Criterion B Details

# Default ecosystem code for template development.
# This line is replaced by build_ecosystem_pages.py for each ecosystem.
ecosystem_code = 'Desierto'

Import Python modules.

import os
import yaml
from pathlib import Path
from lonboard import Map
from rle.core import Ecosystems, criterion_b_status, rle_category
from rle.core.eoo import make_eoo
from rle.core.aoo import make_aoo_grid

Load the country config file.

project_root = os.environ.get('PIXI_PROJECT_ROOT', str(Path('..').resolve()))
config_path = Path(project_root) / 'config' / 'country_config.yaml'
with open(config_path) as f:
    config = yaml.safe_load(f)

# Ecosystem index (the COG pixel value for this ecosystem), looked up by code
# from the canonical index table. Absent if the index has not been built yet.
import csv
ecosystem_index = None
index_csv = Path(project_root) / 'config' / 'ecosystems' / 'index.csv'
if index_csv.exists():
    with open(index_csv) as f:
        for row in csv.DictReader(f):
            if row['code'] == ecosystem_code:
                ecosystem_index = int(row['index'])
                break

Load & Filter Ecosystem Data

Load data for all the ecosystems.

import sys
# Shared config helpers live in scripts/; make them importable from here.
sys.path.insert(0, str(Path(project_root) / 'scripts'))
from _config import ensure_vector_source

source = config['ecosystem_source']
# ecosystem_code_column is optional: fall back to the name column.
ecosystem_column = source.get('ecosystem_code_column') or source.get('ecosystem_name_column')
# Prefer the ecosystem-sorted `optimized_data` copy when configured, so that
# filtering to one ecosystem uses parquet predicate pushdown instead of loading
# the whole national map into memory. Falls back to `data`.
ecosystems = Ecosystems.from_file(
    ensure_vector_source(source.get('optimized_data') or source['data']),
    ecosystem_column=ecosystem_column,
    ecosystem_name_column=source.get('ecosystem_name_column'),
    functional_group_column=source.get('functional_group_column')
)

Filter by the Desierto and check the number of features.

ecosystem = ecosystems.filter(ecosystem_code)
has_data = ecosystem.size() > 0
print(f'{ecosystem.size() = }')
if not has_data:
    from IPython.display import Markdown, display
    display(Markdown(
        f'**No spatial data found for {ecosystem_code}.** '
        f'Criterion B calculations are skipped.'
    ))

ecosystem.size() = 1615

Extent of Occurrence (EOO) (subcriterion B1)

Extent of occurrence (EOO). The EOO of an ecosystem is the area (km2) of a minimum convex polygon – the smallest polygon in which no internal angle exceeds 180° that encompasses all known current spatial occurrences of the ecosystem type.

The minimum convex polygon (also known as a convex hull) must not exclude any areas, discontinuities or disjunctions, regardless of whether the ecosystem can occur in those areas or not. Regions such as oceans (for terrestrial ecosystems), land (for coastal or marine ecosystems), or areas outside the study area (such as in a different country) must remain included within the minimum convex polygon to ensure that this standardised method is comparable across ecosystem types. In addition, these features contribute to spreading risks across the distribution of the ecosystem by making different parts of its distribution more spatially independent.

Calculate EOO

Start by calculating the convex hull of the ecosystem’s distribution.

import geopandas as gpd

if has_data:
    ecosystem_geometry = ecosystem.geometry.union_all()
    gdf_ecosystem_polygons = gpd.GeoDataFrame(geometry=[ecosystem_geometry], crs=ecosystem.geometry.crs)
    hull = ecosystem_geometry.convex_hull
    gdf_hull = gpd.GeoDataFrame(geometry=[hull], crs=ecosystem.geometry.crs)

Display the ecosystem’s distribution and the convex hull.

from lonboard import Map, PolygonLayer
from rle.core.viz import smart_map

if has_data:
    eoo_hull = make_eoo(ecosystem).compute()
    display(smart_map([eoo_hull, ecosystem]))

/home/runner/work/rle-tyler-colombia/rle-tyler-colombia/.pixi/envs/default/lib/python3.11/site-packages/lonboard/_geoarrow/ops/reproject.py:116: UserWarning: Input being reprojected to EPSG:4326 CRS.
Lonboard is only able to render data in EPSG:4326 projection.
  warnings.warn(

if has_data:
    hull_ea = gdf_hull.to_crs("ESRI:54034")
    eoo = hull_ea.geometry.iloc[0].area / 1e6
    print(f'EOO is {eoo:.1f} km2')

EOO is 10956.1 km2

Then calculate the area of the convex hull polygon.

