Assessing the Impact of Turbulence Parameterizations on Typhoon Precipitation Forecast

Improving the typhoon precipitation forecast using the convection-permitting model remains challenging. The impact of cumulus and subgrid-scale turbulence parameterization on precipitation prediction in the grey zone is still debating. This study investigates the influence of cumulus parameterization and a reconstruction and nonlinear anisotropy (RNA) turbulence scheme on Typhoon precipitation prediction by analyzing multiple typhoon cases. The simulation results indicate that incorporating the cumulus and RNA scheme in the CR simulation results in increased domain-averaged precipitation and improved recall scores across various precipitation thresholds. The enhancement is particularly notable for extreme events, as demonstrated by a doubled recall rate at the 110mm threshold for Typhoon Hato compared to simulations without any cumulus scheme. Moreover, the relative error reduction due to using the RNA scheme reveals considerable benefits in predicting extreme precipitation events, with lower error rates for all examined typhoon cases. On the other hand, simulations utilizing the Smagorinsky scheme display marginal progress compared to those without any turbulence scheme. Furthermore, implementing cumulus parameterization can eliminate spurious precipitation while preserving the typhoon structure associated with the RNA scheme. The improved forecasting ability for extreme precipitation events is attributed to the optimal arrangement of dissipation and backscatter. The RNA scheme generates upgradient momentum transport in the lower troposphere, which dynamically reinforces typhoon circulation. The horizontal downgradient (backscatter) mixing of potential temperature in the upper (lower) level heightens the buoyancy flow towards the eyewall and diminishes the depletion of the convective core, which is conducive to developing heavy precipitation.