Bhattacharjee et al. (2026) Spatiotemporal reorganization of drought characteristics across India under changing monsoon variability

Identification

• Journal: Natural Hazards
• Year: 2026
• Date: 2026-03-13
• Authors: Debankana Bhattacharjee, V. Rajendran, C. T. Dhanya
• DOI: 10.1007/s11069-026-08062-4

Research Groups

• Department of Civil and Environmental Engineering, Indian Institute of Technology Delhi, New Delhi, India
• Department of Civil Engineering, Indian Institute of Technology Roorkee, Roorkee, India

Short Summary

This study assesses the spatiotemporal reorganization of drought characteristics across India (1902-2013) using a non-stationary drought index (NSPI) and non-linear trend analysis (EEMD). It reveals a significant post-1950s increase in intrinsic drought duration (~61%) and severity (~62%), driven by changing monsoon variability, with NSPI demonstrating superior detection skill compared to traditional stationary indices.

Objective

• To assess the long-term (1902–2013) spatiotemporal evolution of drought characteristics across India using a non-stationary drought index (NSPI) and non-linear trend analysis (EEMD).
• To demonstrate the superiority of non-stationary frameworks over conventional stationary indices for accurate drought detection and characterization under changing monsoon variability.

Study Configuration

• Spatial Scale: Entire Indian subcontinent (6.5 °N to 38.5 °N and 66.5 °E to 100 °E), covering 3,287,263 square kilometers. Gridded precipitation data at 0.25° x 0.25° resolution. Five representative locations from different climate zones were selected for detailed analysis.
• Temporal Scale: 1902–2013 (112 years) for drought analysis, with a focus on 9-month timescales for drought characteristics. Validation periods include 1951–2013 for national crop yields, 2005–2013 for state-level Rabi yields, and 1981–2013 for ERA5 soil moisture.

Methodology and Data

• Models used:
  • Non-stationary Standardized Precipitation Index (NSPI): Utilizes a time sliding window (TSW) approach to detect non-stationarity in statistical parameters (Gamma distribution), followed by time-varying parameter estimation and marginal distribution computation.
  • Standardized Precipitation Index (SPI): Conventional stationary drought index used for comparative assessment.
  • Ensemble Empirical Mode Decomposition (EEMD): A noise-assisted, non-linear trend analysis method that decomposes signals into Intrinsic Mode Functions (IMFs) to extract non-linear trends (residuals). Uncertainty and statistical significance evaluated using Monte Carlo simulations.
  • Statistical tests: Moving block bootstrap Mann-Kendall test (for non-stationarity), Kolmogorov-Smirnov (KS) test (for Gamma distribution fit), Bayesian inference (for parameter estimates and uncertainties), Welch’s two-sample t-test (for crop yield significance), Bootstrap approach (for yield loss robustness).
• Data sources:
  • Daily gridded precipitation (0.25° x 0.25°) data from January 1902 to December 2013, provided by the Indian Meteorological Department (IMD).
  • National crop yields (1951–2013) from the Ministry of Agriculture and Farmers Welfare, Government of India.
  • State-level Rabi food grain yields (2005–2013) from the World Food and Agricultural Statistical Yearbook 2024.
  • ERA5 soil moisture data (1981–2013).
  • Population density data from the Census of India (2011).

Main Results

• NSPI demonstrated superior drought detection skill compared to SPI, achieving significantly higher Pearson correlations with internal drought signals (e.g., 0.314 vs. 0.189 in Tropical Wet and Dry zone) and robust event capture (F1-scores > 0.70 in Semi-Arid zones) where stationary metrics failed.
• Non-linear trend analysis revealed a statistically significant post-1950s rise of approximately 61% in intrinsic drought duration and approximately 62% in severity. Drought depth also intensified by ~15% by the 2000s after an earlier moderation.
• Spatially, northwestern, western, and eastern India experienced shorter but more intense droughts, while central and southern regions faced longer, milder episodes.
• Intensifying extremes were observed in Semi-Arid and Humid Subtropical zones, and rising frequency in Tropical Wet and Tropical Wet and Dry zones.
• Approximately 70% of India’s population is exposed to intensifying drought hazards. Significant national cereal yield losses were quantified during drought years (e.g., coarse cereals ~14.13%, wheat ~10.15%, rice ~9.45%, pulses ~4.94%). Arid (7.86% loss) and Semi-Arid (7.33% loss) zones showed the highest mean yield losses.
• Hydrological validation with ERA5 soil moisture confirmed synchronous declines during moderate to severe drought periods across all climate zones.

Contributions

• Presents the first long-term (1902–2013) India-wide assessment of evolving drought characteristics using a non-stationary drought index (NSPI) in conjunction with non-linear trend analysis (EEMD).
• Demonstrates the critical importance and methodological superiority of non-stationary frameworks (NSPI) over conventional stationary indices (SPI) for accurate drought detection and characterization under changing monsoon variability.
• Quantifies statistically significant non-linear trends in drought duration (~61% increase) and severity (~62% increase) post-1950s, revealing a structural reorganization of monsoon dynamics.
• Identifies divergent regional drought trajectories and vulnerable hotspots across India's diverse climate zones, providing nuanced insights for adaptive planning.
• Establishes a direct, measurable link between intensifying non-stationary drought signals and significant national agricultural yield losses, highlighting socioeconomic risks and the need for non-stationary drought mitigation frameworks.

Funding

• Prime Minister’s Research Fellowship (PMRF), Government of India.
• Science and Engineering Research Board (SERB), Department of Science and Technology, Government of India, through the SERB Women Excellence Award Scheme (Grant No.: SB/WEA-04/2017).

Citation

@article{Bhattacharjee2026Spatiotemporal,
  author = {Bhattacharjee, Debankana and Rajendran, V. and Dhanya, C. T.},
  title = {Spatiotemporal reorganization of drought characteristics across India under changing monsoon variability},
  journal = {Natural Hazards},
  year = {2026},
  doi = {10.1007/s11069-026-08062-4},
  url = {https://doi.org/10.1007/s11069-026-08062-4}
}