Sensitivity of diurnal variation in simulated precipitation during East Asian summer monsoon to cumulus parameterization schemes

The capability to simulate the diurnal variation of precipitation over East Asia region during the summertime of 2011 is investigated using five different cumulus parameterization schemes with the Weather Research and Forecasting model. A semidiurnal cycle with a 12 h interval over land and a diurnal cycle with a 24 h interval over ocean are commonly found in all simulations, consistent with the observed diurnal cycle. Two observed dominant peaks in the early morning and afternoon are reproduced in all simulations. With overestimated precipitation rate, however, the simulated afternoon peaks occur earlier than the observed peaks by 2 h for the Kain-Fritsch (KF) and Simplified Arakawa-Schubert schemes, and by 3 h for the Betts-Miller-Janjic and Tiedtke schemes. The overestimation of simulated precipitation frequency leads to amplitude and phase errors in the precipitation rate, and the early peak time of simulated precipitation intensity intensifies the phase error in the simulation over land. The KF scheme with alternative trigger function (KFtr) based on moisture advection provides slightly better results in terms of alleviating the overestimated precipitation rate and frequency and delaying the afternoon peaks. Additional sensitivity simulations based on the change of temperature perturbation in the trigger function of the KF and KFtr schemes demonstrate the afternoon peak tends to be delayed as temperature perturbation decreases, implying the significant role of convective initiation frequency in determining diurnal peaks of precipitation. Modulation of temperature perturbation alleviates the precipitation frequency bias, while it could not resolve the precipitation intensity bias.