Methodology

 

The ECoLaSS prototype products for the evolution of Copernicus High Resolution Layers (HRL) focus on the establishment of methods which are suitable for large-scale, automated processing pipelines. The data foundation are dense Sentinel 1 and Sentinel 2 time-series, which, however, are restricted to cover a single year only, in order to accomodate potential yearly updates of HRLs. Moreover, particular attention is paid to implement methods suitable for an improved spatial resolution of 10m per product.

Automated, large-scale EO data analysis requires a number pre-processing steps to be performed, To this end several processing chains were developed to derive readily usable products. For Sentinel 2 this includes atmospheric correction and cloud-masking of Level 1C data. For Sentinel 1 data are processed from GRD to backscatter coefficients with a number of filtering steps to improve data quality.

Based on these pre-processed data, temporal features are derived, which are then used in supervised classification approaches. This involves a) identifying relevant temporal features,  b) fitting classification models and c) production of the final HRL dataset. Lastly, performance and consistency of the results are validated with reference data.

This project has received funding from the European Union’s Horizon 2020 research and innovation programme, under grant agreement no 730008.