Chinese Journal of Polar Research

Chinese Journal of Polar Research ›› 2026, Vol. 38 ›› Issue (2): 207-228.DOI: 10.13679/j.jdyj.20260019

Previous Articles     Next Articles

Construction and evaluation of an Arctic regional atmospheric reanalysis system based on Polar WRF-Gridpoint Statistical Interpolation (GSI)

CHEN Junming1,4, LI Zhe2, CHENG Wei3, DING Minghu1,5, ZHANG Lei1   

  1. 1Chinese Academy of Meteorological Sciences, Beijing 100081, China;
    2Key Laboratory of Climate Resource Development and Disaster Prevention for Gansu Province, College of Atmospheric Sciences, Lanzhou University, Lanzhou 730000, China;
    3Beijing Institute of Applied Meteorology, Beijing 100029, China;
    4Hebei Key Laboratory of Meteorological Artificial Intelligence, Xiong’an Institute of Meteorological Artificial Intelligence, Xiong’an New Area 070001, China;
    5Key Laboratory of Polar Atmosphere-Ocean-Ice System for Weather and Climate of the MOE, Department of Atmospheric and Oceanic Sciences & Institute of Atmospheric Sciences, Fudan University, Shanghai 200438, China
  • Received:2026-04-09 Revised:2026-06-02 Online:2026-06-30 Published:2026-07-13

PDF (PC)

20

Abstract:

This study developed a high-resolution regional atmospheric reanalysis system for the Arctic using the Polar WRF mesoscale numerical model and the Gridpoint Statistical Interpolation (GSI) variational data assimilation system. The system assimilates multi-source observations, including surface observations, radiosonde data, and satellite remote sensing data, to generate a 10 km-resolution Arctic regional atmospheric reanalysis dataset for 2012–2016. The system was evaluated using surface and radiosonde observations, ERA5 reanalysis data, and GPM precipitation products. The results show that: (1) for near-surface variables, WRF performs better than ERA5 in simulating 2 m air temperature and 10 m wind speed, with root mean square error (ERMSE) values of 2.008 ℃ and 1.707 m·s–1, respectively, lower than the corresponding ERA5 values of 2.196 ℃ and 2.018 m·s–1; however, the ERMSE values for 2 m dew-point temperature and sea-level pressure are 1.984 ℃ and 1.303 hPa, respectively, higher than those of ERA5. (2) For upper-air variables, WRF shows advantages in geopotential height and relative humidity, with relative humidity ERMSE values of 11.5%~12.4%, lower than the ERA5 values of 18.0%~19.5%; however, the ERMSE values for temperature and wind speed are slightly higher than those of ERA5. (3) For precipitation, WRF can reproduce the spatial pattern characterized by higher precipitation over the peripheral seas and lower precipitation over the central Arctic Basin. The smallest errors occur in July, with mean error (EME) and ERMSE values of 0.04 mm·d–1 and 0.57 mm·d–1, respectively; the ERMSE values in January and October are 1.11 mm·d–1 and 1.03 mm·d–1, respectively, indicating larger precipitation magnitude errors in winter and autumn. (4) A case study of the strong Arctic cyclone in August 2012 shows that WRF can capture the 500 hPa low-pressure center, cold-core structure, humidity distribution, and peripheral high-wind-speed belt. Overall, the system demonstrates good applicability in reconstructing the atmospheric state and precipitation spatial structure over the Arctic.


Key words: Arctic, atmospheric reanalysis dataset, Polar WRF model, GSI data assimilation system, data evaluation