Upgrade titiler-pgstac

[j08lue]

Large backlog

TiTiler-PgSTAC is now at 0.8.0 and we are still at

veda-backend/raster_api/runtime/setup.py

Line 9 in e4b02a5

"titiler.pgstac==0.2.3",

This is blocking a number of features that were implemented upstream in TiTiler, which we need in VEDA, most notably

• Compute area-weighted averages #226

Complications

There have been some breaking API changes in TiTiler between the latest version and what we have, so any update will need to be coordinated with dependants - the VEDA Dashboard and the EO Dashboard.

Acceptance criteria

• Mapped out breaking changes
• Deployed to a "staging" stack for the Dashboard(s) to build against
• Upgraded TiTiler-PgSTAC

[vincentsarago]

Most of the breaking changes (Some of them only concern the main application of titiler-pgstac so it might not be relevant):

• Use /collections/{collection_id}/items/{item_id} prefix for Item endpoint.

# Before
{endpoint}/stac/info?collection=collection1&item=item1

# Now
{endpoint}/collections/collection1/items/item1/info

• change tile url path parameter order from /tiles/{searchid}/{TileMatrixSetId}/{z}/{x}/{y} to /{searchid}/tiles/{TileMatrixSetId}/{z}/{x}/{y}

# Before
{endpoint}/mosaic/tiles/20200307aC0853900w361030/0/0/0

# Now
{endpoint}/mosaic/20200307aC0853900w361030/tiles/0/0/0

• replace /{searchid}/{z}/{x}/{y}/assets endpoints by /{searchid}/tiles/{z}/{x}/{y}/assets

• use /api and /api.html for documentation (instead of /openapi.json and /docs)

• add landing page /

• replace - by _ in query parameters

  • coord-crs -> coord_crs
  • dst-crs -> dst_crs

• remove max_size default for mosaic's /statistics [POST] endpoint breaking change

• add /bbox and /feature [POST] optional endpoints

[anayeaye]

I created a test stack with the new titiler and a copy of the staging database that I needed for testing the veda-docs
https://test-stac.delta-backend.com/
https://test-raster.delta-backend.com/

[hanbyul-here]

Adding more changes that were commented on the original PR (Thanks @anayeaye for thoroughly going through the changes!)

• Use algorithm instead of post_process: feat(raster)!: update titiler-pgstac from 0.2.3 to 0.8.0 #239 (comment)

[anayeaye]

Veda-config PR with algorithm change.

[j08lue]

We need another version upgrade for rio-tiler to include this bug fix, please:

• validate shape in coverage cogeotiff/rio-tiler#653

This is hopefully the last missing change to enable us to do equal-area zonal statistics. 🚀

Happening here:

• feat: update titiler and titiler-pgstac requirement #246

[vincentsarago]

Note:

any re-deploy will include the new rio-tiler fix (because rio-tiler is not pin to specific version 🙏)

#246 will be nice to have if we want to use the custom algorithm instead of eqa projection!

[anayeaye]

@j08lue @slesaad ghgc-dev(smce) and veda-dev are both running rio-tiler 6.2.6 (w/validate shape in coverage)

[j08lue]

Awesome, @anayeaye and @vincentsarago, this now works:

import requests

RASTER_API_URL = "https://dev.ghg.center/api/raster"

ITEM_ASSET_URL = "s3://veda-data-store-staging/no2-monthly/OMI_trno2_0.10x0.10_202109_Col3_V4.nc.tif"

AOI = {
  "type": "FeatureCollection",
  "features": [
    {
      "type": "Feature",
      "properties": {},
      "geometry": {
        "coordinates": [
          [
            [
              -115,
              82
            ],
            [
              -115,
              -82
            ],
            [
              -43,
              -82
            ],
            [
              -43,
              82
            ],
            [
              -115,
              82
            ]
          ]
        ],
        "type": "Polygon"
      }
    }
  ]
}

result = requests.post(
    f"{RASTER_API_URL}/cog/statistics",
    params={
        "url": ITEM_ASSET_URL,
        "dst_crs": "+proj=cea"
    },
    json=AOI["features"][0],
)
result.raise_for_status()
result.json()

and gives different results when I set dst_crs or not.

[j08lue]

I can also confirm that the results of the case with +proj=cea agree with the weighted average calculated with the scientist-provided method: