Sensitivity of the WRF model to boundary layer and land surface parameterizations: comparisons with Differential Absorption Lidar and Eddy Covariance measurements

At the Institute of Physics and Meteorology at the University of Hohenheim (UHOH) an advanced model system has been developed for seamless prediction of weather and climate from nowcasting to the decadal time scales based on the Weather Research and Forecasting model coupled with Noah with multi-physics options (WRF-NOAH-MP). Using this system, land-surface-atmosphere feedback processes down to the grey zone will be investigated. In this study seven simulations were carried out with WRF in 36-hour cycles over a 25-day period for western and south-western Germany, with 2 km horizontal grid spacing and 89 vertical levels. The sensitivity of WRF to three “local” [Mellor-Yamada-Janjic (MYJ), Mellor-Yamada-Nakanishi-Niino, level 2.5 (MYNN 2.5) and quasi-normal scale elimination (QNSE)] and two “non-local” [the asymmetric convective model, version 2 (ACM2) and Yonsei University (YSU)] PBL schemes, and to two land surface models [LSM; Noah and Noah-MP]. For process studies and model verification we used high resolution absolute humidity measurements (with a range of 15-150 m) performed with the differential absorption lidar (DIAL) of UHOH. In this study we demonstrate the high potential of these measurements for studying the grid-cell averaged structure of the humidity profile, height of the convective PBL, residual moist layer and development of the PBL in the course of the day. For evaluating the model performance at the land surface, we compared simulated surface variables with the measurements at six Eddy covariance(EC)stations located in SW Germany. This study shows significantly higher sensitivity of WRF to the LSMs than to the PBL schemes, which is evident not only in the lower PBL, but it extends to the PBL entrainment zone and even up to lower troposphere. Noah-MP produces higher and drier convective PBL then Noah, which is showed to be associated primarily to entrainment of air from above the PBL. The MYNN 2.5 PBL scheme exhibits slightly better representation of the surface scalar variables by reducing the biases and being somewhat better correlated with measurements, especially in respect to the other two “local” schemes. Furthermore, it outperforms the YSU “non-local” scheme in representing the convective PBL structures. This study is the first step within the ongoing research project which includes the downscaling to gray zone and sensitivity studies in various stability environments.