band | Name | μm | Res |
|---|---|---|---|
| 1 | Coastal Aerosol | 0.43–0.45 | 30 m |
| 2 | Blue | 0.45–0.51 | 30 m |
| 3 | Green | 0.53–0.59 | 30 m |
| 4 | Red | 0.64–0.67 | 30 m |
| 5 | NIR | 0.85–0.88 | 30 m |
| 6 | SWIR 1 | 1.57–1.65 | 30 m |
| 7 | SWIR 2 | 2.11–2.29 | 30 m |
| 8 | Pan | 0.50–0.68 | 15 m |
- True Colour: 4, 3, 2 — natural image.
- False Colour (FCC): 5, 4, 3 — vegetation appears red.
- SWIR: 7, 5, 4 — urban, bare soil, geology.
6 Image Classification
| Aspect | Supervised | Unsupervised |
|---|---|---|
| Training data | User provides | Not needed |
| Algorithm | MLC, Min Distance, SVM | K-Means, ISODATA |
| Knowledge | Must know area | Can be exploratory |
| Accuracy | Generally higher | Lower, needs labelling |
7 NDVI
<0 water | 0–0.1 bare | 0.2–0.4 sparse | 0.5–0.8 dense vegetation
8 Change Detection
- Compare two images from different dates.
- Methods: image differencing, ratioing, post-classification comparison, PCA.
- Applications: deforestation, urban growth, flood mapping, glacial retreat.
Draw reflectance vs wavelength (0.4–2.5 μm). Three curves. Label Red Edge, water absorption bands. Explain each curve shape.
Spatial (pixel size), Spectral (bands), Radiometric (bit depth), Temporal (revisit). Give satellite examples.
Active: own energy, day/night, RADAR/LiDAR. Passive: sunlight, limited by clouds, Landsat/SPOT. Draw comparison diagram.
True: 4,3,2. FCC: 5,4,3. SWIR: 7,5,4. Include band table. Explain what appears red/green in each.
Supervised: training data, MLC. Unsupervised: K-Means, ISODATA, user labels after. Use comparison table.
Formula, range. Applications: crop health, forestry, drought. Multi-temporal for seasonal change.
Multi-date images. Pre-processing: geometric + radiometric correction. Methods: differencing, NDVI diff, post-classification. Nepal examples: flood, deforestation.