[WIP] Burn severity maps using Sentinel-2 L2A to fix #114

[annefou]

Original jupyter notebook using Element 84 STAC catalog (will change to EOPF as soon as data is available).

[review-notebook-app]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB

[clausmichele]

@annefou do you have an estimate of the time required to finish this?

[konstntokas]

@annefou
I don't think that the data will become available, since you are requesting data from 2020, but EOPF Sample Service will support only a rolling archive of the last year. So you would need to select a wilf fire event from this year maybe. Just tested your example but for this year, and it seems to be working.

from xcube.core.store import new_data_store
from xcube_eopf.utils import reproject_bbox

store = new_data_store("eopf-zarr")
bbox_4326 = [-8.375, 41.625, -7.875, 41.875]
crs_utm = "EPSG:32629"
bbox_utm = reproject_bbox(bbox_4326, "EPSG:4326", crs_utm)
ds = store.open_data(
    data_id="sentinel-2-l2a",
    bbox=bbox_utm,
    time_range=["2025-08-06", "2025-08-20"],
    spatial_res=10,
    crs=crs_utm,
    variables=["b08", "b12"],
    query={"eo:cloud_cover": {"lt": 20}},
)
print(ds)

[annefou]

Ok, so you mean that if I select a wildfire from 2025, it should work? Then I’ll do that.

[konstntokas]

@annefou please have a look at the coverage maps; they may help to find a suitable example.

@christophreimer can you post the link to the coverage maps; they should be available for the organization members right?

[clausmichele]

@annefou @konstntokas the Grafana dashboard with the coverage map is available using the GitHub account here: https://grafana.infrastructure.eopf.eodc.eu/d/feqi81bevnxfkd/stac-eopf-coverage-and-statistics?orgId=1&from=now-2d&to=now&timezone=utc&var-collection=sentinel-2-l2a

[clausmichele]

@annefou you requested the data to @senmao here: EOPF-Sample-Service/eopf-sample-data#12 (comment)

So if it's not available on STAC there is probably be an issue.

[konstntokas]

Checked again, it is an issue with v2.6.0, the STAC items are available; see example item. The property "eo:cloud_cover" is missing since v2.6.0. @mario-winkler are you already aware of this? I think this property is important for many users who would like to filter for cloudy days.

The following code

from xcube.core.store import new_data_store
from xcube_eopf.utils import reproject_bbox

store = new_data_store("eopf-zarr")
bbox_4326 = [-8.375, 41.625, -7.875, 41.875]
crs_utm = "EPSG:32629"
bbox_utm = reproject_bbox(bbox_4326, "EPSG:4326", crs_utm)
ds = store.open_data(
    data_id="sentinel-2-l2a",
    bbox=bbox_utm,
    time_range=["2020-08-06", "2020-09-06"],
    spatial_res=10,
    crs=crs_utm,
    variables=["b08", "b12"],
    # query={"eo:cloud_cover": {"lt": 20}},
)
print(ds)

finds the items, but then breaks because "grid:code" is not in the properties in v2.6.0 which will be fixed. See EOPF-Sample-Service/xcube-eopf#41.

So best would be to find data from the general collection, because these are in v2.5.6.

[mario-winkler]

The property "eo:cloud_cover" is missing since v2.6.0.

@konstntokas Yes I am aware of this. There is also a CPM issue here

From the STAC perspective I would recommend to not use products which were converted with 2.6.0 at all.

[clausmichele]

@annefou are there any updates?

[annefou]

My understanding is that it will be fixed so I thought it is better to wait. The alternative is to take another date, for instance in 2025.