Wind input to wildfire spread prediction in complex terrain: the regional wind sets the direction, and errors concentrate in the extremes

Derived data and code that reproduce the numbers, Tables 1–4 and Figs. 2–5 of: Min S (submitted) Wind input to wildfire spread prediction in complex terrain: the regional wind sets the direction, and errors concentrate in the extremes. Natural Hazards. Contents - data/derived: evaluation tables for about 100 wildfires in South Korea (2015–2025) — wind errors at withheld stations for three wind inputs (station-initialised WindNinja, single-value WindNinja, interpolation), and observed and predicted initial spread directions for 21 fires. - code/verify_numbers.py: recomputes every number in the manuscript and reports OK or DIFF for each claim. - code/make_figures.py: regenerates Figs. 2–5. Usage (from the package root) pip install -r requirements.txt python code/verify_numbers.py python code/make_figures.py Column definitions are in DATA_DICTIONARY.md. Code: MIT licence. Data: CC BY 4.0.