Evaluation of ERA5 Reanalysis Precipitation Over the Democratic People's Republic of Korea

ABSTRACT

Accurate precipitation data are critical for climate monitoring, hydrological modeling, and disaster risk management, yet ground‐based observations remain sparse in topographically complex regions such as the Democratic People's Republic of Korea (DPRK). This study presents a comprehensive station‐based validation of ERA5 reanalysis precipitation over the DPRK for the period 1981–2020, using quality‐controlled observations from 39 meteorological stations. Statistical metrics (correlation coefficient, relative bias, root mean square error), Mann–Kendall trend analysis, and Empirical Orthogonal Function (EOF) decomposition were employed to assess spatiotemporal performance across seasonal, topographic, and interannual scales. Results indicate that ERA5 reproduces the annual climatology and seasonal cycle with moderate skill (annual CC = 0.77) but exhibits systematic, intensity‐dependent biases: reliable performance for stations classified as having moderate annual precipitation totals (< 800 mm year ⁻¹ annual mean), overestimation of high‐magnitude summer events, and persistent false‐alarm biases for light winter precipitation. The < 800 mm year ⁻¹ threshold represents a station‐level classification based on 40‐year annual means, not a monthly threshold applied to individual observations. Performance degrades systematically with elevation, with high‐altitude stations (> 700 m) showing the largest positive bias (+34.5%) and weakest correlation. A critical discrepancy emerges in long‐term trends: while observations indicate predominantly positive precipitation trends (+1.1 mm year ⁻¹ mean slope), ERA5 yields consistently negative trends (−3.3 mm year ⁻¹ ), with directional agreement at only 14% of stations. Despite point‐scale biases, EOF analysis confirms that ERA5 reliably captures dominant spatial patterns of interannual variability (first three modes explaining > 84% of observed variance; pattern correlations r  = 0.84–0.86). We recommend that raw ERA5 precipitation not be used directly in hydrological applications over the DPRK without seasonally stratified, elevation‐aware bias correction. These findings provide a foundational reference for the rational application of reanalysis data in regional climate assessment, water resource management, and climate change adaptation planning in data‐sparse, topographically complex monsoon regions.