GLEXIS

GLEXIS is the Global Extreme event Indicators from ISIMIP Simulations dataset and code archive. It provides a global, multi-model set of monthly climate-extreme indicators derived from daily bias-adjusted ISIMIP3b atmospheric forcing for 1981-2100.

The dataset is designed for comparative exposure, impact-model, and adaptation-analysis workflows that need spatially explicit, multi-hazard, multi-model climate-extreme information on a common historical baseline.

Scope

GLEXIS covers:

• Historical simulations for 1981-2014.
• Future projections for 2015-2100 under SSP1-2.6, SSP3-7.0, and SSP5-8.5.
• Five CMIP6 models: GFDL-ESM4, IPSL-CM6A-LR, MPI-ESM1-2-HR, MRI-ESM2-0, and UKESM1-0-LL.
• Valid land grid cells on the native 0.5 degree ISIMIP3b grid north of 60 degrees S.
• Monthly raster products and country-level tabular summaries for selected products.

Thresholds are calibrated to a fixed 1990-2010 historical baseline and then applied consistently to historical and future periods. This fixed-baseline design makes changes in event frequency interpretable relative to the same reference climatology.

Indicators

GLEXIS includes five headline meteorological hazard indicators:

• Hot days: daily maximum near-surface air temperature above the local historical p95 threshold. This is a daily frequency indicator, not a heatwave spell metric.
• Cold-extreme days: daily minimum near-surface air temperature below the local historical p01 threshold. This is not a fixed-threshold frost-day index.
• Heavy-rain days: daily precipitation above the local historical p95 threshold and at least 1 mm d-1. This is a meteorological precipitation indicator, not a flood indicator.
• High-wind days: daily mean near-surface wind speed above the local historical p95 threshold and at least 8 m s-1. This uses daily mean wind speed, not observed gusts.
• Standardized water-balance drought indicator: annual precipitation-minus-PET water balance standardized against the 1990-2010 baseline and classified below the local p05 threshold. This is an annual water-balance deficit indicator, not canonical SPEI-3.

The drought indicator is annual in resolution. In the monthly raster archive, a drought-classified year is represented uniformly across all 12 months for archive consistency.

Products

The full data archive is deposited separately from this source-code repository because the raster products are too large for GitHub.

The manuscript data record describes:

• 4,080 historical GeoTIFF files: 5 models x 34 years x 24 files.
• 30,960 future GeoTIFF files: 5 models x 3 SSPs x 86 years x 24 files.
• Twelve monthly bands per GeoTIFF, January through December.
• WGS84 spatial reference (EPSG:4326), 0.5 degree pixel size, and 720 x 300 raster dimensions.
• Event-count layers and conditional minimum, mean, and maximum magnitude layers.
• Historical country-level annual and monthly ensemble CSV products for 178 countries.
• Global area-weighted annual summaries across historical and future periods.

Country-level CSV column definitions are documented in release_metadata/csv_data_dictionary.csv.

Repository Contents

• R/production/: production scripts for historical and future GLEXIS processing.
• R/thresholds/: archived threshold-generation scripts.
• release_metadata/: implementation definitions, data dictionaries, file manifests, correction logs, validation summaries, and software provenance.
• data/README.md: notes on the separately deposited data archive.
• CITATION.cff: software citation metadata for GitHub and Zenodo-style releases.
• LICENSE: MIT license for the code in this repository.

Reproducibility

The authoritative implementation details are recorded in:

• release_metadata/implementation_definition.csv
• release_metadata/hazard_variant_definition.csv
• release_metadata/targeted_corrections.csv
• release_metadata/calendar_harmonization_audit.csv
• release_metadata/software_provenance.csv

Key reproducibility conventions from the manuscript:

• ISIMIP3b daily forcing is bias-adjusted against W5E5.
• Thresholds use a fixed 1990-2010 baseline.
• Gregorian leap years are harmonized by removing the nominal 29 February slot for consistent day-of-year indexing.
• UKESM1-0-LL keeps its native 360-day calendar.
• GeoTIFF count-layer NoData is encoded as -9999; conditional magnitude NoData is encoded as NaN.
• Six targeted single-cell/month corrections are documented in the release metadata.

Usage Notes

GLEXIS is suitable for:

• Monthly multi-hazard forcing for impact models.
• Multi-model and multi-scenario comparison of climate-hazard frequency trajectories.
• Cross-hazard co-occurrence and compound-extreme screening.
• Historical trend characterization relative to the fixed 1990-2010 baseline.

Important limitations:

• GLEXIS indicators are meteorological hazard screening metrics, not impact or risk estimates.
• Free-running GCM simulations should not be compared year-by-year with specific observed disaster events.
• The high-wind indicator uses daily mean wind speed; wind-damage studies should treat it carefully.
• Wind and drought validation confidence is lower than for temperature and precipitation indicators.
• The ensemble mean may not represent any individual model well in regions with high inter-model spread.

Data Availability

The formal GLEXIS data record is available at DOI: 10.5281/zenodo.20734236.

The main GLEXIS data-link browser is available at:

https://andrenakhavali.github.io/GLEXIS_data_links_public/

The raster GeoTIFF files, tabular CSV products, threshold NetCDF files, and full release metadata are deposited separately from this code repository.

See data/README.md and release_metadata/release_file_manifest.csv for the expected data-package structure.

Code Availability

The production and threshold-generation scripts are released in this repository under the MIT License. The main production scripts are:

• R/production/GLEXIS_historical.R
• R/production/GLEXIS_future_with_wind_gust.R
• R/production/extreme_thresholds_wind_with_gust.R

Citation

Please cite the GLEXIS data descriptor manuscript, the archived software release, and the formal data record: DOI: 10.5281/zenodo.20734236. Repository-level citation metadata are provided in CITATION.cff.

License

Code is released under the MIT License. Data licensing is stated separately in the formal data repository record.