climate-lama-engine 0.1.0

Lean probabilistic climate risk calculation engine

climate-lama-engine

Lean probabilistic climate risk calculation engine. Clean-room reimplementation of CLIMADA's core math — no geospatial dependencies, no I/O, just numpy and scipy.

Runtime deps: numpy>=1.24, scipy>=1.11 Python: >=3.10

---

Install

uv sync --group dev

Or with pip:

pip install -e ".[dev]"

---

Run tests

uv run pytest

---

Use on another machine

Clone the repo and sync the environment — the lockfile guarantees identical versions:

git clone <repo-url>
cd climate-lama-core
uv sync --group notebooks

Then open VS Code, create or open any notebook, and select the .venv inside climate-lama-core/ as the kernel. import climate_lama_engine as cc will work from any notebook on the machine, not just the ones in this repo.

To update later:

git pull
uv sync --group notebooks

---

Notebooks

Documentation and manual testing notebooks live in notebooks/.

Setup

uv sync --group notebooks
uv run jupyter notebook

Or with pip:

pip install -e .
pip install -r notebooks/requirements.txt
jupyter notebook

Then open any notebook from the notebooks/ folder in your browser.

Notebook overview

Notebook	Description
00_quickstart.ipynb	10-minute end-to-end walkthrough: hazard to EAD to measure
01_hazard.ipynb	Hazard deep dive: marginal frequencies, sparse matrix internals
02_exposures.ipynb	Exposures: centroid assignment, insurance parameters
03_impact_functions.ipynb	MDR, MDD, PAA -- why they are separate, JRC flood function
04_impact_calculation.ipynb	Core engine: all output attributes, identity checks, spatial risk map
05_measures.ipynb	Adaptation measures: intensity reduction, retrofits, early warning
06_cost_benefit.ipynb	Cost-benefit analysis: BCR, discount rates, risk metric sensitivity
07_flood_greece_pattern.ipynb	Realistic JRC flood scenario on a 50x50 Greece grid
08_climada_comparison.ipynb	Side-by-side verification against upstream CLIMADA

---

Quick example

import numpy as np
from scipy import sparse
import climate_lama_engine as cc

rps = np.array([10, 100, 500])
intensity = sparse.csr_matrix(np.array([
    [0.0, 0.5, 1.0, 0.0, 0.2],
    [0.0, 1.2, 2.5, 0.5, 0.8],
    [0.0, 2.0, 4.0, 1.0, 1.5],
]))
hazard = cc.Hazard.from_rp_maps(
    haz_type="RF", intensity_unit="m",
    return_periods=rps, intensity=intensity,
    centroid_lat=np.array([37.9, 38.0, 38.1, 38.2, 38.3]),
    centroid_lon=np.array([23.7, 23.8, 23.9, 24.0, 24.1]),
)

impf = cc.ImpactFunc(
    id=1, haz_type="RF",
    intensity=np.array([0.0, 0.5, 1.0, 2.0, 3.0, 5.0]),
    mdd=np.array([0.0, 0.02, 0.07, 0.25, 0.50, 0.80]),
    paa=np.array([0.0, 0.20, 0.40, 0.70, 0.90, 1.00]),
)
impfset = cc.ImpactFuncSet([impf])

exposures = cc.Exposures(
    value=np.array([500_000, 1_200_000, 800_000, 300_000]),
    centroid_idx=np.array([1, 2, 3, 4]),
    impf_id=np.array([1, 1, 1, 1]),
    value_unit="EUR",
)

result = cc.ImpactCalc(hazard, exposures, impfset).impact()
print(f"EAD: EUR {result.aai_agg:,.0f}")

levee = cc.Measure(name="Flood Levee", haz_type="RF", haz_inten_a=0.8, cost=2_000_000)
mod_haz, mod_impf = levee.apply(hazard, impfset)
result_with_levee = cc.ImpactCalc(mod_haz, exposures, mod_impf).impact()
print(f"EAD with levee: EUR {result_with_levee.aai_agg:,.0f}")

---

License

Apache 2.0