Uncertainty Quantification and Optimization of Precipitating Hydrometeor Parameters for Winter Precipitation in a Cloud Microphysics Scheme
This study investigates model uncertainty associated with 13 key parameters in the Weather Research and Forecasting Double-Moment 6-class (WDM6) bulk-type cloud microphysics scheme and identifies optimized parameter sets through a Bayesian optimization. The 13 parameters define the relationtionships of fall velocity–diameter and mass–diameter, and shape parameters in the drop size distribution of rain, snow, and graupel. Their perturbed ranges are derived from two-dimensional video disdrometer observations collected during the International Collaborative Experiments for the PyeongChang 2018 Olympic and Paralympic Winter Games (ICE-POP 2018) campaign. Based on the parameter ranges, an ensemble of 256 simulations is constructed to quantify the sensitivity of microphysical processe tendencies, hydrometeor mixing ratios, and precipitation to parameter uncertainties. Three distinct precipitation cases during ICE-POP 2018—cold-low, warm-low, and air–sea interaction—are selected to explore the parameter uncertainties and optimization. The results of the perturbed parameter ensemble simulations reveal substantial variability in key microphysical processes, hydrometeor mixing ratios, and the dominant precipitation type in each case. A parameter associated with the mass–diameter relationship of snow is identified as one of the most influential factor. Correlation analysis shows that parameter-induced changes in precipitation can help reduce simulation errors in regions with strong positive biases. Additionally, experiments using optimized parameter sets, identified through Bayesian optimization, show improvements in precipitation simulations. The root mean square error is reduced by 26.9%, 30.2%, and 15.2% in the cold-low, warm-low, and air–sea interaction cases, respectively. These results underscore the value of ensemble-based sensitivity analysis and parameter optimization frameworks for improving simulation accuracy and reducing model uncertainty in cloud microphysics schemes.
- Research Organization:
- Pacific Northwest National Laboratory (PNNL), Richland, WA (United States)
- Sponsoring Organization:
- USDOE
- Grant/Contract Number:
- AC05-76RL01830
- OSTI ID:
- 3000083
- Report Number(s):
- PNNL-SA-217070
- Journal Information:
- Atmospheric Research, Journal Name: Atmospheric Research Vol. 330
- Country of Publication:
- United States
- Language:
- English
Similar Records
A Theory for the Balance between Warm Rain and Ice Crystal Processes of Precipitation in Mixed-Phase Clouds