CFE agenda 091117 - Laserfiche WebLink

Browse Search

among atmosphere, oceans, land surface, and ice in order to estimate the likelihood of changes in temperature, precipitation, and other climate factors (Hayhoe et al. 2010). These models are complex, as they simulate the climate system in three dimen- sions (Viner 2000) (Figure 1 -2). Atmosphere -only GCMs were the first generation of climate models, and were used to simulate the equilibrium response of the climate system to a doubling of atmospheric CO, (Viner 2000). More recent models build on the AGCMs with coupled Atmosphere -Ocean General Circulation Models (AOGCMs). AOGCMs are more complex and incorporate additional factors such as sea ice, evapotranspiration over land, and the feedback interactions between the ocean and atmo- sphere (Randall et al. 2007, Hayhoe et al. 2010). Most importantly, these models are able to dynami- cally model the ocean, which has a significant impact on the climate system as a whole. There have been major advances in the development of climate models over the last 20 years, and current models provide a reliable guide to future conditions at a coarse scale, given a particular scenario (Randall et al. 2007). One way GCMs are evaluated is by simulating the historic climate using past observed concentrations of greenhouse gas emissions, and then compare those model outputs to the observed climate (Weart 2009). Climate models have success- fully reproduced the main features of the current climate, including temperature changes over the last hundred years, as well as the main features of the Holocene period (6,000 years ago) and the Last Glacial Maximum (21,000) years ago (Weart 2009). By evaluating models against past climate data, scientists are able to identify potential causal mecha- nisms of climate change, and use that information to project the main features of the future climate (Jones 2000). Models are continually tested and scrutinized, and there are ongoing improvements in computational ability as well as resolution. The ability of AOGCMs to simulate extreme events, such as hot and cold spells, has also improved, although the frequency and amount of precipitation falling in intense events are underestimated (Randall et al. 2007). Models are able to project some climate variables, such as temperature, with a higher degree of confidence than other variables, such as precipitation. However several decades of develop- ment have resulted in a robust and unambiguous picture of significant global warming in response to increasing greenhouse gases (Randall et al. 2007) (Figure 1 -3). There are more than 20 climate models included in the third phase of the Coupled Model Intercom - parison Project (CMIP3), which was developed to serve the IPCC Working Group I for the Fourth Assessment Report (Meehl et al. 2007). Some of these models are better at reproducing observed climate and trends over the past century in partic- ular geographic regions than others (Hayhoe et al. 2010). However, for the purposes of analyzing the potential impacts of climate change, the multimodel ensemble average provides a more robust picture of future climate conditions than any one model (Pierce et al. 2009). Furthermore, choosing one model for use requires a detailed understanding of the climate dynamics in the region of interest (Hayhoe et al. 2010). In most cases, when evaluating the potential impacts of climate change in a given region it is best to use the multi -model ensemble average instead of choosing one or two (Pierce et al. 2009). to Io I D revv r!; c, "a / rl ed rrlat'e r l ("�i°Ie <, i l 1, <,e, One of the drawbacks of the current generation of GCMs is that the resolution is fairly course, upwards of several hundred kilometers (K. Hayhoe et al. 2010). To develop projections of regional climate changes based on global concentrations of green- house gas emissions, the global climate models must be downscaled to transform the large -scale output generated to a regional scale. The main approaches to downscaling are statistical and dynamical down - scaling.