Attribution of extreme weather events at UNESCO World Heritage Sites

Dataset description This dataset provides the results and Python codes used to quantify how human-driven climate change has altered extreme heat (Tx90p), heavy rainfall (R95p), and consecutive dry days (CDD) indices at UNESCO World Heritage Sites. Files included: CulturalWHS_Attribution.csv— A master table for 1,049 WHSs that are inscribed with cultural values. For each site it lists location, inscription details, World Bank income group, IPCC reference region, and computed changes in three extreme-weather indices (Tx90p, R95p, CDD): baseline change relative to preindustrial (1850-1900) under historical forcings (henceforth the "Historical"), natural-only forcings (henceforth the "Hist-NAT"), and SSP2-4.5 forcings. anthropogenic influence (historical forcings contrast natural-only forcings in 2000-2014). CulturalWHS_Attribution_codebook.csv — companion metadata table that defines every column in the master dataset for quick reference. Anthropogenic influence on extreme weather events.ipynb — Runs Steps 1–5: computes Tx90p, R95p, and CDD indices, computes baseline and anthropogenic changes, and draws the global and regional maps. Time series at 12 example WHSs.ipynb — Runs Step 6: creates the 30-year-smoothed time-series plots with uncertainty bands for the twelve World Heritage Sites that covers diverse heritage and geographic types. Data acquisition The calculations call external datasets: the MPI-GE climate ensemble (Olonscheck, 2023), the UNESCO World Heritage Site list, the World Bank income classifications, and the IPCC WGI Atlas reference-region shapefiles (IPCC, 2021). The World Heritage Sites list and attributes are acquired from World Heritage List (https://whc.unesco.org/en/list/xlsx/?2024). Data acquired in December, 2024. The climate data (daily maximum temperature and daily rainfall) are acquired from the Max Planck Institute Grand Ensemble through the Earth System Grid Federation platform (https://aims2.llnl.gov/search/cmip6/). Data acquired in September, 2024. The IPCC reference regions used for regional aggregation and visualisation were obtained from the publicly available GIS repository of the IPCC WGI Interactive Atlas (https://github.com/IPCC-WG1/Atlas/tree/main/data-sources), which provides polygon shapefiles and usage scripts. The income categories and GDP per capita data are acquired from The World Bank Country and Lending Groups (https://datahelpdesk.worldbank.org/knowledgebase/articles/906519-world-bank-country-and-lending-groups). Data acquired in December, 2024. Tabular data processing (in Microsoft Excel) For WHSs categorized as “transboundary”, retain only one entry per country per WHS (i.e., exclude duplicate rows from the same WHS ID). Merge World Bank income categories and GDP per capita data to WHS entries based on State Party (country) names. Computation and visualisation (in Python environment) All codes are written in Python 3.10 environment and use xarray, xclim, pandas, and matplotlib. Extreme indices: Compute Tx90p, R95p, and CDD under three scenarios (Historical, Hist-NAT, and ssp245) using the xclim python package. Baseline change calculation: Compute the change in each index between: recent past (2000-2014) and the preindustrial baseline (1850-1900) for both Historical and Hist-NAT; Future scenario (SSP2-4.5, 2070-2100) and the same preindustrial period. Extract the changes in extreme indices at WHSs to table. Plot: the global maps showing the difference between each scenario and the preindustrial baseline. Anthropogenic influence: Estimate the anthropogenic influence by subtracting Hist-nat from Historical for the period 2000–2014. Plot: Global anthropogenic influence (Historical contrasts Hist-NAT). Plot: Anthropogenic influence at WHSs only, by displaying WHS-containing grid cells. Regional WHS Exposure Summary: Compute regional exposure by averaging anthropogenic influence over grid cells that contain WHSs, grouped by IPCC-defined regions. Plot: Regional exposure map with hexagons sized by the number of WHSs in each IPCC region and colored by average anthropogenic influence at WHS grid cells. Socioeconomic impact stratification: Classify WHSs into impact groups (i.e., quartiles) based on the anthropogenic influence percentiles. Contrast these impact groups across World Bank-defined income categories and produce comparison plots. WHS time series plots: For 12 selected WHSs, plot smoothed time series (30-year rolling mean) for Tx90p, R95p, and CDD across Historical, Hist-nat, SSP2-4.5 scenarios. Result dataset All the calculated results related to this research are stored in the Excel file: CulturalWHS_Attribution.csv. This sheet contains the full list of 1,049 World Heritage Sites (WHSs) used in the analysis. All data are derived from integrating World Heritage Site attributes with climate model outputs under three scenarios (Hist-Nat, Hist-ALL, SSP2-4.5) for three extreme weather indices (Tx90p, R95p, CDD). Only WHSs with cultural or mixed value were retained; natural-only sites were excluded. Transboundary WHSs appear once per country. UNESCO WHS List: WHS ID, name, geographic coordinates, inscription date, and inscription criteria. World Bank data: country name, income group, GDP per capita. IPCC Atlas: assigned reference region. Research results: anthropogenic influence and baseline-relative change for three extreme indices (Tx90p, R95p, CDD). Please contact Shixin Zhao ([e-mail]; [e-mail]) for any queries.