Hajani (2025) Uncertainty in stationary and nonstationary IFD curves with future projections in Australia

Identification

• Journal: The Science of The Total Environment
• Year: 2025
• Date: 2025-12-03
• Authors: Evan Hajani
• DOI: 10.1016/j.scitotenv.2025.181127

Research Groups

Civil Engineering Department, College of Engineering, University of Duhok

Short Summary

This study updates Intensity-Frequency-Duration (IFD) curves for six Australian stations using 45 years of annual maximum rainfall data, comparing stationary and non-stationary models with ENSO to quantify uncertainty in future rainfall extreme projections.

Objective

• To update Intensity-Frequency-Duration (IFD) curves for six Australian stations, comparing stationary and non-stationary approaches (including the El Niño–Southern Oscillation (ENSO) index as a covariate) to quantify uncertainty in rainfall extreme projections for hydrological design and infrastructure planning.

Study Configuration

• Spatial Scale: Six Australian stations.
• Temporal Scale: 45 years (1980–2024) of annual maximum rainfall (AMR) data; analysis of short-duration rainfall intensities (e.g., 1-hour) derived from 24-hour AMR values; return periods up to 100 years.

Methodology and Data

• Models used: Generalized Extreme Value (GEV) distribution (identified as optimal), Gumbel distribution, Log-Pearson III distribution; stationary and non-stationary models (incorporating temporal trends and the ENSO index); Monte Carlo simulation and bootstrap resampling for uncertainty quantification.
• Data sources: 45 years (1980–2024) of annual maximum rainfall (AMR) data from six Australian stations.

Main Results

• The Generalized Extreme Value (GEV) distribution was identified as the optimal model across all study sites.
• Non-stationary models incorporating ENSO predicted 1-hour rainfall intensities up to 18 % higher than stationary models for 100-year return periods.
• Uncertainty quantification revealed widening confidence intervals with increasing return periods, particularly for short durations. For instance, the 1-hour 100-year estimate at Carnegie exhibited a ±38 % relative uncertainty under the non-stationary model versus ±22 % for the stationary case.
• Including ENSO as a covariate allows the model to reflect interannual climatic variability but increases uncertainty for long return periods, especially where observations of extremes are sparse.

Contributions

• Provides updated Intensity-Frequency-Duration (IFD) curves for Australian stations, explicitly incorporating non-stationarity due to temporal trends and the influence of the El Niño–Southern Oscillation (ENSO).
• Quantifies the uncertainty associated with both stationary and non-stationary IFD curves, highlighting the increased uncertainty for long return periods when including additional climate covariates like ENSO.
• Demonstrates the impact of large-scale climate drivers (ENSO) on rainfall extremes and its significance for flood risk and infrastructure planning in Australia.
• Contributes to the understanding of the heterogeneous nature of rainfall extremes by examining spatially diverse Australian stations exhibiting both increasing and decreasing trends in annual maximum rainfall.

Funding

Not specified in the provided text.

Citation

@article{Hajani2025Uncertainty,
  author = {Hajani, Evan},
  title = {Uncertainty in stationary and nonstationary IFD curves with future projections in Australia},
  journal = {The Science of The Total Environment},
  year = {2025},
  doi = {10.1016/j.scitotenv.2025.181127},
  url = {https://doi.org/10.1016/j.scitotenv.2025.181127}
}