RealClearPolitics - Election 2024 - 2024 Nevada: Multi-Candidate

2024 Pennsylvania: Trump vs. Harris | RealClearPolling

RealClearPolitics - Election 2024 - 2024 New York: Trump vs. Harris

2024 Generic Congressional Vote | RealClearPolling

This repository contains wildfire risk maps generated by replicating the methodology described in the study “Climate Change Risk Indicators for Central Banking: Explainable AI in Fire Risk Estimations” by Burger et al. (2024):

Burger, Csaba and Herzberg, Julika and Nuvoli, Thaïs, Climate Change Risk Indicators for Central Banking: Explainable AI in Fire Risk Estimations (February 01, 2024). Available at SSRN: https://ssrn.com/abstract=4865384 or http://dx.doi.org/10.2139/ssrn.4865384

While we are not the authors of the original study, we provide geospatial outputs based on the same modeling pipeline and input data sources as described in the paper.

These maps were generated within the Alpha-Klima, in collaboration with OS-Climate. This is the first version of the maps; a GitHub repository with the full code to replicate the results will be published soon.

The dataset consists of three netCDF files covering Europe with a 2.5 x 2.5 km resolution. These maps represent estimated probabilities of wildfire occurrence and are intended for use in climate risk analysis, particularly in finance and regional planning contexts.

2. Dataset Descriptions

(a) Historical wildfire risk map (2001–2023)
[alpha_klima_historical_fire_probability_2010.nc]

This raster file contains annual average wildfire occurrence probabilities for the period 2001–2023. Values are derived using a constrained XGBoost model trained on fire flags, Fire Weather Index (FWI), land cover, and proximity to critical infrastructure, following the methodology outlined in Burger et al. The model enforces monotonicity with respect to key variables to ensure physical interpretability.

(b) Forecast wildfire risk map under RCP 4.5 (2024–2050)
[alpha_klima_rcp_4p5_fire_probability_2035.nc]

This file presents projected fire occurrence probabilities under the RCP 4.5 climate scenario. The forecasts assume fixed land cover and infrastructure as of 2022, with only climate variables (e.g. FWI) evolving according to scenario data from the Copernicus Climate Data Store. Probabilities are computed using a recursive Monte Carlo simulation incorporating lagged fire risk.

(c) Forecast wildfire risk map under RCP 8.5 (2024–2050)
[alpha_klima_rcp_8p5_fire_probability_2035.nc]

Similar to (b), this raster provides projected fire probabilities under the more extreme RCP 8.5 scenario. The map highlights regions expected to see the greatest increase in fire risk due to intensifying climatic conditions.