Deng et al. (2025) Enhancing Forecasting Capabilities Through Data Assimilation: Investigating the Core Role of WRF 4D-Var in Multidimensional Meteorological Fields

Identification

• Journal: Atmosphere
• Year: 2025
• Date: 2025-11-12
• Authors: Yujiayi Deng, Xiaotong Wang, Xinyi Fu, Nian Wang, Hongyuan Yang, Shuhui Zhao, Xiurui Guo, Jianlei Lang, Ying Zhou, Dongsheng Chen
• DOI: 10.3390/atmos16111286

Research Groups

• Key Laboratory of Beijing on Regional Air Pollution Control, Beijing University of Technology, Beijing, China

Short Summary

This study systematically evaluated the WRF 4D-Var data assimilation system, implementing a novel two-layer nested "assimilation-forecast" workflow, and found significant improvements in multidimensional meteorological forecasts across different seasons and underlying surface types, with benefits persisting for approximately 12 hours.

Objective

• To systematically evaluate the effectiveness of the WRF 4D-Var data assimilation system in enhancing the accuracy of numerical weather prediction (NWP) for multidimensional meteorological fields, particularly over complex terrain and across different seasons and underlying surface conditions.

Study Configuration

• Spatial Scale: Two-layer nested domains: Domain 1 (outer) covering most of China with 27 km × 27 km horizontal resolution; Domain 2 (inner) targeting eastern China with 9 km × 9 km horizontal resolution.
• Temporal Scale: Two one-month simulation periods (February and June 2019) representing winter and summer conditions. Daily 6-hour WRF 4D-Var assimilation window (00:00 to 06:00 UTC) followed by a 24-hour forecast.

Methodology and Data

• Models used: Weather Research and Forecasting (WRF) model v4.6 and WRF 4D-Var data assimilation system v4.6.
• Data sources:
  • Observations for assimilation: NCEP ADP global upper air and surface weather observations (SYNOP/METAR, SHIP/BUOY, TEMP/RAOB, AIREP/AMDAR).
  • Initial and lateral boundary conditions: ECMWF ERA5 reanalysis (0.25° × 0.25° horizontal resolution, hourly temporal resolution).
  • Observations for evaluation: National Climate Data Center (NCDC) for near-surface variables (2 m temperature, 2 m relative humidity, 10 m wind speed, 10 m wind direction); University of Wyoming’s 12-hour resolution radiosonde dataset for upper-air variables (temperature, relative humidity, wind speed, wind direction at 500 hPa and other standard pressure levels).

Main Results

• Near-surface variables: 4D-Var significantly improved correlation coefficients for near-surface variables. In winter, improvements were 2.9% for temperature, 14.5% for relative humidity, 6.6% for wind speed, and 10.4% for wind direction. In summer, improvements were even greater: 13.3% for temperature, 5.8% for relative humidity, 35.3% for wind speed, and 42.3% for wind direction.
• Upper-air variables: 4D-Var considerably enhanced atmospheric vertical profiling, with the middle troposphere (300–700 hPa) showing the most pronounced improvement rates for temperature, relative humidity, and wind speed.
• Underlying surface types: Water bodies exhibited the strongest and most stable positive assimilation response. Grassland showed systematic negative effects on humidity forecasts, and cropland showed similar negative effects on winter temperature forecasts, indicating limitations in current parameterization schemes.
• Background field changes: In February, assimilation led to a regionally coherent warming (0–2 °C) and dehumidification of the 2 m fields, with northward adjustments in 10 m wind direction. In June, adjustments were more complex with alternating positive and negative increments and finer spatial distribution, reflecting multi-scale summer weather processes.
• Persistence of benefits: Positive effects persisted throughout the 24-hour forecasts, with the maximum benefit occurring within the first 12 hours. The decay rate of benefits was faster in winter than in summer for all variables.

Contributions

• Developed and implemented a novel two-layer nested "assimilation-forecast" workflow for WRF 4D-Var, overcoming its inherent limitation of not supporting cross-domain online bidirectional nested assimilation, to achieve staged optimization from large to small scales.
• Conducted a systematic and in-depth assessment of WRF 4D-Var performance across different meteorological variables, seasons (winter and summer), and various underlying surface conditions, addressing limitations of previous studies that focused on specific cases or single seasons.
• Provided detailed insights into the spatial heterogeneity of assimilation benefits and their correlation with underlying surface types, identifying specific land cover types (e.g., water bodies, grasslands, croplands) with distinct assimilation responses.
• Clarified the intrinsic relationship between background field adjustments and forecast improvements, and analyzed the decay patterns of assimilation benefits with forecast lead time, providing guidance for optimal assimilation update frequencies (e.g., 12-hour cycle).

Funding

• National Natural Science Foundation of China (No. 42305125)

Citation

@article{Deng2025Enhancing,
  author = {Deng, Yujiayi and Wang, Xiaotong and Fu, Xinyi and Wang, Nian and Yang, Hongyuan and Zhao, Shuhui and Guo, Xiurui and Lang, Jianlei and Zhou, Ying and Chen, Dongsheng},
  title = {Enhancing Forecasting Capabilities Through Data Assimilation: Investigating the Core Role of WRF 4D-Var in Multidimensional Meteorological Fields},
  journal = {Atmosphere},
  year = {2025},
  doi = {10.3390/atmos16111286},
  url = {https://doi.org/10.3390/atmos16111286}
}