Chakma et al. (2025) Copula-based multivariate analysis of hydrological drought over jiabharali sub-basin of Brahmaputra River, India

Identification

• Journal: Scientific Reports
• Year: 2025
• Date: 2025-12-04
• Authors: Bivek Chakma, Deepak Jhajharia, Ghanashyam Singh Yurembam, G. T. Patle, Phuritshabam Robert, Saurav Saha, Aribam Priya Mahanta Sharma, Sanjay Kumar Singh
• DOI: 10.1038/s41598-025-28558-6

Research Groups

• College of Agricultural and Engineering and Post Harvest Technology, Ranipool, Central Agricultural University (Imphal), India
• College of Agriculture, Imphal, Central Agricultural University (Imphal), India
• ICAR Research Complex for NEH Region, Sikkim Centre, Tadong, Gangtok, Sikkim, India
• School of Agriculture, ITM University, Gwalior, India
• Central Water Commission, Department of Water Resources, Ministry of Jal Shakti (Government of India), Delhi, India

Short Summary

This study conducted a copula-based multivariate analysis of hydrological drought in the Jiabharali sub-basin of the Brahmaputra River, India, from 2000 to 2023, revealing that drought severity and duration are strongly correlated and their joint return periods increase with longer time scales, indicating extreme joint drought events.

Objective

• To investigate the occurrence of hydrological drought using the multi-scale Streamflow Drought Index (SDIn) in the Jiabharali (Kameng) sub-basin of the Brahmaputra River.
• To construct a multivariate copula-based hydrological drought model for joint probability distribution, including conditional probability and return periods, between suitable drought variables at multi-time scales (3, 6, 9, and 12-months).

Study Configuration

• Spatial Scale: Jiabharali (Kameng) River, a sub-basin of the Brahmaputra River, flowing between Arunachal Pradesh and Assam, Northeast India. Data collected from Bhalukpong station.
• Temporal Scale: 2000 to 2023 (24 years) for daily streamflow data, analyzed at multi-time scales (3, 6, 9, and 12-months) for drought indices.

Methodology and Data

• Models used:
  • Streamflow Drought Index (SDIn) at multi-time scales (SDIn3, SDIn6, SDIn9, SDIn12).
  • Marginal Probability Distribution Functions (PDFs): Normal, Log-normal, Gamma, Weibull, Exponential distributions.
  • Copula families: Elliptical (Normal, t-copula) and Archimedean (Gumbel, Clayton, Frank, Joe).
  • Goodness-of-fit tests: Kolmogorov-Smirnov (K-S) Test, Anderson-Darling (AD) Test, Root Mean Square Error (RMSE), Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC).
  • Correlation measures: Spearman’s rank (ρ) and Kendall’s Tau (τ).
• Data sources: Daily streamflow (discharge) data from Bhalukpong station of Jiabharali river (2000-2023), provided by the Central Water Commission (CWC), Brahmaputra Division (BD), Assam, India.

Main Results

• Hydrological drought occurrences became more severe and lengthy with increasing SDIn time scales; the highest drought severity of -87.75 (average -21.3) lasted for 71 months in the SDIn12 time scale.
• Significant negative streamflow departures were observed from 2013 to 2018/2019, with monthly percentage departures reaching -70% in August 2016-17.
• The log-normal distribution was the best-fit for drought Severity across all SDIn time scales. For Duration, log-normal (SDIn3, SDIn6, SDIn9) and exponential (SDIn12) were best-fit. For Interval-time, Normal (SDIn3, SDIn6), Gamma (SDIn9), and Weibull (SDIn12) were best-fit.
• A strong positive correlation was found between drought Severity (S) and Duration (D) across all time scales (p=0.00). Correlations involving Interval (I) were generally weak or poor.
• The Normal copula was the best-fit for S-D pairs in SDIn3 to SDIn9, indicating symmetric dependence, while the Gumbel copula was best-fit for SDIn12, suggesting extreme upper tail dependence for longer-term droughts.
• Multivariate joint return periods (TAND SD and TOR SD) for S and D increased with increasing time scale and return period. For a 2-year univariate threshold, TAND SD was 17 years and TOR SD was 11 years for SDIn3. For 50 years at SDIn12, TAND SD was 1467 years and TOR SD was 1136 years.
• Conditional probability P(D>d|S>s) showed a strong likelihood at longer time scales (SDIn12), indicating that longer return periods (20, 50, 100 years) are associated with more severe droughts of shorter duration compared to shorter return periods.

Contributions

• Uniquely fills research gaps in hydrological drought investigation using copula theory in the Jiabharali sub-basin of the Brahmaputra River, Northeast India.
• Provides valuable insights into the correlation between different drought variables and models joint drought return period behaviors over multiple time scales.
• Assists policymakers and water planners in making timely, informed decisions for drought risk management and mitigation efforts in the region.
• Emphasizes the importance of incorporating joint probability assessments in drought risk planning, especially for extreme joint drought events.

Funding

• National fellowship for Higher Education of ST students (NFST), Ministry of Tribal Affairs, Government of India, New Delhi (grant to the first author for five years since October 28, 2022).

Citation

@article{Chakma2025Copulabased,
  author = {Chakma, Bivek and Jhajharia, Deepak and Yurembam, Ghanashyam Singh and Patle, G. T. and Robert, Phuritshabam and Saha, Saurav and Sharma, Aribam Priya Mahanta and Singh, Sanjay Kumar},
  title = {Copula-based multivariate analysis of hydrological drought over jiabharali sub-basin of Brahmaputra River, India},
  journal = {Scientific Reports},
  year = {2025},
  doi = {10.1038/s41598-025-28558-6},
  url = {https://doi.org/10.1038/s41598-025-28558-6}
}