Sentinel-2 Imagery Comparison

Explore the difference between a true-color Sentinel-2 satellite image and its corresponding land cover classification. Drag the slider bar to reveal more of either the original satellite view or the classified land cover map.

True Color

Land Cover Classification

Satellite imagery offers invaluable insights into our planet. The Sentinel-2 mission, a part of the European Union's Copernicus programme, provides high-resolution optical imagery for land monitoring, emergency services, and security. Below, we detail how these specific image products are created.

True Color Composite

A True Color composite image is created by combining specific spectral bands to visually represent the Earth's surface as it would appear to the human eye. This type of composite is often referred to as a "natural color" image.

• Red Band (B4): Corresponds to the red part of the electromagnetic spectrum. Healthy vegetation absorbs red light strongly, while bare soil and urban areas reflect it more.
• Green Band (B3): Corresponds to the green part of the electromagnetic spectrum. Vegetation reflects green light, which is why healthy plants appear green.
• Blue Band (B2): Corresponds to the blue part of the electromagnetic spectrum. Water bodies tend to absorb blue light, while aerosols and haze scatter it.

By assigning the Red band to the Red color channel of the display, the Green band to the Green color channel, and the Blue band to the Blue color channel (i.e., **RGB = B4, B3, B2**), we get an image that closely resembles what we would see from space. This makes it ideal for general visualization, identifying urban areas, bare ground, and recognizing cloud formations.

Land Cover Classification

A Land Cover Classification image is a thematic map derived from **Sentinel-2 L2A satellite data**, where each pixel is assigned to a specific land cover class based on its spectral characteristics. Unlike True Color images, which are direct representations of light reflection, classification images are the result of advanced data analysis and machine learning algorithms.

• Data Source: The classification is typically performed using multiple spectral bands from Sentinel-2, often including:
  • Visible bands (Blue, Green, Red)
  • Near-Infrared (NIR) bands (e.g., B8, B8A): Highly sensitive to vegetation health and density.
  • Shortwave Infrared (SWIR) bands (e.g., B11, B12): Useful for distinguishing between different types of vegetation, soil moisture, and bare ground.
• Classification Process: Algorithms analyze the unique spectral signature (how different land cover types reflect and absorb light across various bands) of each pixel. This involves training the algorithm with known samples of different land cover types (e.g., forest, water, urban, agriculture).
• Output: Each pixel is then assigned a color that represents its classified land cover type (e.g., dark green for dense forest, light green for grassland, grey for urban areas, blue for water, etc.). The specific colors used depend on the classification scheme.

Land Cover Classification images are crucial for environmental monitoring, urban planning, agriculture management, and understanding ecological changes over time. They provide a simplified yet powerful thematic view of the landscape.