Increasing Sentinel-2 spatial resolution to 2.5 meters/pixel

Pan-sharpening of sentinel-2 by ALOS PRISM 2.5m panchromatic orthorectified imagery using image fusion algorithms to increase spatial resolution from 10 meters to 2.5 meters/ pixel.

The final product can produce 10000 scale maps on different band composites, the example here applied on sentinel-2_L2A composite of 12/4/2 in RGB which very common in geological mapping.

ALOS PRISM is an optical instrument which observes the earth surface with a spatial resolution of 2.5 m in visible bands. Its data are used for creating high-precision digital elevation models (DEM).

The instrument has three optical systems for acquiring topographic data with altitude information. It acquires data in three directions―forward, nadir and backward―along the satellite track at the same time.
Read more about ALOS products…

PRISM data can be accessed from ESA earth online and download ALOS Africa products through the ESA LDS. The data is freely available to registered EO-SSO users.

4 thoughts on “Increasing Sentinel-2 spatial resolution to 2.5 meters/pixel

  1. Nicely done.
    However, if you don’t need the standalone image, but are working within a GIS, you can just ‘blend’ the Sentinel image layer with the underlying ALOS greyscale layer. For example, in QGIS:
    1. Open the Properties dialog of the Sentinel layer
    2. On the Symbology page, in the Color Rendering dialogue, select Blending mode ‘Multiply’ and click Apply – Adjust Brightness and Contrast as required
    The resulting image is similar in colour dynamics to a fused image. Much better than adjusting the transparency of the overlying layer, as this reduces the colour strength of the latter.

    You can do the same using a ‘live’ Google Earth or Bing maps feed in an underlying layer, which in many places has 1-metre ground resolution or less. The Color Rendering of the feed must be set to Greyscale, of course.

    1. Yes, Sure this is another way, But about using Google Earth or Bing maps its too difficult as this imagery not ortho-rectified, for best result you should use ortho image

Leave a Reply