Uncertainty Quantification And Parameter Tuning In Convection Schemes In Climate Models And The Physical Impacts Of Simulated Convection On Regional And Global Climate

Climate models always have various parameters with large ranges of their possible values. The calibrations of these parameters are important not only for the reduction of model uncertainties but also for better understanding the atmospheric dynamical and physical processes. In this thesis, we apply Uncertainty Quantification (UQ) and parameter tuning in both Regional Climate Model (RCM) and Global Climate Model (GCM) using a stochastic importance sampling algorithm which can progressively converge to the optimal parameters. Several key parameters in Kain-Fritsch convection scheme in the Weather Research and Forecasting (WRF, i.e. RCM) model and in Zhang-McFarlane deep convection scheme in Community Atmosphere Model version5(CAM5, i.e. GCM) are examined, respectively. Our purposes are to explore the utility of high-resolution observed precipitation for improving simulations of regional patterns, evaluate the transferability of UQ and parameter tuning across physical processes, spatial scales, and climatic regimes, improve the modeling of convective precipitation (i.e. overestimated convective precipitation), and evaluated the impact of improved deep convection on the global circulation and climate. The RCM results show that the precipitation bias in the model can be significantly reduced when optimal parameters identified by the sampling algorithm are used. The model performance is found to be sensitive to downdraft-and entrainment-related parameters and consumption time of Convective Available Potential Energy (CAPE). Simulated convective precipitation decreases as the ratio of downdraft to updraft flux increases. Larger CAPE consumption time results in less convective but more stratiform precipitation. The simulation using optimal parameters obtained by constraining only precipitation generates positive impact on the other output variables, such as temperature and wind. By using the optimal parameters obtained at25-km simulation, both the magnitude and spatial pattern of simulated precipitation are improved at12-km spatial resolution. The optimal parameters identified from one region also improve the simulation of precipitation when the model domain is moved to another region with a different climate regime. These results suggest that benefits of optimal parameters determined through vigorous mathematical procedures are transferable across processes, spatial scales, and climatic regimes to some extent.The GCM results show that the simulated convective precipitation is most sensitive to the parameters of CAPE consumption time scale, parcel fractional mass entrainment rate and maximum downdraft mass flux fraction. Using the optimal parameters constrained by the Tropical Rainfall Measuring Mission (TRMM) observed convective precipitation remarkably improves the simulation of convective to stratiform precipitation ratio and rain rate spectrum (i.e. frequency of occurrence). When the convections are suppressed, precipitation tends to be more confined to the regions with strong atmospheric convergence. As the optimal parameters are used, the positive impacts on some aspects of the atmospheric circulation and climate, e.g. reduction of double-ITCZ, improved East Asian monsoon precipitation, and improved annual cycles of the cross-equator jets, can be found, as a result of the redistribution of latent heat release from the modified convective system. The positive impacts of the optimized parameters derived from the2-degree simulations are found to transfer to the1-degree simulations to some extent.This UQ and parameter tuning study with both RCM and GCM provide some useful insights on the sources of model biases among different physical processes and motivate further work to assess the strategies for UQ and parameter tuning through improved understanding of the internal physical and dynamical processes of the atmospheric system at both regional and global scale.