Evaluating the Land-Atmosphere-Cloud Interaction in HRRR using New York
State Mesonet
Abstract
Key Points: The HRRR model near surface thermodynaic biases are
seasonally dependent. A systematic warm and dry bias present during the
warm season. The summer warm and dry biases over farmland are
consistently larger than forest. Hydrological bias of spring snow melt
might eventually lead to the summer warm and dry biases over farmland.
Abstract The High-Resolution Rapid Refresh (HRRR) model version 3
meteorological fields were examined using data from New York State
Mesonet (NYSM) sites from an entire year. In this work, the land
surface, atmosphere and cloud coupling systems are evaluated as an
integrated system. The physical processes influencing soil hydrological,
surface thermodynamic processes from surface fluxes to boundary layer
convection are investigated from both temporal (seasonal and diurnal)
and spatial perspectives. Results show that the model 2m surface biases
are seasonally dependent, with warm and dry bias present during the warm
season, and an extreme cold bias during the night in winter. The summer
warm bias includes both a land-surface-induced bias and a cloud-induced
bias. Inacurate representation of energy partition and soil hydrological
process across different land use types and hydrological bias of spring
snow melt in the land surface model is the main source of the
land-surface induced bias. A feedback loop linking cloud presence,
radiative flux changes and temperature contributes to the cloud-induced
bias. The positive solar radiation bias increases from clear sky to
overcast sky conditions, beyond simply the lack of aerosols in the
current version. The most significant bias occurs during overcast and
thick cloud conditions associated with frontal passage and
thunderstorms.