Evaluation of Stochastic Perturbed Parameterization Tendencies on Convective-Resolving Ensemble Forecasts of Heavy Rainfall Events in New York and Taiwan

Heavy rainfall episodes often cause significant societal disruption; therefore, it is imperative to understand the predictability of these events. One way to understand forecast predictability is to employ ensemble forecasts, which can account for uncertainty both in the initial conditions and the model formulation. Stochastic methods, such as stochastic perturbed parameterization tendencies (SPPT), represent one way of parameterizing model uncertainty by randomly perturbing individual physics tendencies at each time step during forecast simulation. While this method has been used extensively in global ensemble prediction studies, it has received relatively less attention for high-resolution heavy precipitation forecasts, including how to optimally tune it for grid spacings below 10 km.

This research aims to understand the predictability of heavy precipitation events and improve ensemble forecasts of these events by employing SPPT to convective-resolving WRF forecasts. Two case studies from different parts of the world, but similar synoptic characteristics (i.e., moist southwesterly flow from the deep tropics), are studied. The first case is the remnants of Tropical Storm Lee, which brought heavy rainfall to New York on 6-8 Sep 2011, while the second case is a Meiyu-related heavy rainfall event in Taiwan on 16 June 2008 during SoWMEX-IOP8 in 2008. The talk will document the sensitivity of probabilistic forecasts of these events to modifications in the SPPT parameters (i.e., perturbation length and time scales, and amplitude). In addition, the independent SPPT (iSPPT) scheme will be used to understand how internal uncertainty in the microphysical, radiation, and PBL parameterizations modulate the predictability of these events.