Quantification of the Importance of Parameterized Deep Convective Characteristics to Successful Simulation of Convective Transport

Deep moist convection is an important transport mechanism in the tropopause and lower stratosphere. The wide range of spatial and temporal scales deep convective processes span can result in some processes not being explicitly resolved in current model simulations, thereby requiring parameterization. Evaluation of parameterized convective processes such as entrainment and vertical transport, however, is complicated by the inaccurate timing and location of simulated convection.

To identify ways to improve convective parameterization and vertical transport, this study evaluates the relative importance of (1) correct convective timing and location, (2) correct characterization of the convective entrainment profile, and (3) correct convective cloud depth to subsequent simulation of transport. Simulations of convection observed during the Deep Convective Clouds and Chemistry (DC3) field campaign are conducted using the Weather Research and Forecasting Coupled with Chemistry Model (WRF-Chem) with the Kain-Fritsch (KF) convective parameterization. A lightning data assimilation (LDA) technique is used to trigger convection within the KF scheme where lightning was observed, thereby improving simulated convective timing and location. A new modification to this technique that incorporates assimilation of satellite-estimated cloud top height is tested to improve the depth of the parameterized convection. A final sensitivity test is conducted by modifying the parameterization of entrainment to more closely mimic idealized, high-resolution WRF-Large Eddy Simulations (LESs) of the same storm system.

The results of these three tests, each designed to improve a different aspect of parameterized deep convection within a simulation, are compared to an unmodified control simulation as well as DC3 observations and fields from the idealized WRF-LESs. Mixing ratios of carbon monoxide within storm outflow aloft from both the test simulation and DC3 aircraft measurements are compared. Entrainment profiles are estimated from the same aircraft measurements, from the WRF-LESs using inclusion of tracers, and directly output from the WRF-Chem test simulations. Total convective mass flux from both the LESs and WRF-Chem tests are also directly compared. These comparisons allow quantification of which of the three relevant convective characteristics: timing and location, entrainment profile, and/or cloud depth, contributed most to improvement of the simulated convective transport.