Cloud-resolving models (CRMs) link synoptic-scale circulation patterns to sub-kilometer-scale cloud structure and precipitation using non-hydrostatic dynamics in concert with parameterizations of sub-grid turbulence, microphysics, and radiative processes. To diagnose these models with observational data, one must simultaneously measure the meteorological drivers of cloud formation, the cloud properties themselves, and the spatiotemporal distribution of precipitation generated by the clouds. Reanalysis data alone fail to provide adequate diagnostics, since they typically estimate cloud properties and precipitation from the large-scale flow using coarser parameterizations than the CRMs themselves.
We present a new, combined dataset at ~8-km and hourly resolution over the eastern and Midwestern CONUS from 2002-2020 that bridges this gap.
The dataset provides ERA-5 reanalysis data for fields relevant to cloud and precipitation formation, GOES-East geostationary satellite data for cloud structure, and ground-based NEXRAD radar data for precipitation, all on the same space-time grid.
We demonstrate the added value in using observed precipitation and cloud fields instead of reanalysis with a particular eye toward capturing extreme events. The dataset can be leveraged to build and validate realistic stochastic parameterizations of precipitation in future climate models.