Direct calculation of EOO

EOO can also be calculated directly using …

if has_data:
    ecosystem.eoo

Verify that the area returned by calling make_eoo(ecosystem).compute().area_km2 is the same as the area of the convex hull polygon.

if has_data:
    assert ecosystem.eoo == eoo

Area of Occupancy (AOO) (subcriterion B2)

The protocol for this adjustment includes the following steps:

Intersect AOO grid with the ecosystem’s distribution map.

Calculate extent of the ecosystem type in each grid cell (area) and sum these areas to obtain the total ecosystem area (total area).

Arrange grid cells in ascending order based on their area (smaller first). Calculate accumulated sum of area per cell (cumulative area).

Calculate cumulative proportion by dividing cumulative area by total area (cumulative proportion takes values between 0 and 1)

Calculate AOO by counting the number of cells with a cumulative proportion greater than 0.01 (i.e. exclude cells that in combination account for up to 1% of the total mapped extent of the ecosystem type).

AOO Calculation Details

Intersect AOO grid and ecosystem map

Intersect AOO grid with the ecosystem’s distribution map

from pathlib import Path
from rle.core.aoo import make_aoo_grid_cached

if has_data:
    # Prefer a prebuilt grid cache (e.g. a gs:// URI) so the national AOO grid
    # is not recomputed during CI renders — computing it from the full national
    # ecosystem map peaks at many GB of RAM. Fall back to a local cache when no
    # prebuilt cache is configured. Build one with `pixi run build-caches`.
    cache_path = (
        source.get('aoo_grid_cache_url')
        or (Path(project_root) / '.cache' / 'aoo_grid.parquet')
    )
    aoo_grid = make_aoo_grid_cached(ecosystems, cache_path=cache_path)
    aoo_grid_filtered = aoo_grid.filter_by_ecosystem(ecosystem_code)

Visualize variations in the AOO grid.

from matplotlib.colors import LinearSegmentedColormap
from lonboard.colormap import apply_continuous_cmap
from rle.core.aoo import slugify_ecosystem_name

ecosystem_column = slugify_ecosystem_name(ecosystem_code)
if has_data:
    cmap = LinearSegmentedColormap.from_list("white_red", ["white", "red"])
    values = aoo_grid_filtered.grid_cells[ecosystem_column].values
    normalized = (values - values.min()) / (values.max() - values.min())
    colors = apply_continuous_cmap(normalized, cmap)
    display(smart_map([(aoo_grid_filtered, {"get_fill_color": colors}), ecosystem]))

Calculate grid cell area and total area

Calculate extent of the ecosystem type in each grid cell (area) and sum these areas to obtain the total ecosystem area (total area).

if has_data:
    keep = ['geometry', 'grid_col', 'grid_row', ecosystem_column]
    gdf = aoo_grid_filtered.grid_cells[keep]
    display(gdf)

	geometry	grid_col	grid_row	Desierto
0	POLYGON ((-72.67371 11.56174, -72.67371 11.654...	-810	127	0.042463
1	POLYGON ((-72.67371 11.65401, -72.67371 11.746...	-810	128	0.045059
2	POLYGON ((-72.58387 11.56174, -72.58387 11.654...	-809	127	0.017003
3	POLYGON ((-72.58387 11.65401, -72.58387 11.746...	-809	128	0.033131
4	POLYGON ((-72.49404 11.56174, -72.49404 11.654...	-808	127	0.005736
...	...	...	...	...
109	POLYGON ((-71.14657 12.02342, -71.14657 12.115...	-793	132	0.119539
110	POLYGON ((-71.14657 12.11584, -71.14657 12.208...	-793	133	0.044757
111	POLYGON ((-71.14657 12.2083, -71.14657 12.3007...	-793	134	0.052335
112	POLYGON ((-71.05674 12.02342, -71.05674 12.115...	-792	132	0.104662
113	POLYGON ((-71.05674 12.11584, -71.05674 12.208...	-792	133	0.013462

114 rows × 4 columns

The column Desierto contains the (fractional) area of the ecosystem in each grid cell.

