Rahman et al. (2026) Characterization of drought in the Arabian Peninsula: A multi-index approach

Identification

• Journal: Journal of Hydrology Regional Studies
• Year: 2026
• Date: 2026-02-20
• Authors: Ghani Rahman, Abdur Rashid Jamalzi, Vivek Anand Voora, Sina Borzooi, Subin Kang, Hyun-Han Kwon
• DOI: 10.1016/j.ejrh.2026.103250

Research Groups

• Department of Civil and Environmental Engineering, Sejong University, Seoul, Republic of Korea
• Department of Geography, University of Gujrat, Punjab, Pakistan
• IVL Swedish Environmental Research Institute, Stockholm, Sweden

Short Summary

This study characterizes drought in the Arabian Peninsula (AP) from 1975 to 2024 using a multi-index approach (SPI, SPEI, EDDI) and ERA5 reanalysis data, revealing a marked increase in drought frequency and severity, particularly in the Southeast and Southwest zones, primarily driven by dewpoint temperature, precipitation, and maximum temperature.

Objective

• To assess long-term drought dynamics in the Arabian Peninsula using a multi-index framework (SPI, SPEI, EDDI) based on ERA5 reanalysis data, evaluating spatiotemporal patterns, identifying asymmetric distributional shifts, characterizing the joint probability of duration and severity, and determining dominant meteorological drivers.

Study Configuration

• Spatial Scale: The Arabian Peninsula (over 3.2 million square kilometers), divided into four homogeneous zones: Northeast (NE), Northwest (NW), Southeast (SE), and Southwest (SW).
• Temporal Scale: 1975–2024 (50 years).

Methodology and Data

• Models used:
  • Drought Indices: Standardized Precipitation Index (SPI), Standardized Precipitation-Evapotranspiration Index (SPEI), Evaporative Demand Drought Index (EDDI), computed at 3-, 6-, and 12-month accumulation windows.
  • Trend Analysis: Modified Mann–Kendall (MMK) test, Sen’s slope estimator, Innovative Trend Analysis (ITA) with Improved Visualization for ITA (IV-ITA), Pettitt test.
  • Drought Characterization: Run theory.
  • Dependence Modeling: Copula functions (Gaussian, Clayton, Frank, Gumbel).
  • Driver Attribution: SHAP (Shapley Additive exPlanations)-based Extreme Gradient Boosting (XGBoost) regression model.
  • Spatial Homogenization: Rotated Empirical Orthogonal Function (REOF).
  • Correlation: Pearson correlation coefficient.
• Data sources: ERA5 reanalysis dataset from the European Centre for Medium-Range Weather Forecasts (ECMWF) at a spatial resolution of 0.25 degrees. Key meteorological variables include precipitation, 2-meter air temperature (maximum and minimum), dewpoint temperature, surface pressure, solar radiation, and wind speed, aggregated to a monthly timescale.

Main Results

• Major drought events occurred in 1983–1984, 1999–2002, 2007–2009, 2014–2015, and 2021–2023, with the Southeast (SE) and Southwest (SW) zones identified as most drought-prone.
• SPEI and EDDI consistently indicated greater drought intensity, earlier onset, and persistence across all timescales compared to SPI, underscoring the increasing influence of rising temperature and evaporative demand.
• The Modified Mann–Kendall test and Sen’s slope revealed significant drying trends, particularly at 12-month timescales in the SW and SE zones.
• Innovative Trend Analysis (IV-ITA) showed systematic and widespread redistribution of hydroclimatic conditions, with statistically significant negative shifts in both dry and wet extremes for SPEI and EDDI, and a notable decrease in wet spells for SPI. Shifts in high-value distributions consistently exceeded those in low-value distributions (e.g., Δhigh/Δlow ratios from 6.3:1 to 2.6:1).
• SHAP-based XGBoost analysis identified dewpoint temperature (mean absolute SHAP ≈ +0.14), precipitation (mean absolute SHAP ≈ +0.11), and maximum temperature (mean absolute SHAP ≈ +0.08) as the dominant meteorological drivers of drought severity (SPEI-12 and EDDI-12).
• Joint probability assessment using Copula functions indicated that multi-year droughts (up to approximately 25 months duration, with severities around 30 index-months) have return periods of about 20 years, highlighting their significant risk.
• The SE zone exhibited the highest frequency of extreme droughts (EDDI up to 3.7%, SPEI up to 3.5%), while SPI showed no extreme droughts in this zone.
• Pearson correlation analysis showed strong positive correlations between SPEI-12 and EDDI-12 (r ≈ 0.98) across all zones, reflecting the dominant role of evaporative demand in long-term drought variability.

Contributions

• This study provides the first comprehensive, multi-index drought assessment for the entire Arabian Peninsula, integrating the Evaporative Demand Drought Index (EDDI) within a combined monitoring framework.
• It applies Innovative Trend Analysis (ITA) with improved visualization to identify asymmetric distributional shifts and complex non-linear drought dynamics, complementing traditional trend detection methods.
• The research characterizes the joint probability of drought duration and severity using Copula functions, offering a more robust approach to risk management.
• It quantifies the relative contributions of multiple meteorological drivers to drought variability using an interpretable SHAP-based XGBoost model, providing novel insights into the underlying physical mechanisms.
• The findings emphasize the critical need for multi-index drought monitoring and region-specific adaptation strategies to enhance climate resilience in arid environments.

Funding

• Cooperative Research Method and Safety Management Technology in National Disaster (grant 2022-MOIS63–001(RS-2022-ND641011)) funded by the Ministry of the Interior and Safety (MOIS, Korea).

Citation

@article{Rahman2026Characterization,
  author = {Rahman, Ghani and Jamalzi, Abdur Rashid and Voora, Vivek Anand and Borzooi, Sina and Kang, Subin and Kwon, Hyun-Han},
  title = {Characterization of drought in the Arabian Peninsula: A multi-index approach},
  journal = {Journal of Hydrology Regional Studies},
  year = {2026},
  doi = {10.1016/j.ejrh.2026.103250},
  url = {https://doi.org/10.1016/j.ejrh.2026.103250}
}