Skip to main content

Surface Reflectance Satellite Data

Surface Reflectance (SR) is reflectance corrected for atmospheric effects, giving an accurate representation of the Earth's surface properties. SR data underpins a range of remote-sensing analyses, from vegetation monitoring and land-use mapping to change detection and crop yield estimation.

Piksel provides five SR products from two main constellations: the Landsat family (USGS) and Sentinel-2 (ESA).

Product Comparison

ProductSensorResolutionCoverageRevisitBest for
Landsat 5TM30 mMar 1984 – Jun 201316 daysHistorical analysis, 1984–2013
Landsat 7ETM+30 mApr 1999 – Jan 202416 daysLong-term time series (SLC-off after May 2003)
Landsat 8OLI30 m (15 m pan)Mar 2013 – present16 daysRegular updates, consistent with L9
Landsat 9OLI-230 m (15 m pan)Oct 2021 – present16 days14-bit radiometry, paired with L8
Sentinel-2MSI10–60 mJan 2017 – present5 daysHighest resolution, red-edge bands, precision agriculture

Combining Landsat 8 + 9 provides an effective revisit time of 8 days. Adding Sentinel-2 allows the same area to be observed up to several times per week.

Choosing a Product

  • Need a long historical record? — Use Landsat 5 for 1984–2013 or Landsat 7 for 1999–2024. Both products have 30 m resolution and are consistent with each other.
  • Need the highest resolution?Sentinel-2 provides 10 m pixels in the RGB and NIR bands.
  • Need regular updates and a short revisit time? — Combine Landsat 8, Landsat 9, and Sentinel-2.
  • Need red-edge bands for advanced vegetation analysis?Sentinel-2 provides three red-edge bands (B5, B6, B7).
  • Need surface temperature rather than reflectance? — See Surface Temperature.

About Surface Reflectance

Surface reflectance (BOA — bottom-of-atmosphere) differs from top-of-atmosphere reflectance (TOA — top-of-atmosphere) because it has been corrected for:

  • Scattering by aerosols and air molecules
  • Absorption by water vapour and other atmospheric gases
  • Adjacency effects from surrounding objects

Each product uses a different correction algorithm: LEDAPS for Landsat 5/7, LaSRC for Landsat 8/9, and Sen2Cor for Sentinel-2. These algorithm differences produce small variations between products that need to be considered when combining cross-sensor datasets for time-series analysis.