Risk indicators — quickstart (live)¶

The risk-indicators backend pulls country/admin-indexed risk screening from three sources behind one interface:

• GFDRR ThinkHazard! — a hazard rating (Very low → High) per hazard type and admin division; public, no key.
• INFORM Risk (JRC) — the composite humanitarian-risk index and its sub-dimensions, scored 0–10 per country; public, no key.
• Global Forest Watch Data API — forest indicators and admin geometry; needs a free API key (covered in the Global Forest Watch notebook).

Unlike the raster backends there is no bounding box or grid — you name a dataset id in variables= and a country with country= (an ISO3 code), and you get back a tidy pandas.DataFrame. This notebook covers the two public sources (ThinkHazard! and INFORM), fetched live.

In [ ]:

import tempfile

from earthlens.core import EarthLens

OUT = tempfile.mkdtemp()  # tabular downloads also write a CSV here

import tempfile from earthlens.core import EarthLens OUT = tempfile.mkdtemp() # tabular downloads also write a CSV here

ThinkHazard! — one hazard for a country¶

variables=["thinkhazard:flood_river"] selects the river-flood rating; country="KEN" names the country as an ISO3 code. The backend resolves that ISO3 to ThinkHazard's numeric ADM0 division code and returns one row with the qualitative level / level_title.

In [ ]:

flood = EarthLens(
    data_source="risk-indicators",
    variables=["thinkhazard:flood_river"],
    country="KEN",
    path=OUT,
).download()

flood

flood = EarthLens( data_source="risk-indicators", variables=["thinkhazard:flood_river"], country="KEN", path=OUT, ).download() flood

The level is ThinkHazard's coded rating (VLO/LOW/MED/HIG) and level_title its human label; admin_code is the resolved ADM0 code (133 for Kenya). download() also wrote the row to a CSV under path.

All eleven hazards at once¶

The thinkhazard:all dataset returns every hazard ThinkHazard! screens for the division in a single call — handy for a quick risk profile.

In [ ]:

profile = EarthLens(
    data_source="risk-indicators",
    variables=["thinkhazard:all"],
    country="KEN",
    path=OUT,
).download()

print("hazards:", len(profile))
profile

profile = EarthLens( data_source="risk-indicators", variables=["thinkhazard:all"], country="KEN", path=OUT, ).download() print("hazards:", len(profile)) profile

INFORM Risk — a composite country score¶

inform:risk is the headline INFORM composite index (0–10). The "inform" alias and the "risk-indicators" key both reach the same backend, so either works as data_source=.

The first call downloads JRC's release workbook (a few MB) and caches it, so the later INFORM cells reuse it rather than re-downloading.

In [ ]:

risk = EarthLens(
    data_source="inform",
    variables=["inform:risk"],
    country="KEN",
    path=OUT,
).download()

risk

risk = EarthLens( data_source="inform", variables=["inform:risk"], country="KEN", path=OUT, ).download() risk

One row per country: indicator_score is the 0–10 composite (higher = more at risk), iso3 the country. The last three columns record where the numbers came from: source is release (JRC's published workbook, the default and the current release) or api, workflow_id names the model release an API row used, and validity_year carries the workbook's release year — the API leaves it at 0. Pass source="api" or workflow_id= to switch.

Omit country= and INFORM returns every country — the whole global table in one frame, ready to rank or join.

In [ ]:

world = EarthLens(
    data_source="inform",
    variables=["inform:risk"],
    path=OUT,
).download()

print("countries:", len(world))
world.sort_values("indicator_score", ascending=False).head(5)

world = EarthLens( data_source="inform", variables=["inform:risk"], path=OUT, ).download() print("countries:", len(world)) world.sort_values("indicator_score", ascending=False).head(5)

The country= selector¶

country= is always an ISO3 code ("KEN", "NGA", "BRA", …). What happens to it depends on the source:

• ThinkHazard! resolves the ISO3 to a GAUL 2015 ADM0 division code (the numeric key its API uses) via the catalog's bundled admin_codes: table. For a sub-national division, pass a raw admin_code= instead.
• INFORM filters its global table to that one ISO3 (omit it for all countries, as above).
• GFW uses the ISO3 as the SQL / geostore key (see the GFW notebook).

Takeaway¶

• EarthLens(data_source="risk-indicators", variables=[<id>], country=<ISO3>) returns a tidy DataFrame — no bbox, no grid.
• ThinkHazard! gives qualitative hazard ratings per division (thinkhazard:all for the full profile); INFORM gives 0–10 composite scores (omit country= for the whole world).
• Both are public and keyless. The third source, Global Forest Watch, needs a free key — see the catalog & behaviour notebook for the full dataset list and the GFW notebook for the keyed forest indicators.