Transferability of regional climate models over different climatic domains

• Main Author: Gbobaniyi, Emiola Olabode
• Other Authors: Hewitson, Bruce
• Format: Doctoral Thesis
• Language: English
• Published: University of Cape Town 2014
• Subjects: Environmental and Geographical Science
• Online Access: http://hdl.handle.net/11427/4854

Includes abstract. === Includes bibliographical references (p. 119-144). === In the continuing quest to improve climate model predictions to meet the increasing demand for knowledge on the regional effects of global climate change, it is pertinent to increase our understanding of how the underlying processes of climate are represented in the models we use to make these predictions. Concerted efforts in model evaluations and intercomparison have provided numerous insights into various model biases which plague current state-of-the-art regional climate models (RCMs). Model evaluation and assessment is crucial to model development and understanding how physical processes are represented in models is necessary for improving model parameterizations. This thesis explored model transferability as a new approach for systematic process-based intercomparison of RCMs. It investigated an untested transferability hypothesis which states that “for non-monsoon regions experiencing weak synoptic scale forcing, the height of the cloud base is correlated with the daytime surface fluxes”. An initial transferability experiment was conducted over Cabauw, the Netherlands (51.97°N, 4.93°E) to assess the models’ skill in resolving the diurnal and seasonal cycles and to investigate the simulated connections between surface and hydrometeorological variables over a non-monsoon station. The ability of models to resolve these cycles correctly is a good metric of their predictive capabilities. The data used for the study comprises three-hourly surface observations for the period October 2002 – December 2004 from the Coordinated Enhanced Observing Period (CEOP) measuring campaigns of the Global Energy and Water Cycle Experiment (GEWEX) and three-year simulations (2002 -2004) from five RCMs (CLM, GEMLAM, MRCC, RCA3 and RSM). In simulating seasonal and diurnal cycles of CBH and surface variables, the European models (CLM and RCA3) demonstrate a clear home advantage over the North American models (GEMLAM, MRCC and RSM). Principal component analysis revealed that the models couple the cloud base height with surface fluxes as in observations and that this coupling is not sensitive to changes in wind speed. This study found that summer daytime loadings gave the strongest couplings of variables. Three major processes were identified over Cabauw. First and most dominant is the surface energy process which couples sensible and latent heat with net radiation. The second process is thermodynamic, coupling temperature and surface moisture (specific humidity), and the third is a dynamic process which couples pressure and wind speed. A model intercomparison was then carried out across the six midlatitude domains to test the validity of the Cabauw findings. In observations, CBH is well coupled with the surface fluxes over Cabauw, Bondville, Lamont and BERMS, but coupled only with temperature over Lindenberg and Tongyu. All the models (except GEMLAM) simulated a good coupling with surface fluxes at all stations. In GEMLAM, there is no coupling between CBH and surface fluxes at any station. In less homogenous domains of the study, a very slight decrease in the strength of coupling is seen in most of the models, under strong large scale forcing. This would suggest that the coupling between cloud base height and surface fluxes in the models is possibly more influenced by radiative forcing than by synoptic controls. This second study confirmed the findings at Cabauw that the simulated cloud base is correlated with surface energy fluxes and the sign of the correlations in the models is as in observations. This finding is important for the modeling community as it establishes the fact that the models are actually simulating the direction of influence of surface fluxes and possibly, soil water variability, on cloud processes.