| Summary: | MSc (Environmental Sciences), North-West University, Potchefstroom Campus, 2014 Global Circulation Models (GCMs) provide the basis of our capacity to simulate, understand and predict climate variability and change. These models are based on established physical laws and have proven fidelity for assessing changes to global quantities (Randall et&al. 2007; Bengtsson et&al. 2007; Gualdi et&al. 2008; Oouchi et&al. 2006; Smith et&al. 2010; Sugi et&al. 2002; Zhao et&al. 2009). However, GCMs typically are of too coarse a resolution to directly infer climatology of high=impact weather at local scales and it is common to downscale over regions of interest using either statistical techniques or dynamical downscaling (Regional Climate Models = RCMs). RCMs are also based on established physical laws, with the added benefit of high=resolution enabling them to better simulate local effects and high=impact weather. One of their weaknesses is the cost of running the models, making it near impossible to run enough simulations to fully define uncertainty. This research utilizes the Weather Research and Forecasting Model (WRF; Skamarock et&al. 2008) as a climate model, using GCMs to downscale to regional scales (Bender et& al. 2010; Knutson et& al. 2007, 2008; Walsh et&al. 2004). The paper “Modeling"High.Impact"Weather" and" Climate:" Lessons" from" a" Tropical" Cyclone" Perspective”' (Chapter 6: Done et&al. 2013), presented here describes the development and implementation of the WRF model as a regional climate model. This paper also addresses the lessons learned and some best practices for using WRF as a regional climate model. It is known that GCMs suffer from biases (Liang et&al. 2008; Xu and Yang 2012). Unfortunately, biases that may be acceptable at global scales may irretrievably change = or even destroy = extreme weather signals, when used as driving data for RCMs (Ehret et& al. 2012). The focus of this study is not to merely run the WRF model as an RCM, but finding improved ways to utilize these biased GCM data as RCM drivers. The paper Bias"Corrections"of"Global"Models"for"Regional"Climate"Simulations"of" High.Impact" Weather”" (Chapter 5: Bruyere et& al. 2013) presented here describes in detail the problems associated with driving RCMs with GCM data containing biases. A new bias correction method, whereby the climate change signal and variability are retained from the GCM while removing the systematic mean errors, is presented in this paper. Statistical models encapsulate empirical relationships and enable inferences of extremes from low= resolution data. These, combined with low=resolution global models provide a low=cost method of downscaling and assessing uncertainty. The major disadvantage is that they do not directly encapsulate the laws of physics. Building on work previously done by Emanuel and Nolan (2004) and Emanuel (2010), the North Atlantic basin is used as a test case to develop an improved basin specific empirical tropical cyclone genesis index. This is presented in the paper “Investigating"the"Use"of"a"Genesis"Potential"Index"for" Tropical" Cyclones" in" the" North" Atlantic" Basin” (Chapter 3: Bruyere et& al. 2012). Using this initial development, similar indices are developed for the other tropical cyclone basins. This work is presented in the paper “Exploring"Genesis"Potential"Indices”"(Chapter 4: Bruyere and Holland 2014). A Hybrid Statistical=Dynamical approach provides an attractive way to harness the strengths from both statistical techniques and nested regional climate models. With this approach dynamical models are used as a baseline, while the statistical models provide additional local information and an improved assessment of uncertainty. With the use of this hybrid statistical=dynamical approach we can make better inferences regarding the effect climate change will have on rare and small=scale extreme events, with tropical cyclones used as a single example NCAR, Research Partnership to Secure Energy for America (RPSEA) and NSF EASM Doctoral
|
|---|