Kulkarni et al. (2025) Global assessment of socio-economic drought events at the subnational scale: a comparative analysis of combined versus single drought indicators

Identification

• Journal: Hydrology and earth system sciences
• Year: 2025
• Date: 2025-09-15
• Authors: Sneha Kulkarni, Yohei Sawada, Yared Bayissa, Brian Wardlow
• DOI: 10.5194/hess-29-4341-2025

Research Groups

• Department of Civil Engineering, School of Engineering, The University of Tokyo, Tokyo 113-8656, Japan
• Department of Ecology and Conservation Biology, Texas A&M University, College Station, TX 77843, USA
• School of Natural Resources, Center for Advanced Land Management Information Technologies, University of Nebraska-Lincoln, Lincoln, NE 68583, USA

Short Summary

This study globally assesses socio-economic drought events at the subnational scale by comparing a novel combined drought indicator (CDI) with single-parameter indices, finding that CDI significantly outperforms individual indices in identifying GDIS-documented socio-economic drought impacts.

Objective

• To understand the link between global drought hazards and their socio-economic impacts at the subnational scale using GDIS data.
• To develop a new global combined drought indicator (CDI) to enhance the precision and reliability of agro-climatological and socio-economic drought assessment, and to assess its performance in detecting GDIS drought events.
• To check the performance of commonly used traditional drought indicators (SPI, STI, NDVI, and SSMI) in identifying subnational socio-economic impacts of droughts (i.e., their association with GDIS).

Study Configuration

• Spatial Scale: Global, subnational level, with all datasets resampled to a consistent resolution of 0.10° × 0.10°.
• Temporal Scale: Monthly, covering the period from 2001 to 2021. Analysis primarily focused on drought events lasting 2 months or longer, but also included shorter durations and lag effects up to 3 months prior to the event.

Methodology and Data

• Models used:
  • Standardised Precipitation Index (SPI)
  • Normalised Difference Vegetation Index (NDVI)
  • Standardised Soil Moisture Index (SSMI)
  • Standardised Temperature Index (STI)
  • Combined Drought Indicator (CDI), developed using Principal Component Analysis (PCA) for weighted integration of input variables.
  • Z-score statistics for standardization of indices.
• Data sources:
  • Rainfall: Climate Hazards Group Infrared Precipitation with Station (CHIRPS) data (0.05° × 0.05°).
  • Temperature: ERA5-Land monthly temperature dataset (0.1° × 0.1°).
  • Soil moisture: ERA5-Land monthly volumetric soil moisture dataset (0.1° × 0.1°), using the top three layers (0–100 cm).
  • NDVI: Moderate Resolution Imaging Spectroradiometer (MODIS) (1 km × 1 km).
  • Socio-economic drought events: Geocoded Disaster (GDIS) dataset, based on EM-DAT, providing geocoded disaster locations at a subnational level.

Main Results

• Out of 1641 GDIS drought events (lasting ≥ 2 months), the Combined Drought Indicator (CDI) successfully identified 1550 (94.5 %) events when the drought threshold was set to ≤ -1, outperforming all individual indices.
• When considering a less stringent threshold of < 0, CDI identified 1635 (99.63 %) of the 1641 GDIS events.
• By extending the analysis window to include data from 3 months prior to the actual event period and using a threshold of < 0, CDI achieved 100 % detection of GDIS events (both for all 2142 events and for the 1641 events lasting ≥ 2 months).
• For the 1641 GDIS events (threshold ≤ -1, actual event period), individual indices detected: NDVI 1541 (93.9 %), SPI 1458 (88.8 %), STI 1439 (87.7 %), and SSMI 1376 (83.9 %).
• The highest frequency of severe droughts (greater than seven events) based on CDI was observed in sub-Saharan Africa and South Asia. CDI also captured persistent droughts in Argentina, Brazil, the Horn of Africa, western India, and North China.
• CDI consistently demonstrated superior detection accuracy across all Köppen climate zones (arid, tropical, temperate, and cold) and major land cover types (forest, shrubland and savanna, agriculture and cropland, settlement, and barren lands) compared to single-parameter indices.
• CDI exhibited strong spatial correlations with SSMI and NDVI, particularly in North America, southern parts of Africa, and the Indian subcontinent, with correlations showing seasonal variability.

Contributions

• This study provides the first global assessment of socio-economic drought events at the subnational scale, leveraging the Geocoded Disaster (GDIS) dataset to directly link drought hazards with their socio-economic repercussions.
• It introduces and validates a novel global Combined Drought Indicator (CDI) that integrates meteorological (rainfall, temperature) and agricultural (NDVI, soil moisture) variables using an objective Principal Component Analysis (PCA) based weighting approach.
• The research demonstrates the superior capability of the CDI over traditional single-parameter drought indices (SPI, STI, NDVI, SSMI) in accurately identifying global subnational socio-economic drought impacts.
• It offers a comprehensive global framework for drought impact assessment, integrating agro-climatological hazard data with socio-economic impacts, which can inform improved drought management strategies and policy-making worldwide.
• The study highlights the complex and regionally varying relationship between drought hazards and socio-economic vulnerability, noting disparities between developed and less developed nations.

Funding

• Japan Aerospace Exploration Agency (grant no. ER3AMF106)
• JSPS KAKENHI (grant nos. 21H01430 and 24K17352)
• Japan Society for the Promotion of Science (grant no. 21H01430)
• Katsu Kimura Research Award

Citation

@article{Kulkarni2025Global,
  author = {Kulkarni, Sneha and Sawada, Yohei and Bayissa, Yared and Wardlow, Brian},
  title = {Global assessment of socio-economic drought events at the subnational scale: a comparative analysis of combined versus single drought indicators},
  journal = {Hydrology and earth system sciences},
  year = {2025},
  doi = {10.5194/hess-29-4341-2025},
  url = {https://doi.org/10.5194/hess-29-4341-2025}
}