Abstract: Convection-permitting (CP) simulations were evaluated during the wet season from October 2009 to March 2010, characterized by numerous extreme daily rainfall events in southeastern South America. We analyzed six simulations using WRY (versions 3.8.1 and 3.9.0) and RegCM4, considering parameterized convective processes (20 km resolution, non-CP) and a nested CP domain (4 km resolution, with convective parameterization switched off). First, we assessed the performance of six gridded datasets (CMORPH, ERA5, GSMAP, MSWEP, PERSIANN, and TRMM) in reproducing 3-h accumulated precipitation observed in 54 stations. The station data exhibited highly diverse diurnal cycles of precipitation, with different times for maximum and minimum, often showing double peaks. The most intense peak is observed at 09:00 am local time (LT), representing nighttime precipitation. CMORPH and ERA5 showed better agreement with the observations. The non-CP simulations failed to capture the diversity of diurnal cycles observed, which is better captured by CP simulations. When comparing non-CP and CP, WRF-CP shifts the afternoon peaks (at 12:00 pm and 03:00 pm LT) to the morning (at 06:00?09:00 am LT), while RegCM4-CP shifts some peaks from dawn (at 03:00 am LT) to the morning (at 09:00 am LT). Both shifts, along with the increase in the diversity of diurnal cycles, are notable features to obtain better agreement of CP simulations with local observations. The observed morning peak of mean precipitation diurnal cycles is primarily influenced by higher frequency of heavy rainfall events rather than by their intensities.

Autoría

Descargar referencia