Anvari et al. (2025) A nonstationary framework for hydrological drought assessment in Iran

Identification

• Journal: Journal of Hydrology Regional Studies
• Year: 2025
• Date: 2025-12-13
• Authors: Sedigheh Anvari, Jesper Rydén, Ameneh Mianabadi
• DOI: 10.1016/j.ejrh.2025.103009

Research Groups

• Department of Ecology, Institute of Science and High Technology and Environmental Science, Graduate University of Advanced Technology, Kerman, Iran
• Unit of Applied Statistics and Mathematics, Department of Energy and Technology, Swedish University of Agricultural Sciences, Uppsala, Sweden

Short Summary

This study introduces a Nonstationary Standardized Runoff Index (NSRI) using the GAMLSS framework to enhance hydrological drought assessment in Iran's Halil-Rud Basin. It finds that nonstationary models, incorporating hydroclimatic covariates, consistently outperform stationary models by more accurately capturing spatiotemporal drought variability and moderating extreme drought estimates, especially in human-impacted downstream regions.

Objective

• To develop a Non-Stationary Standardized Runoff Index (NSRI) tailored for hydrological drought assessment in Iran's changing climate.
• To model the non-stationarity of climate-related variables using both temporal and hydro-climatic covariates within the Generalized Additive Models for Location, Scale, and Shape (GAMLSS) framework.
• To evaluate the proposed NSRI against its stationary counterpart (SSRI) in terms of temporal and spatial drought characteristics, including severity, duration, and intensity.

Study Configuration

• Spatial Scale: Halil-Rud Basin, Kerman Province, Iran (approximately 22,255 km², with a focus on 7224 km² up to the Jirof Dam). Data from three hydrometric stations (Pole-Baft, Meidan, Kenaroyeh) and three synoptic stations (Baft, Jiroft, Kahnooj).
• Temporal Scale: 40-year period from 1980 to 2019, using monthly and 12-month cumulative runoff data.

Methodology and Data

• Models used:
  • Nonstationary Standardized Runoff Index (NSRI)
  • Stationary Standardized Runoff Index (SSRI)
  • Generalized Additive Models for Location, Scale, and Shape (GAMLSS) framework with time-varying Gamma distribution.
  • Mann-Kendall (MK) trend test
  • Augmented Dickey-Fuller (ADF) test
  • Pearson correlation analysis for covariate selection
  • Akaike Information Criterion (AIC) for model comparison and selection
  • Run-length theory for drought characteristic delineation
• Data sources:
  • Monthly precipitation and air temperature records from three synoptic stations (Baft, Jiroft, Kahnooj) provided by the Iran Meteorological Organization (IMO).
  • Historical monthly runoff data from three hydrometric stations (Pole-Baft, Meidan, Kenaroyeh) managed by the Iran Ministry of Energy.
  • Gridded climate data (precipitation, mean, maximum, and minimum temperature) from the Climatic Research Unit (CRU) database (0.5-degree spatial resolution).
  • Hydroclimatic covariates: precipitation (P), temperature (T), potential evapotranspiration (PET), and antecedent runoff (R).

Main Results

• All three hydrometric stations exhibited a statistically significant decreasing trend in runoff (p-values < 0.05), with an average Sen’s slope of approximately -0.011.
• The Augmented Dickey-Fuller (ADF) test confirmed non-stationary behavior in the cumulative runoff series at all stations.
• Runoff showed positive correlations with precipitation (0.20–0.37) and strong positive correlations with antecedent runoff (>0.94), while exhibiting negative correlations with temperature and potential evapotranspiration (–0.31 to –0.45).
• Nonstationary models consistently outperformed the stationary baseline (M0) across all stations, yielding substantially lower Akaike Information Criterion (AIC) values.
• Model M25 (incorporating temperature and antecedent runoff) provided the best fit for Pole-Baft station for most months. Model M26 (incorporating potential evapotranspiration and antecedent runoff) was optimal for Meidan station for most months. For Kenaroyeh, M25 was optimal in early/late hydrological year, and M26 dominated summer months.
• The Nonstationary Standardized Runoff Index (NSRI) detected more drought events (29 total) compared to the Stationary Standardized Runoff Index (SSRI) (26 total) across the study period.
• NSRI indicated a higher average peak drought intensity (-2.49) compared to SSRI (-2.40).
• At the downstream Kenaroyeh station, NSRI revealed a significantly longer maximum drought duration (130 months vs. 100 months for SSRI) and a higher maximum severity (-101.13 vs. -64.78), highlighting increased non-stationarity and human impacts.
• Drought frequency analysis showed NSRI indicating substantially higher frequencies of moderate and severe droughts at Meidan and Kenaroyeh, reflecting the influence of non-stationarity and anthropogenic factors.

Contributions

• Introduces a novel Non-Stationary Standardized Runoff Index (NSRI) framework for hydrological drought assessment in semi-arid regions, specifically tailored to Iran's changing climate.
• Demonstrates the superior performance and robustness of nonstationary modeling (using GAMLSS with time-varying Gamma distribution parameters and hydroclimatic covariates) over traditional stationary approaches.
• Quantifies the risk of overestimating extreme drought severity and underrepresenting drought frequency when stationary assumptions are applied, particularly in downstream regions influenced by anthropogenic activities.
• Provides a robust, data-driven framework that can inform policymakers for enhanced drought monitoring, risk assessment, and adaptive water resource management strategies.
• Methodologically contributes to statistical modeling by exploring non-stationary parameterization within the GAMLSS framework, supported by rigorous model comparison using AIC and diagnostic tools like worm plots.

Funding

• Graduate University of Advanced Technology (Institute of Science and High Technology and Environmental Science) (Grant No. 1402/3641).

Citation

@article{Anvari2025nonstationary,
  author = {Anvari, Sedigheh and Rydén, Jesper and Mianabadi, Ameneh},
  title = {A nonstationary framework for hydrological drought assessment in Iran},
  journal = {Journal of Hydrology Regional Studies},
  year = {2025},
  doi = {10.1016/j.ejrh.2025.103009},
  url = {https://doi.org/10.1016/j.ejrh.2025.103009}
}