Qiu et al. (2025) Quality control of the composite radar quantitative precipitation estimation product for Great Britain

Identification

• Journal: Journal of Hydrology
• Year: 2025
• Date: 2025-12-12
• Authors: Xiaonuan Qiu, Adèle C. Green, Stephen Blenkinsop, H. J. Fowler
• DOI: 10.1016/j.jhydrol.2025.134755

Research Groups

• School of Engineering, Newcastle University, Newcastle upon Tyne, United Kingdom
• Tyndall Centre for Climate Change Research, Newcastle University, Newcastle upon Tyne, United Kingdom

Short Summary

This study develops a novel quality control (QC) framework for radar Quantitative Precipitation Estimation (QPE) products in Great Britain to systematically detect and correct underestimation (e.g., beam blockage) and overestimation errors. The framework significantly improves the accuracy of radar QPE, reducing RMSE by 29% and increasing the correlation coefficient by 31% compared to rain gauge observations, while preserving real extreme rainfall events.

Objective

• To develop a new, systematic, and holistic quality control (QC) framework for radar Quantitative Precipitation Estimation (QPE) products that operates directly on rainfall values, reduces reliance on ancillary datasets, and addresses a broad spectrum of error types in a unified approach.

Study Configuration

• Spatial Scale: Great Britain (GB), covering the domain of a composite product from 18 UK Met Office C-band rain radars. Data is gridded at 1 km resolution.
• Temporal Scale: Hourly resolution, from 1 January 2006 to 8 October 2018, accumulated from 5-minute radar snapshots.

Methodology and Data

• Models used:
  • A novel rule-based Quality Control (QC) framework comprising:
    • Blockage Area Detection Method
    • Blockage Recovery Method (using linear interpolation)
    • Normal Rainfall Field Classifier
    • Relaxed-threshold Check (combining basic methods)
    • Strict-threshold Check (combining basic methods)
    • Bad Neighbor Check
    • Gauss Rainfall Interpolation Method (based on Gauss–Seidel method for Laplace equation)
  • Basic methods for overestimation detection: Maximum Difference Method, Robust Score Method, Zero Difference Method, Flexible Zero Difference Method, Expand Area Method, High-ranking Rainfall Method, Isolated Rainfall Method, Good Neighbor Method.
• Data sources:
  • UK composite radar QPE (NIMROD System, 1 km resolution, hourly) from the Centre for Environmental Data Analysis (CEDA).
  • Hourly rain gauge records (~1300 gauges across GB) for verification.
  • A separate dataset of 208 manually verified extreme rainfall events for impact assessment.

Main Results

• Radar QPE errors in Great Britain increase with elevation, distance from radar, and rainfall intensity.
• Radar QPE frequently underestimates high-intensity rainfall and fails to detect many events ≥40 mm h−1. Underestimation occurs 1.71 times more frequently than overestimation for rainfall ≥0.2 mm h−1.
• All 18 GB radar stations suffer from beam blockage, with a mean blockage ratio of 21.69%.
• The proposed QC framework significantly improves radar QPE accuracy:
  • Root Mean Square Error (RMSE) reduced from 0.546 mm to 0.386 mm (29% reduction).
  • Correlation Coefficient (CC) increased from 0.552 to 0.725 (31% increase).
  • Mean Absolute Error (MAE) slightly reduced from 0.0942 mm to 0.0934 mm (0.8% reduction).
• For identified "bad" rainfall data (0.37% of total), RMSE was reduced by 81%, MAE by 53%, and CC increased by 476%.
• The framework effectively removes unrealistic large rainfall amounts (e.g., maximum radar rainfall reduced from 427.0 mm to 96.9 mm, aligning with the gauge maximum of 91.8 mm).
• It successfully preserves real extreme rainfall events, with only 3.62% of 387 verified extreme hourly events being wrongly identified as "bad."

Contributions

• Development of a novel, systematic, and holistic rule-based QC framework for radar QPE that directly processes rainfall values, reducing reliance on external ancillary data.
• Comprehensive addressing of both underestimation (beam blockage) and overestimation (malfunction, ground clutter, electronic noise) errors within a unified framework.
• Quantitative assessment of radar QPE error characteristics across Great Britain, highlighting the prevalence of underestimation.
• Demonstrated significant improvements in radar QPE accuracy (RMSE, CC) for Great Britain, crucial for hydrological and climate research, while ensuring the retention of true extreme rainfall events.
• The framework's design, requiring minimal geographical and meteorological knowledge, makes it broadly applicable to various radar network settings globally.

Funding

• Flood Hydrology Improvements Programme, Environment Agency
• CAMELS-GB-v2 (UK Floods and Droughts Research Infrastructure and a UKRI Future Leaders Fellowship, reference MR/V022857/1)
• China Scholarship Council, China (reference No. 202108440108)
• Co-Centre for Climate + Biodiversity + Water funded by UKRI (reference NE/Y006496/1)
• ‘Co-developed Environmental Solutions to Mitigate the Impact of Temperature Extremes on the Health of Vulnerable Populations’ project funded by NERC (reference NE/Y503241/1)

Citation

@article{Qiu2025Quality,
  author = {Qiu, Xiaonuan and Green, Adèle C. and Blenkinsop, Stephen and Fowler, H. J.},
  title = {Quality control of the composite radar quantitative precipitation estimation product for Great Britain},
  journal = {Journal of Hydrology},
  year = {2025},
  doi = {10.1016/j.jhydrol.2025.134755},
  url = {https://doi.org/10.1016/j.jhydrol.2025.134755}
}