Niedzielski et al. (2025) Relationship between skills of multimodel hydrologic ensemble predictions and atmospheric circulation patterns: a case study from the Nysa Kłodzka river basin (SW Poland)

Identification

• Journal: Acta Geophysica
• Year: 2025
• Date: 2025-12-11
• Authors: Tomasz Niedzielski, Hanna Ojrzyńska, Bartłomiej Miziński, Maciej Kryza, Waldemar Spallek
• DOI: 10.1007/s11600-025-01733-z

Research Groups

• Institute of Geography and Regional Development, Faculty of Earth Sciences and Environmental Management, University of Wrocław, Poland

Short Summary

This study investigated the relationship between short-term hydrologic ensemble prediction skills and atmospheric circulation patterns in the Nysa Kłodzka river basin, finding that the most skillful predictions were associated with wet humidity and prevailing northerly air mass advection.

Objective

• To verify the hypothesis that specific atmospheric circulation patterns exist for which short-term hydrologic ensemble predictions (with lead times from minutes to hours) are skillful.

Study Configuration

• Spatial Scale: Upper Nysa Kłodzka river basin (southwestern Poland), with a contributing basin area of 1744 km². Forecasts were computed at 11 hydrologic gauges.
• Temporal Scale: The experiment was conducted from 1 September 2013 to 3 December 2016. Short-term predictions had lead times ranging from 15 minutes to 180 minutes (3 hours), with 12 intermediate 15-minute steps, and were updated every 15 minutes.

Methodology and Data

• Models used: HydroProg hydrologic ensemble prediction system, integrating up to six ensemble members:
  • Vector Autoregressive model (VAR)
  • Univariate Autoregressive model (AR)
  • Autocovariance method (ACV)
  • Topography-based Hydrological Model (TOPMODEL)
  • Artificial Neural Network Autoregressive model (ANNAR)
  • Artificial Neural Network Autoregressive model with regularisation (ANNARreg)
  • Multimodel ensemble (weighted sum of the above).
• Data sources:
  • Hydrometeorological data from 20 automatic hydrologic gauging stations and 12 automatic weather stations of the Local System for Flood Monitoring of Kłodzko county (LSOP), with a 15-minute sampling interval.
  • Atmospheric circulation types derived from "Die objektive Wetterlagenklassifikation" (modified by Ojrzyńska 2015), based on:
    • Global NCEP/NCAR Reanalysis data (2.5° × 2.5° horizontal resolution, 4 times per day).
    • ERA5 reanalysis data (0.25° × 0.25° horizontal resolution) for comparison.
  • Prediction skill was quantified using the Nash-Sutcliffe efficiency (NSE).
  • Statistical analysis included the two-sample proportion test and a bootstrap approach for chi-squared statistics.

Main Results

• The most skillful short-term water level predictions (classified as good or satisfactory, NSE ≥ 0.36) were associated with the wet humidity type and prevailing northerly advection of air masses.
• One circulation type (NEACW) exhibited good prediction skills (median NSE = 0.82), while 12 circulation types showed satisfactory performance (median NSE between 0.36 and 0.75).
• Among the 13 best-performing circulation types, the wet humidity type was most frequent (77% of satisfactory cases), and northerly advection prevailed, particularly for the highest-ranked predictions.
• Heavy rainfall events (exceeding 6 mm/h) occurred for approximately 69% of the good/satisfactory circulation types, compared to 46% for unsatisfactory types.
• The identified relationship between prediction skills and atmospheric circulation patterns was qualitative rather than quantitative, suggesting its use for evaluating forecast uncertainty/reliability rather than direct statistical model refinement.
• Results were largely consistent when using both NCEP/NCAR and ERA5 reanalysis datasets for circulation type determination.

Contributions

• This study is among the first to investigate the influence of time-varying atmospheric circulation patterns on the skills of short-term (minutes to hours) hydrologic ensemble predictions.
• It identified specific meteorological conditions (wet humidity type and northerly air mass advection) that are conducive to highly skillful short-term multimodel ensemble water level forecasts in a mountainous, flood-prone river basin.
• The findings provide a qualitative framework that can be integrated into multi-criteria hydrologic decision-making processes to assess forecast uncertainty and reliability.

Funding

• National Science Centre, Poland (Sonata programme, grant no. 2011/01/D/ST10/04171).

Citation

@article{Niedzielski2025Relationship,
  author = {Niedzielski, Tomasz and Ojrzyńska, Hanna and Miziński, Bartłomiej and Kryza, Maciej and Spallek, Waldemar},
  title = {Relationship between skills of multimodel hydrologic ensemble predictions and atmospheric circulation patterns: a case study from the Nysa Kłodzka river basin (SW Poland)},
  journal = {Acta Geophysica},
  year = {2025},
  doi = {10.1007/s11600-025-01733-z},
  url = {https://doi.org/10.1007/s11600-025-01733-z}
}