Invalid time values

[olafveerman]

https://earthdata.nasa.gov/covid19/explore/sf?map=-122.3218%2C37.7334%2C11.63&layers=no2-diff&lState=nightlights-hd%7C0%7C0%2Cdetections-ship%7C0%7C0%2Cno2-diff%7C0%7C0

results in:

• NO2 Diff
• Agriculture

[olafveerman]

For agriculture, the API returns an empty array for the date domain.

For NO2 Diff, the tile url gives a 500 internal server error. The legend is of type gradient, the tile url has a gamma value. These are incompatible. Change tile url to remove gamma.

[olafveerman]

Tracking NO2 in a separate issue.

Config for agriculture returned from the API:

{
    "id": "agriculture",
    "name": "Agriculture",
    "type": "raster-timeseries",
    "isPeriodic": true,
    "timeUnit": "month",
    "domain": [],
    "source": {
      "type": "raster",
      "tiles": [
        "https://8ib71h0627.execute-api.us-east-1.amazonaws.com/v1/{z}/{x}/{y}@1x?url=s3://covid-eo-data/agriculture-cropmonitor/CropMonitor_{date}.tif&resampling_method=nearest&bidx=1&color_map=custom_cropmonitor"
      ]
    },
    ...
}

Originally I thought this was due to the date formatting of the COG, which does not follow the convention of the other monthly datasets. But even after changing all the filenames on S3, the domain is still empty.

CropMonitor_2020_10_28.tif -> CropMonitor_202010.tif

[leothomas]

Agriculture looks for data in s3 subfolder s3://covid-eo-data/agriculture-cropmonitor (as opposed to s3://covid-eo-data/agriculture). The reason for this was because originally, the s3://covid-eo-data/agriculture/ folder had a bunch of different files (other can Cropmonitor* which were causing invalid dates to be returned, so I moved all agriculture/Cropmonitor* files to agriculture-cropmonitor. See covid-api, issue 53

Two possible fixes:

1. Now that the agriculture/ folder has been cleaned up to contain only Cropmonitor* files, we update the s3_location of the dataset's metadata to point to agriculture instead of agriculture-cropmonitor

• Requires no moving of files in S3
• Requires a re-deploy of the API

2. Copy all files in agriculture to agriculture-cropmonitor as a temp fix

• Requires no redeploy of API
• Requires duplicating (temporarily) files in S3

[olafveerman]

🤦 I missed the -cropmonitor. Files are moved and renamed. The agriculture layer is working again.

Thanks @leothomas