Ryu et al. (2026) RRBF-KMA: high-resolution radar–gauge merged precipitation dataset for South Korea (2016–2024)

Identification

• Journal: Earth system science data
• Year: 2026
• Date: 2026-10-26
• Authors: Soorok Ryu, Joon Jin Song, Kyo Sun Lim, GyuWon Lee
• DOI: 10.5194/essd-18-7679-2026

Research Groups

• BK21 Weather Extremes Education & Research Team, Department of Atmospheric Sciences, Center for Atmospheric Remote sensing (CARE), Kyungpook National University, Daegu, Republic of Korea
• Department of Statistical Science, Baylor University, Waco, TX, USA
• School of Earth and Environmental Sciences, Seoul National University, Seoul, Republic of Korea

Short Summary

This study introduces RRBF-KMA, a high-resolution (0.5 km, 10 min) radar–gauge merged precipitation dataset for South Korea (2016–2024). The dataset, generated using a temporally constrained bias correction and Residual Radial Basis Function (RRBF) interpolation, provides consistent precipitation fields crucial for hydrological and meteorological applications.

Objective

• To develop and validate a long-term, high-resolution (0.5 km spatial, 10 min temporal) radar–gauge merged precipitation dataset for South Korea (2016–2024) to support hydrological and meteorological applications requiring sub-hourly rainfall information.

Study Configuration

• Spatial Scale: 0.5 km spatial resolution on a regular grid of 2305 pixels × 2881 pixels, covering South Korea.
• Temporal Scale: 10 min temporal resolution, covering the period from 1 January 2016 to 31 December 2024 (9 years).

Methodology and Data

• Models used:
  • Radar–gauge merging framework: Temporally constrained radar–gauge bias correction scheme with an exponentially weighted moving average (EWMA) for smoothing.
  • Residual interpolation: Residual Radial Basis Function (RRBF) method with an exponential RBF kernel (shape parameter s = 16 km).
  • Radar rainfall estimation: Operational Z–R relationship (Z = 148 R^1.59) adopted by the Korea Meteorological Administration (KMA).
  • Validation: Leave-one-out cross-validation (LOOCV).
• Data sources:
  • Radar: Nationwide composite reflectivity fields (Hybrid Surface Rainfall, HSR) from the KMA API Hub, provided at 5 min intervals with 500 m × 500 m resolution, derived from 11 KMA weather radars (10 S-band, 1 C-band).
  • Rain gauges: Automatic Weather System (AWS) and Automated Synoptic Observing System (ASOS) networks operated by KMA (approximately 550 AWS, 100 ASOS stations), providing rainfall observations at 1 min temporal resolution with 0.5 mm (AWS) and 0.1 mm (ASOS) measurement resolution.

Main Results

• The RRBF-KMA dataset provides continuous, high-resolution precipitation fields for South Korea from 2016 to 2024 at 0.5 km spatial and 10 min temporal resolution.
• Compared to radar-only Z–R estimates, the RRBF method significantly reduced bias and improved correlation with rain gauge observations. For 10 min accumulations, the Pearson correlation coefficient (COR) increased from 0.7854 to 0.8318, and the mean absolute error (MAE) decreased from 0.62 mm to 0.50 mm, with normalized bias error (NBE) near zero (1.06%).
• For 60 min accumulations, the RRBF method achieved a COR of 0.9386 and an MAE of 0.93 mm/h, outperforming Z–R (COR = 0.8762, MAE = 1.53 mm/h) and mean field bias (MFB) (COR = 0.8949, MAE = 1.41 mm/h) estimates.
• Long-term validation (2016–2024) for hourly rainfall ≥ 0.1 mm/h showed that RRBF reduced normalized mean absolute error (NMAE) from 53.38% (Z–R) to 28.85% and increased COR from 0.7703 to 0.9251, with NBE reduced from -13.13% to 0.58%.
• The RRBF estimates maintained near-zero bias (within ±3.3%) and high correlation (COR > 0.8699) across various rainfall intensity thresholds, including higher rates (up to 5.0 mm/h).
• Seasonal analysis revealed higher correlations in summer (JJA) and autumn (SON) (>0.92) compared to winter (DJF) (~0.86), with larger dispersion during winter.
• Spatially, RRBF estimates exhibited reduced bias magnitude and more uniform correlation patterns across stations compared to radar-only products, particularly in complex terrain and coastal regions.
• The dataset robustly preserves interannual variability, identifying 2017 as the driest and 2020 as the wettest year, without abrupt discontinuities.
• The merging procedure is computationally efficient, requiring approximately one minute per 10 min analysis, which supports near-real-time implementation.

Contributions

• Public release of RRBF-KMA, a unique long-term (9-year), high-resolution (0.5 km spatial, 10 min temporal) radar–gauge merged precipitation dataset for South Korea, addressing a critical need for detailed rainfall information in the region.
• Development and application of an innovative radar–gauge merging framework that combines a temporally constrained bias adjustment (using an exponentially weighted moving average) with Residual Radial Basis Function (RRBF) interpolation, effectively reducing spurious temporal variability and preserving fine-scale radar-derived variability while incorporating gauge-based corrections.
• Comprehensive validation demonstrating the robust and consistent performance of the RRBF method across various temporal scales, rainfall intensity thresholds, and seasons over a 9-year period, highlighting its superior accuracy compared to radar-only estimates.
• The dataset serves as a valuable resource for a wide range of applications, including extreme rainfall analysis, hydrological modeling, urban flood forecasting, and satellite precipitation validation, particularly in a region characterized by complex terrain and highly variable precipitation.

Funding

• Kyungpook National University Development Project Research Fund, 2025.
• Korea Meteorological Administration Research and Development Program under grant no. RS-2023-00237740.
• National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIT) (grant no. RS-2026-25488670).

Citation

@article{Ryu2026RRBFKMA,
  author = {Ryu, Soorok and Song, Joon Jin and Lim, Kyo Sun and Lee, GyuWon},
  title = {RRBF-KMA: high-resolution radar–gauge merged precipitation dataset for South Korea (2016–2024)},
  journal = {Earth system science data},
  year = {2026},
  doi = {10.5194/essd-18-7679-2026},
  url = {https://doi.org/10.5194/essd-18-7679-2026}
}