Khoie et al. (2026) Characterising runoff processes for Australia: insights from a parsimonious rainfall-runoff event identification method

Identification

• Journal: Hydrology and earth system sciences
• Year: 2026
• Date: 2026-09-23
• Authors: Mohammad Masoud Mohammadpour Khoie, Danlu Guo, Conrad Wasko
• DOI: 10.5194/hess-30-5947-2026

Research Groups

• School of Engineering, ANU College of Systems and Society, Australian National University, Canberra, Australian Capital Territory, Australia
• Institute for Water Futures, The Australian National University, Canberra, Australian Capital Territory, Australia
• School of Civil Engineering, The University of Sydney, Sydney, New South Wales, Australia

Short Summary

This study introduces the Robust Variance-based Event Identification Method (RVEIM), a parsimonious and robust approach for identifying rainfall-runoff events. It demonstrates RVEIM's superior performance in reducing uncertainty compared to existing methods and provides the first comprehensive characterization of rainfall-runoff event characteristics across 467 Australian catchments, revealing a strong climate gradient in runoff coefficients.

Objective

• To develop and validate the Robust Variance-based Event Identification Method (RVEIM) for rainfall-runoff event identification, aiming to reduce uncertainty and improve transferability compared to existing methods.
• To apply RVEIM across 467 Australian catchments to characterize rainfall-runoff processes and understand their variability across diverse hydro-climatic conditions.

Study Configuration

• Spatial Scale: 467 unregulated Hydrologic Reference Stations (HRS) catchments across Australia, encompassing diverse hydro-climatic conditions. Catchment areas range from 95.6 square kilometres to 11,903.4 square kilometres. Eight representative catchments were selected for detailed analysis.
• Temporal Scale: Daily streamflow and rainfall data spanning at least 30 years, extending through February 2024.

Methodology and Data

• Models used:
  • Robust Variance-based Event Identification Method (RVEIM) – proposed method, relying on two parameters (moving variance window length d_var and baseflow filter parameter alpha).
  • Benchmarking methods: Local Maxima method (six parameters) and Detrending Moving-average Cross-correlation Analysis (DMCA) (two parameters).
  • Baseflow separation: Lyne and Hollick (1979) digital baseflow filter.
  • Missing streamflow data infilling: GR4J model.
  • Rainfall event detection: Peak Over Threshold (POT) approach (1 millimetre threshold, 1 day minimum difference).
• Data sources:
  • Streamflow: Daily records from the Australian Bureau of Meteorology’s Hydrologic Reference Stations (HRS).
  • Rainfall: Daily gridded data from the Australian Water Availability Project (AWAP) at 0.05° × 0.05° (approximately 5 kilometres × 5 kilometres) resolution, processed using the "AWAPer" R package.
  • Evapotranspiration and Soil Moisture: Bureau of Meteorology (AWAP).

Main Results

• RVEIM demonstrated significantly lower uncertainty in rainfall-runoff event characteristics (annual number, length, mean volume) with a standard deviation within ±15% of the mean across 8 representative catchments, compared to 43% to 93% for benchmarking methods.
• RVEIM consistently produced a lower percentage of physically implausible runoff coefficients (RCs > 1), generally below 4% for most catchments, indicating higher reliability than the local maxima method.
• The Empirical Cumulative Distribution Function (ECDF) of RCs derived from RVEIM and DMCA showed significantly narrower uncertainty bands across parameter choices compared to the local maxima method.
• A comprehensive summary of rainfall-runoff event characteristics across 467 Australian catchments revealed a strong climate gradient in event-scale runoff coefficients.
• In drier regions (desert, grassland), ECDFs of RCs showed a sharp increase at low RC values, suggesting a dominance of infiltration-excess runoff generation.
• In wetter regions (subtropical, temperate, equatorial), ECDFs rose more gradually towards higher RCs, indicating a higher potential for runoff generation and predominance of saturation-excess processes. Tropical catchments exhibited an intermediate pattern.
• Catchments in western Tasmania, despite longer rainfall events, exhibited shorter runoff events, attributed to steeper slopes promoting quick flow paths.
• Catchments in northern Western Australia showed the lowest mean rainfall volume but relatively larger mean runoff volume, suggesting lower losses due to lower evapotranspiration and higher soil moisture.

Contributions

• Introduction of the Robust Variance-based Event Identification Method (RVEIM), a novel, parsimonious (two parameters), robust, and transferable method for rainfall-runoff event identification.
• RVEIM simultaneously detects and pairs rainfall-runoff events using a time-variant search window, which better mimics natural hydrological processes and reduces the risk of mispairing.
• Demonstrated significantly lower uncertainty and higher reliability (fewer physically implausible RCs) of RVEIM compared to two widely used benchmarking methods (Local Maxima and DMCA).
• Provided the first comprehensive, Australia-wide summary of rainfall-runoff event characteristics and the distribution of event-scale runoff coefficients across 467 catchments.
• Revealed systematic shifts in runoff coefficient distributions across climate regions, indicating contrasts in dominant runoff generation mechanisms (e.g., infiltration-excess versus saturation-excess).

Funding

• National Computational Infrastructure (supported by the Australian Government)
• Sydney Horizon Fellowship (The University of Sydney, for Dr. Conrad Wasko)

Citation

@article{Khoie2026Characterising,
  author = {Khoie, Mohammad Masoud Mohammadpour and Guo, Danlu and Wasko, Conrad},
  title = {Characterising runoff processes for Australia: insights from a parsimonious rainfall-runoff event identification method},
  journal = {Hydrology and earth system sciences},
  year = {2026},
  doi = {10.5194/hess-30-5947-2026},
  url = {https://doi.org/10.5194/hess-30-5947-2026}
}