Analysis Ready Data

On Friday (15th September 2017) I attended the second open meeting (of two) linked to a project focused on the generation of Sentinel-2 Analysis Ready Data (ARD). The project partners were Aberystwyth University, Defra (through the JNCC) and the Satellite Applications Catapult (who acted as hosts of this meeting) and the point of the meeting was to update interested parties on progress and find out more about users needs.

The meeting started with a technical review from Dr Pete Bunting about the development undertaken as part of the project to include Sentinel-2, of the open-source arcsi software he maintains. The presentation was a comprehensive review of how the software creates atmospherically corrected datasets. In addition, Pete outlined some of the areas that need further consideration for the generation of Sentinel-2 ARD more broadly: improved cloud masking algorithms, understanding illumination correction algorithms in different datasets, applying algorithms to multiple tiles of data, etc.  Next, Frederica Moscato presented on how the Catapult used the NiFi framework to scale up the processing of the incoming satellite data, and subsequently investigated the use of Open Data Cube to display the ARD.

The technical aspects of the project were put into context with a couple of talks given by Ferran Gascon of ESA.  He first outlined the specifications being suggested as part of the CEOS CARD4L project, which looks to help data providers offer their data to users in a more standardised manner. Then he outlined the Level 2 data processing of Sentinel-2 imagery that is being undertaken by ESA and offered through the Collaborative Nodes.

Following lunch, Dr Richard Hilton provided an overview of the importance of ARD, how it is being, and in future might be, put to operational use,  and what some of the key considerations should be about producing ARD i.e. storage, archiving, costs and funding, standards etc. The final part of the meeting was a break-out session where the attendees discussed some of these issues and fed back to the group as a whole.

Geoger thinks…

Having access to standardised ARD (in whatever form that takes) is critical in promoting the uptake of satellite imagery in the wider user community. At present, too much time and money from project budgets is committed to creating usable datasets. Often these data are full of gaps where cloud cover obscures the site of interest, and it is almost always the case that users download a lot more data than they require i.e. entire images, when their site of interest might only be a few kms in diameter. Being able to trust the source of an ongoing archive of analysis ready data, and only access the parts of that dataset that you need, would free up more time and money for the applied aspects of satellite data use. The work being undertaken as part of this project is incredibly exciting and could really benefit the UK user community, generating new projects and use cases from the provision of these data.


One thought on “Analysis Ready Data

  1. Good Summary. Except perhaps as a proxy evidence for polar wandering phenomenon, long term archives of cloudy optical images (would) have very little other applications. In my opinion, so-called “ARD” must have a minimum requirement that it has nil or minimal cloud and shadow, because while zero landcover analysis can be done with a cloudy image, a great deal can still be done just fine with S2 Level-1C image with few simple thresholding trials. Folks have been atcor’ing satellite images for last 4 decades.

Comments are closed.