Document the CCDC acquisition

[jordancaraballo]

Below are the documented steps from Matt, including links and steps. This should be documented somewhere so SRLite can be useful for someone that does not have a Matt to generate the CCDC files (or at least until we get it working with HLS).

I ran this from an ABR server and on GEE (we have a setup where GEE exports to google drive get automatically copied to our server).

First, I combined some information from Paul's TOA footprint gpkg and the NASA NGA inventory geodatabase, and calculated some extra fields including UTM zone. Result is saved to a shapefile that is then uploaded to GEE.

Then in GEE, I generate a CCDC model asset that covers the footprints of interest. I have one that covers all of Alaska, and three patches in Senegal. You can also calculate this CCDC on-the-fly for each footprint but it ends up being orders of magnitude slower. There should eventually be a precomputed global CCDC asset that you may be able to point to instead of generating it ourselves.

Then in GEE I extract modeled reflectance matching the footprint, date, and UTM zone of the TOA footprints. This is exported to google drive.

On the ABR server I fixed up the GEE exported geotiff (tagging the nodata value). Then I did some file management to organize the CCDC outputs into the expected folder structure.

The set then gets put into a tar.gz archive, copied to ADAPT, and unpacked.

Lastly I make it world readable on ADAPT.

The Rmd file includes links to the GEE script.

adapt_footprint_gpkg_to_ccdc_Senegal_20230309.Rmd.zip

[jordancaraballo]

In our workflow:

• we would not rely on Google Drive, and would use geedim to pull it locally to ADAPT
• run it from ADAPT instead of ABR servers
• fix the nodata values on the fly

[jordancaraballo]

Authentication from the CLI:

# get credentials
credentials = ee.ServiceAccountCredentials(gee_account, gee_key)
ee.Initialize(credentials)  # gd initialize does not take service account
print("Initialized")