With the increase in computation power in recent years, global storm resolving models (GSRMs), which have ultra-high horizontal resolutions of 1-5 km and are capable of simulating convective storms directly, are now feasible to produce simulations beyond a month. In light of these new global ultra-high-resolution simulations, a question that naturally arises is what aspects of precipitation simulations are better captured by these GSRM models compared to other CMIP-class models, and what problems remain. In this study, scientists at Lawrence Livermore National Laboratory, investigated how well these GSRMs simulate daily and sub-daily precipitation statistics, and compared their performance with coarser-resolution models (~25-500 km).
The GSRMs from the DYAMOND initiative (DYnamics of the Atmospheric general circulation Modeled On Non-hydrostatic Domains) provide a great opportunity for the community to investigate detailed cloud and convection processes and their interactions with large-scale environment under the ultra-high-resolution configuration. It is hoped that further study of these simulations to better understand the key processes relevant to convection and precipitation biases will provide guidance for future model development of both GSRMs and coarse resolution GCMs. Ultimately, the demonstrated superiority of GSRMs could help better inform how these statistics of large precipitation events will change with a warming climate.
Compared to two satellite rainfall datasets, GSRMs (~4km) convincingly exhibit superior performance for statistics of heavier rain rate events including their diurnal cycle, spatial propagation, and the amount contributed by intense precipitation, but not for statistics of weaker or shorter duration precipitation. Both high- and standard-resolution models from CMIP6/HighResMIP (~50km) and CMIP6/AMIP (~100km), respectively, fail to simulate the correct phase and amplitude of the diurnal cycle of precipitation and the propagating convection in the Central US, but high-resolution models show relative improvement in the distribution of precipitation frequency and amount, especially for intense precipitation.