Sensitivity of Arctic climate to spectral nudging in the Regional Arctic System Model

Friday, 19 December 2014
Mimi Hughes1, John J Cassano2, Andrew Roberts3 and Wieslaw Maslowski3, (1)University of Colorado at Boulder, Boulder, CO, United States, (2)Univ Colorado, Boulder, CO, United States, (3)Naval Postgraduate School, Monterey, CA, United States
Large-domain regional climate models often suffer from biases in their atmospheric circulation, moisture, and temperature fields due, in part, to poor representation of planetary scale waves and stratospheric processes. Spectral nudging – where the largest scales of a finite-area model are incrementally nudged toward a reference dataset – is one method of reducing these biases. A few recent publications have investigated the impact of spectral nudging on atmospheric regional climate simulations, however testing of nudging within a fully coupled system has not been performed. Thus to eliminate atmospheric biases in the Regional Arctic System Model and explore the sensitivity of the coupled system to the degree of data assimilation and thus atmospheric constraint, we perform a series of sensitivity experiments applying spectral nudging to WRF with the WRF Four-dimensional Data Assimilation (FDDA) package. Using FDDA allows us to tightly constrain RASM-WRF by its atmospheric reanalysis boundary conditions (ERA Interim), which already include direct data assimilation of a very large quantity of observations (i.e., satellite radiances, radiosondes, etc).

With well-configured nudging in RASM-WRF and standalone WRF, we can virtually eliminate atmospheric biases in highly constrained and well-observed variables (e.g. SLP). However, both coupled and uncoupled simulations still have some large biases compared to atmospheric reanalysis in variables that aren’t directly nudged in WRF or as well-observed (e.g., radiation and cloud cover). Using a series of multi-decade WRF and RASM simulations driven with reanalyses, we examine the sensitivity of large-scale atmospheric climate representation (i.e., SLP biases, etc) and the surface interface (heat fluxes, etc), as well as sea ice diagnostics (e.g., extent and concentration), to nudging in our large-domain, coupled system. We show that with no nudging RASM’s climate moves to a very unrealistic state, but that with over-constrained nudging we get unrealistic interchanges of fluxes at the surface interface (since the reanalysis forcing has different sea surface temperatures and weather than RASM’s ocean), highlighting the importance of nudging configuration in regional climate model simulations.