SI_LULC_pipeline notebook about eo-learn HOT 9 CLOSED

1. The description states that we took the shape from the Natural Earth (as a .shp), then loaded the shape, applied a 500m buffer and saved it. You can find it in the repo under example_data/svn_border.geojson
2. In the cell where the selection is visualized you can also see which cell the corresponding ID belongs to. The ID 616 was chosen since the area contains all of the main landcover classes. I just re-rean the notebook and don't get an error, did you perhaps change something?

from eo-learn.

Also, feel free to report any oddities you find, we're always keen to improve the examples

from eo-learn.

Thanks for your quick reply! Ah, cool. I actually did not even come so far in the tutorial yet and hence did see the selection visualisation.

One more thing: If I apply the code to another country (e.g. Germany) what do I need to change? The selection visualisation displays all bboxes shifted to the right of the actual country borders. I guess it has to do with the crs?

Thanks again!

from eo-learn.

Sounds like you should double-check that the CRS match, yes

from eo-learn.

Actually the CRS matches. Did not change anything compared to the Slovenia example.

country = country.to_crs(epsg=32633).buffer(500)

Plotting the bbox_gdf looks weird...

country crs:

`Derived Projected CRS: EPSG:32633
Name: WGS 84 / UTM zone 33N
Axis Info [cartesian]:

• E[east]: Easting (metre)
• N[north]: Northing (metre)
  Area of Use:
• name: Between 12°E and 18°E, northern hemisphere between equator and 84°N, onshore and offshore. Austria. Bosnia and Herzegovina. Cameroon. Central African Republic. Chad. Congo. Croatia. Czechia. Democratic Republic of the Congo (Zaire). Gabon. Germany. Hungary. Italy. Libya. Malta. Niger. Nigeria. Norway. Poland. San Marino. Slovakia. Slovenia. Svalbard. Sweden. Vatican City State.
• bounds: (12.0, 0.0, 18.0, 84.0)
  Coordinate Operation:
• name: UTM zone 33N
• method: Transverse Mercator
  Datum: World Geodetic System 1984 ensemble
• Ellipsoid: WGS 84
• Prime Meridian: Greenwich`

Country plot looks good:

from eo-learn.

The notebook uses UtmZoneSplitter so for large countries you'll get the bboxes with different CRS (corresponding to the appropriate UTM zone, see here). The notebook assumes that all the BBoxes share the same CRS when the geodataframe is constructed.

from eo-learn.

Thanks! And is there an easy way to circumvent this problem of different CRS and still get the same result?

from eo-learn.

Working over multiple CRS brings a whole set of issues. However for this example I believe you should only have problems when displaying results (the bbox_gdf seems to only be used for plotting purposes).
Another angle is also that here we focus on only 9 EOPatches, partially to conserve data, and also because we the LGBM model for predictions we don't need to have an enormous amount of data. So you can just focus on a sub-area of Germany in order to run this notebook.

from eo-learn.

Assuming it was resolved due to inactivity

from eo-learn.