Unit
- Electromagnetic spectrum: Remote sensing relies on the electromagnetic spectrum, from visible light to microwaves. Each wavelength interacts differently with Earth materials, enabling detection of distinct features like vegetation health or soil moisture.
- Radiant energy and sources: The sun acts as the primary source for passive systems, emitting energy that reflects off surfaces. Active systems, such as radar, generate their own energy and measure the return signal.
- Atmospheric interactions: As energy passes through the atmosphere, it undergoes scattering and absorption. Rayleigh scattering affects shorter wavelengths, making the sky blue, while absorption by gases like water vapor limits usable bands.
- Spectral signatures: Different materials reflect and absorb energy at specific wavelengths. Vegetation shows high reflectance in near-infrared and low in red, forming a distinct spectral curve used for classification.
- Resolution concepts: Spatial resolution defines the ground area per pixel. Spectral resolution refers to the number and width of bands. Temporal resolution is the revisit frequency of a satellite. Radiometric resolution captures the sensitivity to signal strength.
- Passive vs. active systems: Passive sensors detect natural radiation; they require sunlight and operate during daytime. Active sensors emit pulses and measure returns, working day or night and through clouds (e.g., radar).
- Platforms for sensing: Satellites provide global coverage but limited revisit. Aircraft offer higher resolution on demand. Drones enable ultra-high resolution for local studies, each with trade-offs in cost and coverage.
- Data formats and storage: Remote sensing data is stored as raster images, often in GeoTIFF format. Bands are stacked, and metadata includes georeferencing, acquisition date, and sensor calibration.
- Image preprocessing: Atmospheric correction removes scattering and absorption effects to retrieve surface reflectance. Geometric correction aligns imagery to a map coordinate system, correcting distortions from sensor geometry and terrain, ensuring accurate spatial location for subsequent analysis.
- Classification methods: Unsupervised algorithms like K-means group pixels by spectral similarity without training data. Supervised methods such as maximum likelihood use labeled training sites to assign land cover classes. Accuracy is assessed through confusion matrices and kappa coefficients.
- Vegetation indices beyond NDVI: The Enhanced Vegetation Index (EVI) corrects for atmospheric and canopy background effects, improving sensitivity in dense biomass. The Normalized Difference Water Index (NDWI) uses green and near-infrared bands to detect water bodies and monitor moisture stress.
Code Example
Python
import rasterio
from rasterio.plot import show
with rasterio.open('landsat8_sample.tif') as src:
print('CRS:', src.crs)
print('Transform:', src.transform)
print('Band count:', src.count)
red = src.read(3)
nir = src.read(4)
ndvi = (nir - red) / (nir + red + 1e-10)
show(ndvi, cmap='RdYlGn')This snippet loads a Landsat 8 scene, reads red (band 4) and near-infrared (band 5) data, computes the Normalized Difference Vegetation Index, and displays it. NDVI values range from -1 to 1, with healthy vegetation showing high positive values.
Quick Revision
- The electromagnetic spectrum provides the foundation for remote sensing, with each wavelength band suited to different applications.
- Atmospheric scattering and absorption affect the quality and usability of remote sensing data.
- Spectral signatures enable material identification by analyzing reflectance patterns across wavelengths.
- Understanding resolution types helps in selecting appropriate data for specific analysis tasks.