3.3    Modifications to the MYNN PBL and Surface Layer Scheme for      WRF-ARW

Olson, Joseph B. and John M. Brown, National Oceanic and Atmospheric Administration/ESRL/GSD/AMB

The Mellor-Yamada-Nakanishi-Niino (MYNN) planetary boundary layer (PBL) scheme was introduced into the Advanced Research version of the Weather Research and Forecasting model (WRF-ARW) a few years ago as an alternative turbulent kinetic energy (TKE)-based scheme. The unique features of this scheme include closure constants tuned to match large eddy similations, a more elaborate mixing length formulation, and an option to run at both closure levels 2.5 and 3.0. The MYNN surface layer (SFC) scheme originated from an old version of the Yonsei University (YSU) SFC scheme, with few modifications to customize the performance with the MYNN PBL scheme. Since implementation, many shortcomings have been noted, such as a warm-bias over desert/bare soil regions, production of negative TKE, high 10-m wind speed bias, excessive polar fog in cool season nocturnal conditions, and excessive low-level clouds over the ocean. This work presents modifications made to the MYNN PBL and SFC scheme to alleviate these problems.

Beginning with the negative TKE problem, the method of Canuto et al. (2008) is implemented, which results in an elimination of the critical Richardson number, allowing the mixing of momentum to persist in stable conditions, similar to the QNSE or TEMF PBL schemes. This modification fixes the negtive TKE problem but results in a distinct behavioral change that necessitates successive effort to reduce an over-diffusive nature in stable conditions. Changes to fundamental closure constants and the mixing lengths return the mixing behavior to near previous performance, while further effort is made to improve other biases noted above. Furthermore, modifications to the SFC scheme reduces the high 2-m temperature bias over the southwest U.S. as well as the high-bias in low-level clouds over the oceans. With these improvements, the MYNN PBL/SFC schemes become a realistic candidate for implementation into the Rapid Refresh and High-Resolution Rapid Refresh forecast systems.