Kašpar et al. (2025) Improving design precipitation estimates by combining estimates from high-resolution adjusted radar data and long-term ombrographic measurements

Identification

• Journal: Atmospheric Research
• Year: 2025
• Date: 2025-10-11
• Authors: Marek Kašpar, Filip Hulec, Miloslav Müller, Lenka Crhová
• DOI: 10.1016/j.atmosres.2025.108557

Research Groups

• The Czech Academy of Sciences, Institute of Atmospheric Physics, Prague, Czechia
• Charles University, Faculty of Science, Prague, Czechia
• Czech Hydrometeorological Institute, Prague, Czechia

Short Summary

This study introduces an innovative geostatistical merging method for design precipitation estimates, combining high-resolution adjusted radar data and long-term ombrographic measurements. Applied to Czechia, the approach yields enhanced spatial accuracy and reliability, crucial for robust flood risk management and infrastructure planning.

Objective

• To propose and verify an innovative synergy method for estimating design precipitation by merging two design precipitation fields: one derived from adjusted high-resolution radar data and another from long-term ombrographic measurements.
• To leverage the advantages of both datasets (spatial detail from radar, reliability from gauges) while mitigating their limitations (radar uncertainty, gauge sparsity) to produce more robust and spatially accurate design precipitation estimates.

Study Configuration

• Spatial Scale: The entire territory of Czechia (approximately 79,000 km²), with radar data on a 1 km by 1 km grid and ombrographic data from 164 stations.
• Temporal Scale:
  • Accumulation Durations: 1 hour, 6 hours, and 24 hours.
  • Return Periods: 2 years, 5 years, 10 years, 20 years, 50 years, and 100 years.
  • Radar Data Series: 20 years (2002–2021) of 10-minute precipitation intensities, restricted to the frost-free period (April–October).
  • Ombrographic Data Series: Ranging from 20 to 70 years (1951–2022) of 1-minute digitized records.

Methodology and Data

• Models used:
  • Generalized Extreme Value (GEV) distribution for fitting annual maxima of precipitation totals.
  • L-moment-based index storm procedure for GEV parameter estimation.
  • Region-of-Influence (ROI) method for stabilizing local extreme-value estimates.
  • Empirical Bayesian Kriging (EBK) for interpolating ratios between ombrographic and adjusted radar design precipitation totals.
  • Polynomial Depth-Duration-Frequency (DDF) curves for ombrographic design precipitation.
• Data sources:
  • Adjusted radar data: 20-year series of 10-minute precipitation intensities on a 1 km by 1 km grid, derived from C-band Doppler radars (Brdy and Skalky) operated by the Czech Hydrometeorological Institute (CHMI). These data were adjusted using daily precipitation totals from approximately 700 CHMI rain gauges.
  • Ombrographic data: Time series of annual maxima of precipitation intensities (2 minutes to 96 hours) from 164 ombrographs and automatic rain gauges across Czechia (1951–2022), provided by CHMI.

Main Results

• Empirical Bayesian Kriging (EBK) applied to the ratios between ombrograph- and adjusted radar-derived design totals was identified as the most effective merging method through leave-one-out cross-validation.
• The combined design precipitation estimates exhibit enhanced spatial accuracy, better reflecting the true spatial gradients of extreme precipitation.
• Resulting design precipitation fields show significant spatial variability at shorter durations (1 hour, 6 hours) due to convective precipitation, while 24-hour design totals are smoother and concentrated in mountainous regions, consistent with stratiform rainfall patterns.
• The merging procedure effectively increases underestimated radar-based short-term design totals; for example, adjusted radar 1-hour design precipitation totals were initially 90.6% (2-year return period) and 86.5% (100-year return period) of the ombrographic averages.
• Final combined design precipitation totals for Czechia range from 12.5 mm to 28.9 mm (1-hour, 2-year) up to 62.4 mm to 235.6 mm (24-hour, 100-year).
• Compared to other existing datasets for Czechia, the combined dataset generally yields higher design precipitation totals (e.g., 7% to 22% higher for 1-hour duration, 2-year to 100-year return periods), providing more conservative and safer estimates.

Contributions

• Introduction of an innovative and generally applicable geostatistical merging method for design precipitation, combining high-resolution adjusted radar data and long-term ombrographic measurements.
• Development of enhanced spatial resolution and reliability for design precipitation estimates, crucial for robust flood risk management, infrastructure design, and adaptation planning in regions with diverse precipitation regimes.
• Refinement of existing radar-only design precipitation estimates by incorporating the reliability of long-term gauge data and a robust merging procedure.
• Demonstration of the importance of radar adjustments and regional frequency analysis in improving design precipitation estimates.
• Addressing and mitigating the systematic underestimation of short-term design precipitation totals inherent in radar-only or adjusted radar-based approaches.

Funding

• Technology Agency of the Czech Republic, grant no. SS02030040 (Prediction, Evaluation and Research for Understanding National sensitivity and impacts of drought and climate change for Czechia (PERUN)).
• Johannes Amos Comenius Programme (P JAC), project no. CZ.02.01.01/00/22_008/0004605 (Natural and anthropogenic georisks).

Citation

@article{Kašpar2025Improving,
  author = {Kašpar, Marek and Hulec, Filip and Müller, Miloslav and Crhová, Lenka},
  title = {Improving design precipitation estimates by combining estimates from high-resolution adjusted radar data and long-term ombrographic measurements},
  journal = {Atmospheric Research},
  year = {2025},
  doi = {10.1016/j.atmosres.2025.108557},
  url = {https://doi.org/10.1016/j.atmosres.2025.108557}
}