Sum up the areas of each grid cell to get the total area.

if has_data:
    total_area = gdf[ecosystem_column].sum()
    display(total_area)

np.float64(18.354081153884348)

Calculate cumulative area

Arrange grid cells in ascending order based on their area (smaller first). Calculate accumulated sum of area per cell (cumulative area).

if has_data:
    gdf = gdf.sort_values(by=ecosystem_column)
    gdf["cumulative_area"] = gdf[ecosystem_column].cumsum()
    display(gdf)

	geometry	grid_col	grid_row	Desierto	cumulative_area
12	POLYGON ((-72.31438 11.83865, -72.31438 11.931...	-806	130	0.000290	0.000290
26	POLYGON ((-72.04489 11.37728, -72.04489 11.469...	-803	125	0.000315	0.000605
23	POLYGON ((-72.13472 12.02342, -72.13472 12.115...	-804	132	0.003328	0.003933
71	POLYGON ((-71.59573 11.65401, -71.59573 11.746...	-798	128	0.003806	0.007739
7	POLYGON ((-72.40421 11.65401, -72.40421 11.746...	-807	128	0.004220	0.011959
...	...	...	...	...	...
78	POLYGON ((-71.59573 12.30079, -71.59573 12.393...	-798	135	0.455016	16.186432
97	POLYGON ((-71.32623 11.83865, -71.32623 11.931...	-795	130	0.464979	16.651412
86	POLYGON ((-71.5059 12.30079, -71.5059 12.39331...	-797	135	0.480255	17.131666
90	POLYGON ((-71.41607 11.93102, -71.41607 12.023...	-796	131	0.604898	17.736565
82	POLYGON ((-71.5059 11.93102, -71.5059 12.02342...	-797	131	0.617517	18.354081

114 rows × 5 columns

Calculate cumulative proportion

Calculate cumulative proportion by dividing cumulative area by total area (cumulative proportion takes values between 0 and 1)

if has_data:
    gdf["cumulative_proportion"] = gdf["cumulative_area"] / total_area
    display(gdf)

	geometry	grid_col	grid_row	Desierto	cumulative_area	cumulative_proportion
12	POLYGON ((-72.31438 11.83865, -72.31438 11.931...	-806	130	0.000290	0.000290	0.000016
26	POLYGON ((-72.04489 11.37728, -72.04489 11.469...	-803	125	0.000315	0.000605	0.000033
23	POLYGON ((-72.13472 12.02342, -72.13472 12.115...	-804	132	0.003328	0.003933	0.000214
71	POLYGON ((-71.59573 11.65401, -71.59573 11.746...	-798	128	0.003806	0.007739	0.000422
7	POLYGON ((-72.40421 11.65401, -72.40421 11.746...	-807	128	0.004220	0.011959	0.000652
...	...	...	...	...	...	...
78	POLYGON ((-71.59573 12.30079, -71.59573 12.393...	-798	135	0.455016	16.186432	0.881898
97	POLYGON ((-71.32623 11.83865, -71.32623 11.931...	-795	130	0.464979	16.651412	0.907232
86	POLYGON ((-71.5059 12.30079, -71.5059 12.39331...	-797	135	0.480255	17.131666	0.933398
90	POLYGON ((-71.41607 11.93102, -71.41607 12.023...	-796	131	0.604898	17.736565	0.966355
82	POLYGON ((-71.5059 11.93102, -71.5059 12.02342...	-797	131	0.617517	18.354081	1.000000

114 rows × 6 columns

Count AOO cells

Calculate AOO by counting the number of cells with a cumulative proportion greater than 0.01 (i.e. exclude cells that in combination account for up to 1% of the total mapped extent of the ecosystem type).

if has_data:
    aoo = len(gdf[gdf["cumulative_proportion"] > 0.01])
    print(f'AOO is {aoo} cells')

AOO is 95 cells

AOO Calculation (direct call)

if has_data:
    aoo_count = ecosystem.aoo
    print(f'AOO: {aoo_count} grid cells')

AOO: 95 grid cells

Criterion B Summary

Criterion B status (spatial)

Endangered (EN) — Desierto (Desierto), index 44

Status reflects the spatial thresholds for EOO (B1) and AOO (B2) only. A final listing under B1/B2 additionally requires at least one of: (a) an observed or inferred continuing decline; (b) threatening processes likely to cause continuing decline within 20 years; or (c) few threat-defined locations — none of which are derived from the spatial metrics.

Sub-criterion	Metric	Value	Category
B1	EOO	10956 km²	Endangered (EN)
B2	AOO	95 cells	Least Concern (LC)
Overall B	—	—	Endangered (EN)