Implementation and evaluation of SNICAR snow albedo scheme in Noah-MP
(version 5.0) land surface model
Abstract
The widely-used Noah-MP land surface model (LSM) currently adopts snow
albedo parameterizations that are semi-physical in nature with
nontrivial uncertainties. To improve physical representations of snow
albedo processes, a state-of-the-art snowpack radiative transfer model,
the latest version of Snow, Ice, and Aerosol Radiative (SNICAR) model,
is integrated into Noah-MP in this study. The coupled Noah-MP/SNICAR
represents snow grain properties (e.g., shape and size), snow aging, and
physics-based snow-aerosol-radiation interaction processes. We compare
Noah-MP simulations employing the SNICAR scheme and the default
semi-physical Biosphere-Atmosphere Transfer Scheme (BATS) against
in-situ snow albedo observations at three Rocky Mountain field stations.
The agreement between simulated and in-situ observed ground snow albedo
in the broadband, visible, and near-infrared spectra is enhanced in
Noah-MP/SNICAR simulations relative to Noah-MP/BATS simulations. The
SNICAR scheme improves the temporal variability of modeled broadband
snow albedo, with a nearly twofold higher correlation with observations
(r=0.66) than the default BATS snow albedo scheme (r=0.37). The
underestimated variability in Noah-MP/BATS is a result of inadequate
physical linkage between snow albedo and environmental/snowpack
conditions, which is substantially improved by the SNICAR scheme.
Importantly, the Noah-MP/SNICAR model, with constraints of snow grain
size from the MODIS snow covered area and grain size (MODSCAG) satellite
data, physically represents and quantifies the snow albedo and
absorption of shortwave radiation in response to snow grain size,
non-spherical snow shapes, and light-absorbing particles (LAPs). The
coupling framework of the Noah-MP/SNICAR model provides a means to
reduce the bias in simulating snow albedo.