Improved Surface Layer Simulation Using Refined Vertical Resolution in the GFDL Atmospheric General Circulation Model

Surface layer (SL) diagnostic variables in atmospheric models – such as 2-m temperature (T2) and 10-m wind (U10) – are computed by applying Monin-Obukhov (M-O) similarity theory to the lowest model level height (LMH). This assumes the LMH is within SL height (SLH), although most atmospheric general circulation models (AGCMs) hardly satisfy this in stable boundary layers (SBLs) in particular. To assess errors in the SL simulation solely due to the violation of the assumption, one-dimensional offline tests of M-O similarity theory are performed with the LMHs ranging from 1 to 100 m at intervals of 1 m for three SBL cases: weakly stable, transition stability, and very stable regimes. The cases are set up with surface parameters and vertical profiles based on the three Global Energy and Water Exchanges (GEWEX) Atmospheric Boundary Layer Study (GABLS) intercomparison studies. The offline runs showed that T2 and U10 are underestimated by O(0.1 K) and O(0.1 m s⁻¹), respectively, when the LMH is higher than the SLH. The more stable the surface layer, the larger the errors in the SL parameters. Based on the offline tests an additional vertical layer is added in the GFDL AGCM, and the LMH is lowered from 34 m to 18 m in average. Overall, the negative biases in the simulated climatologies of T2 and U10 are reduced in the experiment with lower LMH, regardless of the season.