Numerical simulation and intercomparison of boundary layer structure with different PBL schemes in WRF using experimental observations at a tropical site

In this study the performance of seven PBL parameterizations in the Weather Research and Forecast (WRF-ARW) mesoscale model was tested at the tropical site Kalpakkam. Meteorological observations collected during an intense observation campaign for wind field modeling called Round Robin Exercise (RRE) were used for comparison. High resolution simulations were conducted for a warm summer condition on 22–24 September 2010. The observations included GPS Sonde vertical profiles, surface level data from meteorological towers and turbulent fluxes from sonic anemometers. Sensitivity experiments with seven PBL schemes [Mellor-Yamada-Janjic (MYJ), Mellor-Yamada-Nakanishi-Niino (MYNN), Quasi Normal Scale Elimination (QNSE), Yonsei University (YSU), Asymmetric Convective Model (ACM2), Bougeault-Lacarrére (BL), Bretherton-Park (UW)] indicated that while all the schemes similarly produced the stable boundary layer characteristics there were large differences in the convective daytime PBL. It has been found that while ACM2 and QNSE produced highly unstable and deep convective layers, the UW produced relatively shallow mixed layer and all other schemes (YSU, MYNN, MYJ, BL) produced intermediately deep convective layers. All the schemes well produced the vertical wind directional shear within the PBL. A wide variation in the eddy diffusivities was simulated by different PBL schemes in convective daytime condition. ACM2 and UW produced excessive diffusivities which led to relatively weaker winds, warmer and dryer mixed layers with these schemes. Overall the schemes MYNN and YSU simulated the various PBL quantities in better agreement with observations. The differences in the simulated PBL structures could be partly due to various surface layer formulations that produced variation in friction velocity and heat fluxes in each case.