Zhong et al. (2025) Evaluation of WRF Planetary Boundary Layer Parameterization Schemes for Dry Season Conditions over Complex Terrain in the Liangshan Prefecture, Southwestern China

Identification

• Journal: Atmosphere
• Year: 2025
• Date: 2025-12-31
• Authors: Jinhua Zhong, Debin Su, Zijun Zheng, Wenyu Kong, Peng Fang, Fang Mo
• DOI: 10.3390/atmos17010053

Research Groups

• College of Electronic Engineering, Chengdu University of Information Technology, Chengdu, China
• Liangshan Yi Autonomous Prefecture Meteorological Bureau, Xichang, China

Short Summary

This study evaluates six WRF Planetary Boundary Layer (PBL) schemes against multi-source observations over complex terrain in the Liangshan Prefecture during clear-sky dry seasons. It finds that the QNSE and MYNN2.5 schemes offer the most balanced performance in simulating near-surface variables, vertical profiles, and PBL height, providing guidance for high-resolution modeling in the region.

Objective

• To systematically evaluate the performance of six Weather Research and Forecasting (WRF) Planetary Boundary Layer (PBL) parameterization schemes (ACM2, BL, MYJ, MYNN2.5, QNSE, and YSU) in simulating near-surface meteorological variables, vertical atmospheric profiles, and Planetary Boundary Layer Height (PBLH) over the complex terrain of the Liangshan Prefecture, Southwestern China. The evaluation focuses on clear-sky dry season conditions (spring and winter) using comprehensive multi-source observational data to provide a reference for high-resolution numerical weather prediction models in this region.

Study Configuration

• Spatial Scale:
  • Study area: Liangshan Yi Autonomous Prefecture, Southwestern Sichuan Province, China (26°03′–29°18′ N, 100°03′–103°52′ E), with elevations ranging from 300 m to 4500 m.
  • WRF nested domains: 9 km (158 × 138 grid points), 3 km (286 × 241 grid points), and 1 km (337 × 325 grid points) horizontal resolutions.
  • Vertical resolution: 45 terrain-following levels with a model top at 50 hPa (approximately 20 km altitude).
  • Observation station elevation: 1590.8 m.
• Temporal Scale:
  • Simulation periods: Two clear-sky dry season cases.
    • Spring: 8–11 April 2025 (78-hour simulation, 72-hour analysis).
    • Winter: 24–27 December 2024 (78-hour simulation, 72-hour analysis).
  • Initialization: 18:00 UTC on the day prior to the case period, with a 6-hour spin-up.
  • Radiosonde observations: Twice daily at 07:15 and 19:15 Local Standard Time (UTC + 8 hours).
  • Surface observations: 10-minute averages.
  • Wind Profiling Radar (WPR) data: 6-minute intervals.

Methodology and Data

• Models used:
  • Weather Research and Forecasting (WRF) model version 4.2 with the Advanced Research WRF (ARW) dynamical solver.
  • Planetary Boundary Layer (PBL) parameterization schemes evaluated: Asymmetrical Convective Model version 2 (ACM2), Bougeault–Lacarrère (BL), Mellor–Yamada–Janjic (MYJ), Mellor–Yamada–Nakanishi–Niino Level 2.5 (MYNN2.5), Quasi-Normal Scale Elimination (QNSE), and Yonsei University (YSU).
  • Microphysics scheme: Lin scheme.
  • Cumulus parameterization: Kain-Fritsch (disabled in innermost domains d02, d03).
  • Shortwave radiation scheme: Dudhia scheme.
  • Longwave radiation scheme: Rapid Radiative Transfer Model (RRTM).
  • Land-surface model: Noah.
• Data sources:
  • Initial and lateral boundary conditions: National Centers for Environmental Prediction (NCEP) 0.25° × 0.25° Final Operational Global Analysis (FNL) dataset, available at 6-hour intervals.
  • Observational data for validation (from a common station):
    • Digital GTS13 Radiosonde (RS) system: High-resolution atmospheric profiles (60 m vertical resolution) of temperature (±0.3 °C accuracy), relative humidity (±5%), and pressure (±2.0 hPa).
    • Surface meteorological stations: 10-minute averaged 2 m temperature (T2), 2 m relative humidity (R2), surface pressure, 10 m wind speed (WS10), and 10 m wind direction (WD10).
    • TWP-8 L-band PBL Wind Profiling Radar (WPR): High-temporal-resolution data (6-minute intervals) with 120 m vertical resolution, including power spectrum and instantaneous radial velocity spectra. WPR-derived PBLH was determined through wavelet transform analysis of signal-to-noise ratio (SNR) data.

