Do Double‐Moment Microphysics Schemes Make Reliable Predictions on the Raindrop Number Concentration?: A Squall‐Line Case Study

• Published in: Journal of geophysical research. Atmospheres 2023-05, Vol.128 (9), p.n/a
• Main Authors: Jin, Han‐Gyul, Baik, Jong‐Jin
• Format: Article
• Language: English
• Subjects: Bias; Cloud droplet concentration; Cloud droplets; Cloud microphysics; Cloud models; Clouds; Diameters; disdrometer observation; double‐moment bulk microphysics scheme; Droplets; Geophysics; Graupel; Ground-based observation; Ice particles; Melting; microphysical parameterizations; Microphysics; Modelling; Performance evaluation; Precipitation; Predictions; Rain; Rain water; raindrop number concentration; Raindrop size distribution; Raindrops; Rainfall; Rainfall forecasting; Rainfall simulators; Size distribution; Spheroids; Squalls; Uncertainty; Weather forecasting
• ISSN: 2169-897X; 2169-8996
• DOI: 10.1029/2022JD038394

Utilizing raindrop size distribution observations in the evaluation of a cloud microphysics scheme helps identifying systematic deficiencies of the scheme and improving it. In this study, we evaluate the performance of seven different double‐moment bulk microphysics schemes in predicting raindrop number concentration Nr in squall‐line simulations using ground‐based disdrometer observations. While all seven schemes show good performances in the prediction of rainwater mass, they show significant biases from the observation and severe discrepancies from each other in the prediction of Nr. The Thompson, Thompson aerosol‐aware, and Weather Research and Forecasting double moment 6‐class (WDM6) schemes, where the main source of Nr is melting of snow and graupel, overestimate Nr by four times, which may be caused by overestimation of Nr production via melting. The Morrison and predicted particle properties (P3) schemes, where the autoconversion of cloud droplets into raindrops is an Nr source that is as important as melting, exhibit negative biases of Nr, which may be attributed to insufficient collisional breakup of raindrops. The P3_Nc scheme, the P3 scheme that prognoses cloud droplet number concentration, shows two prominent modes in Nr distribution, one of which consists of overestimated Nr produced by overly active autoconversion. The ice‐spheroids habit model with aspect‐ratio evolution scheme overestimates Nr but not as much as the two Thompson schemes and WDM6 scheme, showing the most realistic prediction of mean raindrop diameter. The high uncertainty in Nr prediction needs to be reduced by carefully selecting parameterizations for important microphysical processes, in order to maximize the advantage of using double‐moment microphysics schemes. Plain Language Summary The raindrop number concentration is one of the variables predicted by weather models, but the accuracy of the prediction has not been well evaluated. This study evaluates the predictions of raindrop number concentration by seven different cloud models, which gives implications on whether the cloud models reasonably represent the cloud microphysical processes where the raindrop number concentration is involved. For the evaluation, the observation data from a ground‐based disdrometer, which measures each raindrop's size and fall speed, are used. In the simulations of a rainfall case, the cloud models all show biased predictions on the raindrop number concentration, in contrast with their pre