1. Status Report on the Open Data Cube and the use of ARD
LSI-VC Meeting (remote presentation)
February 14, 2019
Hanoi, Vietnam
Brian Killough
CEOS Systems Engineering Office
NASA Langley Research Center

2. Where are the Data Cubes?
Operational
Under Development
Expressed Interest
55 countries are interested with 9 in operation and 14 coming soon!

3. Moving toward ARD
Analysis Ready Data (ARD) is no longer a desire of global users, but is now becoming a requirement and an expectation
The Open Data Cube is constantly promoting the use of ARD in its global data cubes, as nearly all of those cubes use ARD.
ARD is critical for stacking interoperable data and performing time series change analyses or interoperable products (e.g. water extent). The ODC team is working these concepts.
The primary issue with ARD is the cost and complexity of producing the ARD ... but things are improving
CEOS, through the LSI-VC, is making great strides in the development of ARD definitions and guidelines and the world is taking notice!

4. Optical ARD Progress Landsat
USGS is moving toward a new ”tile based” system that will produce non-overlapping ARD for easier analysis and ingestion into data cubes. In addition, the data will be hosted on the AWS cloud for easier access. This is all coming in late 2018.
Sentinel-2
ESA-EC has started systematic production of surface reflectance products for the globe. Europe is first, but the rest of world will be ready in late 2018.
Mirror archives at USGS and other hosts (e.g. AWS, Google) will allow much better access to ARD for non-Europe users.

5. Sentinel-1 ARD Progress
ESA/EC does not currently produce simple ARD (gamma nought intensity) from GRD data products. Due to the large file sizes (1GB each), this is a huge burden for common users.
Global production of ARD for SLC data (includes interferometric phase information) is even more difficult due to its complexity and choices in processing.
Google Earth Engine has created simple ARD from GRD data and provides gamma-nought intensity to its users. The SEO has recently developed a script (with support from Google) that will generate a GeoTIFF export and then a data cube from these data holdings. This has worked well for rapid creation of small S1 data cubes.
The SEO is working with e-goes on an approach to rapidly create S1 ARD Data Cubes on the Amazon Cloud (next chart).

6. ALOS ARD Progress
JAXA produces annual PALSAR mosaics (25 meter) ARD products. These cannot be used for short-term time series analyses, but can be used for annual change detection.
JAXA has begun processing the PALSAR stripmap and PALSAR-2 ScanSAR data on a supercomputer and aims to have that completed and publicly available by the end of Stripmap PALSAR-2 is still under discussion.
Global users desire L-band data for forest and agriculture change detection. If such data was available, even at 30-meter scale, it would be a significant global impact.

7. AWS Sentinel-1 Working Group
Amazon is considering hosting S1 ARD as part of their AWS Public Datasets.
The new AWS Sentinel-1 Working Group is tasked with defining the backscatter intensity data product.
The group includes representatives from DigitalGlobe, Radiant Earth Foundation, e-GEOS, NASA CEOS (Brian Killough), Sinergise, Environment Systems, Earth Big Data, SkyWatch, Harris.
Sinergise is offering to process all of the GRD data into COG-format gamma-nought backscatter.
e-geos is investigating processing of SLC data for a 1 to 2 month global rolling archive on AWS.
In the end, they will look to CEOS for our community-driven radar PFS definitions!

8. e-goes Collaboration
e-geos is a commercial company in Rome, Italy. They have managed all of the COSMO-SkyMed data since They currently hold several ESA contracts to exploit Sentinel data.
Domenico Grandoni (e-geos) has demonstrated the CUBEO tool to pre- process raw Sentinel data to ARD format. They currently use SciHub and Copernicus-Hub for their data, but plan to move to AWS in the future.
The CUBEO dashboard selects an AOI, time range, dataset (S1 or S2) and then produces ARD. Output can be viewed and explored via WCS. They are proposing to expand this tool as an “ESA Service” via an ESA tender.
They would like to add an ”ODC format” pipeline option and the potential to push the products to the AWS cloud. This concept will be tested and explored in in collaboration with the SEO.

9. ARD PFS Feedback
The new ceos.org/ard site is an excellent resource and a useful location to capture the ARD objectives and documentation. I have used it often when communicating with global Data Cube and data provider contacts.
Expansion of the radar PFS to include polarimetric decomposition and interferometric coherence would be an excellent addition. These ARD products are complex to produce (use SLC data) but are valuable for land classification and change detection.
Be sure to check out the new “Layman’s Interpretation Guide to L-band and C- band Synthetic Aperture Radar Data” at “Resources” link of the ARD website. This report was funded by the SEO and produced by Ake Rosenqvist. The report will greatly promote the use of radar ARD.
Promotion of ARD at various conferences has also been very productive. The community is seeing the impact of our work and recognizing the value of a common definition.