Analysis of high-resolution cloud simulations using dynamical downscaling and data assimilation

Description

Parametric representations of cloud/precipitation processes continue to be adopted in climate simulations with increasingly higher spatial resolution or with emerging adaptive mesh framework, and it is only becoming more critical that such parameterizations have to be scale-aware. Continuous cloud measurements at DOE's ARM sites have provided a strong observational basis for novel cloud parameterization research at various scales. Despite significant progress in our observational ability, there are important cloud-scale physical and dynamical quantities that are either not currently observable or insufficiently sampled. Outputs from cloud-resolving simulations are often used to provide such cloud-scale fields.

To complement the ARM measurements, the cloud-resolving simulations have to be configured in realistic settings. This is achieved with 3D cloud-resolving dynamical downscaling using Weather Research and Forecasting (WRF) model and multi-domain nesting. However, there is no guarantee that simply cloud-resolving would produce high-confidence cloud-scale data. A number of factors may have important influence on the cloud-resolving simulations. These factors include, but are not limited to, domain size, spatial resolution, model top, forcing data set, model physics and the growth of model errors. In addition to sensitivity experiments to basic model configurations, a multi-scale data assimilation system is further adopted to minimize model errors and improve the simulation of cloud-scale physical and dynamical quantities. The hydrometeor advection that may play a significant role in hydrological process within the observational domain but is often lacking, and the limitations due to the constraint of domain-wide uniform forcing in conventional cloud system-resolving model simulations, are at least partly accounted for in our approach. The analysis focuses on the evaluation of both mean and probability distribution of macrophysical and microphysical properties, as well as eddy transports that are central to the parameterization of sub-grid processes in coarser-resolution models.