Cannot work with line exposures and hazards outside of CRS EPSG:4326

[mncosta]

Describe the bug
When trying to assign centroids of hazards to exposure objects from lines, if they are in a CRS that uses meters as units, function is likely to fail. From the check it is performed in "_nearest_neighbor_antimeridian" I assume it would only work when in a "standard" CRS (e.g., epsg:4326).

To Reproduce
Steps to reproduce the behavior/error (code based on the example from the tutorial:

1. Have exposure object created from lines
2. Have hazard object
3. Have both in a meters-based CRS (e.g., "epsg:28992")
4. Try to assign centroids

Code example:

# Imports
import geopandas as gpd
import pandas as pd
from pathlib import Path

from climada.entity import Exposures
from climada.entity.impact_funcs.storm_europe import ImpfStormEurope
from climada.entity.impact_funcs import ImpactFuncSet
from climada.engine import ImpactCalc
from climada.hazard.storm_europe import StormEurope
import climada.util.lines_polys_handler as u_lp
from climada.util.constants import DEMO_DIR, WS_DEMO_NC

def gdf_lines():
    gdf_lines = gpd.read_file(Path(DEMO_DIR,'nl_rails.gpkg'))
    gdf_lines = gdf_lines.to_crs(epsg=4326).to_crs(epsg=28992) # Reproject to EPSG:28992 Amersfoort / RD New -- Netherlands - Holland - Dutch
    return gdf_lines

# Create a GeoDataFrame with lines (railways) in the Netherlands
exp_nl_lines = Exposures(gdf_lines())
exp_nl_lines.gdf['impf_WS'] = 1
exp_nl_lines.gdf['value'] = 1
exp_nl_lines.gdf.head()

# define hazard
storms = StormEurope.from_footprints(WS_DEMO_NC)
# define impact function
impf = ImpfStormEurope.from_welker()
impf_set = ImpactFuncSet([impf])

# Reproject the hazard to match the exposures CRS
storms.reproject_vector(exp_nl_lines.crs)

# Calculate the impact
imp_deg = u_lp.calc_geom_impact(
    exp=exp_nl_lines, impf_set=impf_set, haz=storms,
    res=100, disagg_met=u_lp.DisaggMethod.DIV, disagg_val=None,
    agg_met=u_lp.AggMethod.SUM
)

# Or alternatively, you can use the below code. It fails in the "assign_centroids".
# exp_pnt = u_lp.exp_geom_to_pnt(
#     exp=exp_nl_lines, 
#     res=100,
#     to_meters=False,
#     disagg_met=u_lp.DisaggMethod.DIV,
#     disagg_val=None
#     )

# exp_pnt.assign_centroids(storms)

Expected behavior
Centroids are assigned to disaggregated exposure lines.

Screenshots
Attached.

Climada Version: 6.0.1

System Information (please complete the following information):

• Operating system and version: Ubuntu 24.04
• Python version: 3.11

Additional observations
To bypass the bug, one can set check_antimeridian to False in the function call, but, the parameter itself cannot be set when doing assign_centroids as kwargs are not passed further down the chain. They become available at some point, but not all the way up to assign_centroids.

[chahank]

@luseverin : is this related to the PR #1020 ?

[peanutfun]

match_coordinates supports kwargs that are forwarded to the respective nearest-neighbor function. These would have to be passed up to match_centroids and assign_centroids. Another option is to automatically detect if the antimeridian needs to be checked (usually when crs.is_geographic == True)

[luseverin]

@mncosta, thanks for reporting! This is indeed a known issue and is currently being taken care of in PR #1020.
I just implemented a fix for this issue, where I turned off the check_antimeridian for non-degree units (thanks @peanutfun for the suggestion). If you now check out the branch feature/remove_implicit_use_crs

cd climada_python
git pull
git checkout feature/remove_implicit_use_crs

you should be able to run your impact calculation using your crs defined in meters. The unit of your coordinate system has been automatically inferred (unit defaults to meters when coords.crs.is_projected == True) allowing the centroids assignement in the inputted crs. Note that this is still under review so there might still be some bugs or issues. Any feedback on this is then welcome!