Some next steps

[cboettig]

• all .json catalog entries should be converted into .R versions that generate (and validate) them. (see example catalog.R)

• base URLs of all href values in links section need to be mapped to use https://github.com/eco4cast/neon4cast-catalog/raw/main/stac/ instead of https://projects.ecoforecast.org/neon4cast-catalog/stac path. Note: it would probably be good to do this in the R code so that the base URL is set as a variable once and re-used, rather than having the base URL always repeated.

• Any links to asset parquet should adopt the s3:// notation shown in

  neon4cast-catalog/stac/aquatics/models/models.R

  Lines 50 to 53 in 2df8259

  	"href"= paste0("s3://anonymous@",
  	"bio230014-bucket01/neon4cast-forecasts/parquet/",
  	"aquatics/model_id=", model_id,
  	"?endpoint_override=sdsc.osn.xsede.org"),

• Flush out collection JSON files with descriptions, metadata on tables, etc

• create scores.R / scores.json stac item JSON (similar to models template like cb_prophet.json), with asset element pointing to s3:// scores notation

Example directory layout

stac
├── aquatics
│   ├── forecasts
│   │   ├── aquatics.json
│   │   └── models
│   │       ├── asset-description.Rmd
│   │       ├── cb_prophet.json
│   │       └── models.R
│   └── scores
│       ├── scores.json
│       └── scores.R
├── beetles
│   └── beetles.json
├── catalog.json
├── catalog.R
├── noaa
│   └── noaa.json
├── phenology
│   └── phenology.json
└── terrestrial
    └── terrestrial.json

[rqthomas]

I think we want https://raw.githubusercontent.com/eco4cast/neon4cast-catalog/main/stac/catalog.json as the raw GitHub address.

[cboettig]

I think we're all good here so I'll close.