Main Results

• Near-Surface Meteorological Variables:
  • The QNSE scheme demonstrated the best overall performance for 2 m temperature (T2) in both spring (RMSE = 1.79 °C, MBE = 0.10 °C, R = 0.94) and winter (RMSE = 2.98 °C, MBE = 1.86 °C, R = 0.89), accurately capturing diurnal variations and large temperature amplitudes. Non-local schemes (ACM2, YSU) tended to overestimate nighttime T2 due to excessive mixing.
  • All schemes showed notable dry biases for 2 m specific humidity (Q2) at night, suggesting an underestimation of moisture accumulation. The BL scheme showed a weak correlation for Q2 in spring (R = -0.09).
  • All schemes captured the diurnal fluctuation pattern of 10 m wind speed (WS10), with the MYJ scheme performing best overall (Spring: RMSE = 1.46 m s⁻¹, MBE = 0.57 m s⁻¹, R = 0.58). All schemes underestimated extreme near-surface wind values.
• Surface Fluxes:
  • The QNSE scheme produced the largest daytime sensible heat flux (HFX) and surface temperature (TSK), indicating more efficient upward heat transfer and higher turbulent exchange efficiency.
  • QNSE also yielded the smallest ground heat flux (GRDFLX), suggesting less net energy storage in the soil due to enhanced partitioning of available surface energy into turbulent fluxes, consistent with dry season conditions.
• Vertical Profiles:
  • QNSE and MYJ schemes best represented the lower-to-middle level thermodynamic structure (potential temperature and specific humidity) compared to radiosonde observations in both seasons.
  • All schemes simulated warmer potential temperature profiles than observations in April, with non-local schemes (YSU, ACM2) being the warmest.
  • Specific humidity was generally underestimated below 1800 m and overestimated above in December.
  • Simulated wind speeds were generally stronger than observed in the lower-to-middle layers but weaker above 1000 m in December.
• Planetary Boundary Layer Height (PBLH):
  • All schemes reproduced the diurnal cycle of PBLH, but with differences in amplitude.
  • In spring, MYNN2.5 (RMSE = 532.85 m, MBE = -72.92 m, R = 0.778) and BL (RMSE = 526.32 m, MBE = -107.28 m, R = 0.840) schemes showed the lowest RMSE and highest correlation with WPR observations.
  • In winter, QNSE performed best in capturing daytime PBLH (RMSE = 669.46 m, MBE = 238.68 m, R = 0.507), despite occasionally producing unrealistic spikes.
  • The MYJ scheme consistently underestimated PBLH in both seasons.
  • PBLH in winter generally reached lower maximum heights (typically below 2000 m) compared to spring (1500–2500 m).
• Overall Performance: No single scheme performed optimally for all variables and seasons. However, QNSE and MYNN2.5 demonstrated the most balanced overall performance across both spring and winter.

Contributions

• This study provides the first systematic evaluation of six WRF PBL parameterization schemes over the complex terrain of the Liangshan Prefecture, Southwestern China, utilizing comprehensive multi-source observational data (radiosondes, surface stations, and wind profiling radar).
• It offers specific recommendations for selecting optimal PBL schemes (QNSE and MYNN2.5) for high-resolution WRF simulations in this complex geographical area, particularly for applications in PBL dynamics, air-quality modeling, and fire-weather forecasting during dry-season conditions.
• The research highlights the distinct scheme-dependent differences and clear seasonal variability in model performance, emphasizing the importance of regional and seasonal assessment for PBL parameterizations.

Funding

• Sichuan Province Science and Technology Plan Project (grant number: 2024YFTX0016)
• Science and Technology Development Fund Project of Sichuan Provincial Key Laboratory of Rainstorm, Drought and Flood Disasters in Plateau and Basin (grant number: SCQXKJYJXMS202210)
• Liangshan Prefecture Science and Technology Plan Project (grant number: 23ZDYF0182)

Citation

@article{Zhong2025Evaluation,
  author = {Zhong, Jinhua and Su, Debin and Zheng, Zijun and Kong, Wenyu and Fang, Peng and Mo, Fang},
  title = {Evaluation of WRF Planetary Boundary Layer Parameterization Schemes for Dry Season Conditions over Complex Terrain in the Liangshan Prefecture, Southwestern China},
  journal = {Atmosphere},
  year = {2025},
  doi = {10.3390/atmos17010053},
  url = {https://doi.org/10.3390/atmos17010053}
}