The impact of surface and PBL schemes on the initiation of convection over the tropical ocean and land

Description

This study aims to identify the weakness of current PBL and surface schemes in capturing the initiation of convection over different underlying surfaces with the goal of developing improved parameterizations. We performed a series of regional simulations with the WRF model using a wide range of PBL, surface, and convection schemes over the Indian Ocean and Amazon regions using data from AMIE/DYNAMO and GoAmazon. The work consists of four portions: 1) we evaluated the simulated precipitation against multiple observations, as well as the simulated vertical profiles of cloud and atmospheric state variables against observations; 2) we conducted simulations with three different PBL schemes (i.e. YSU, MYJ and UW) and compared the boundary layer structure and response to the vertical transport of water vapor within the simulations; 3) we compared two different surface schemes, i.e. MM5 vs. Eta to look at the responses of simulated surface flux and PBL clouds; 4) we conducted another set of simulations similar to (2) and (3) but with nudging applied to force the simulations towards the simulated water vapor. The nudging tendency terms are examined to identify errors in the observed moistening and heating processes. Results show that the PBL and surface schemes have surprisingly large impacts on the simulations, with different characteristics between tropical ocean and land and between wet and dry seasons. All three PBL schemes tested tend to overestimate precipitation and moisture in PBL and free troposphere, and consequently produce larger moist static energy over the tropical ocean. Vertical transport of water vapor is too active in all PBL schemes and moisture nudging tends to suppress the initiation of convection and reduces the overprediction of precipitation. Discrepancies are found in the simulation of the surface moisture flux. The MM5 surface scheme predicts much larger surface moisture flux and consequently yields simulations with larger amounts of PBL moisture and precipitation than using the Eta scheme. When Eta surface scheme is applied over the ocean, the UW PBL scheme predicts a larger surface wind speed than is observed and yields values of moisture flux that are more sensitive to the surface wind speed than was measured resulting in much larger latent heat flux than that in the MYJ scheme. Our results highlight the need for additional studies to better understand issues associated with the surface, PBL, and convective schemes applied in WRF.