Level-3 Analysis Ready Data - Sentinel-2 and Sentinel-1 Mosaics

If you want satellite-based information, but not at the frequency of satellite observations, imagery mosaics are an ideal data source for you. These data collections are created by integration of data from several satellite observations, providing seamless coverage across wide areas, without the noise affecting individual images.

Sentinel-2 mosaics are created every three months based on the cloud-free pixels of all images in the observation period, using the first quartile value to avoid the effect of eventually remaining cloud pixels. Sentinel-1 mosaics are built monthly, using a weighted average of each pixel value, where the weighting is based on the local resolution. This way, the artefacts created by slopes facing towards or away from the sensor are avoided.

The results are spectacular for both collections: swath edges are barely visible, the pixel values truly correspond to the properties of the surface, so image interpretation is a lot easier. This is particularly useful for Sentinel-1 data. In addition to visual evaluation, these mosaics also serve as benchmarks for evaluating change, e.g. in a disaster setting where a clear map of the original situation has to be obtained quickly. Also, these datasets are suitable without further processing as input data for machine learning based products, directly from the API request.

We trust that by further standardizing the pre-processing of earth observation data, we have created products that are particularly useful for operational analysis.

By András Zlinszky, Community Evangelist at Sinergise Solutions