FS6 Cloud and gap free high resolution, high frequency satellite data

Challenge Description

This challenge aims to overcome a significant problem in the usage of high-resolution optical data, which is that, for many areas of the globe, cloudiness, haze, and gaps arising from revisit time limitations reduce the usefulness and information content of these data streams, in particular Sentinel-2 (but also Landsat-8), both by producing noisy seasonal profiles and by introducing gaps in the time series at a specific location.

A smoothing filter developed by WFP (Whittaker filter) is in principle able to minimize most if not all of these problems. Applied to Sentinel-2 it produces a noise and gap-free product, with the advantage that the output can be produced at any arbitrary time interval. This may significantly improve the usability of Sentinel-2 data and could provide much more accurate land cover classifications, for example for crop type mapping applications that WFP is engaged in.

The challenge is focused on the implementation of the filter – the initial application was focused on medium resolution data, but Sentinel-2 much higher spatial resolution and correspondingly much higher data volumes pose difficulties since the Whittaker filter is computationally intensive.

Results

The results below are divided into Software Repositories, Online Services and Data Products Repositories.
In case you would like to get in touch with us about specific results, please use the indicated reference codes (e.g. WFP-01-01-01).

Software Repositories

There are two categories of software repositories generated for the Challenge:

• GitLab repositories are hosting the reference software code, for use by the team members of a challenge only (restricted access)
• GitHub repositories are hosting the public software code (mirrors), for engagement and collaborative work with the broad community of software developers (public access)

Follow the links below to proceed to one or the other, and find an "Ellip Workflows (ewf)" software application implementing a data processing function.
You can adapt this software to your own context, commit a contribution to the related repository, or engage with the BETTER consortium to run it for a specific purpose (new area of interest, ...) by requesting Cloud resources to support (at) terradue.com

Data Processing as-a-Service

There are two categories of "Data Processing as-a-Service" application packages created for the Data Challenge:

• Data Processing Pipelines are controlling full EO data production campaigns, for use by approved members of a challenge only (authorized access),
• On-demand Processing Services are controlling individual EO data processing jobs, also for use by the approved members of a challenge only (authorized access).

Follow the links below to proceed to one or the other, via a dedicated Geobrowser application providing web access to one or the other of the Data Processing services.
You can engage with the BETTER consortium to request authorized access via support (at) terradue.com

Data Product Repositories

There is a dedicated client App publicly available for the visualization of the data products resulting from EO data processing challenges.

The results can consist of:

• Validation data: small amount of data products generated for approval by the Challenge promoter
• Production campaign data: full-time series/collection of data products answering the Challenge promoter question about Earth Observation data being used to help solve global societal challenges