Classification of Warm-Season Precipitation in High-Resolution Rapid Refresh (HRRR) model forecasts over the Contiguous United States

Abstract

This study uses the convective adjustment time scale to identify the climatological frequency of equilibrium and non-equilibrium convection in different parts of the Contiguous United States (CONUS) as modeled by the operational convection-allowing High-Resolution Rapid Refresh (HRRR) forecast system. We find a qualitatively different climatology in the northern and southern domains separated by the 40°N parallel. The convective adjustment time scale picks up the fact that convection over the northern domains is governed by synoptic flow (leading to equilibrium) while locally forced, non-equilibrium convection dominates over the southern domains. Using a machine learning algorithm, we demonstrate that the convective adjustment timescale diagnostic provides a sensible classification that agrees with the underlying dynamics of equilibrium and non-equilibrium convection. Furthermore, the convective adjustment time scale can indicate the model quantitative precipitation forecast (QPF) quality, as it correctly reflects the higher QPF skill for precipitation under strong synoptic forcing. This diagnostic based on the strength of forcing for convection will be employed in future studies across different parts of CONUS to objectively distinguish different weather situations and explore the potential connection to warm-season precipitation predictability.