Search

Interpretation of changes of global temperature is important for our understanding of the climate system and for our confidence in projections for the future. Massive efforts have been devoted to improve the accuracy of reproducing the global temperature by the available climate models, but the hindcasts are still inaccurate. Notwithstanding the need to further advance climate models, one may consider data-driven approaches, providing practically useful results in a simpler and faster way. Without assuming any prior knowledge about physics and without imposing a model structure that encapsulates the existing knowledge about the underlying processes, we hindcast global temperature by automatically identified evolutionary computation models. We use 60 years of records of global temperature and climate drivers, with training and testing periods being 1950-1999 and 2000-2009, respectively. This paper demonstrates that the global temperature observed in the past is mimicked with reasonably good accuracy. Evolutionary computation holds promise for modeling the global climate system, which looks hopelessly complex in classical perspective. [ABSTRACT FROM AUTHOR]

The article focuses on the research on ocean circulation in a warming climate. Accordingly, findings show that the Last Glacial Maximum (LGM) had stronger winds which fit to the common observation that it is windier in the winter than in the summer because there is greater thermal contrast within the atmosphere in the winter hemisphere. Moreover, the paper cited several observations that predict that oceans' circulation will weaken in response to global warming, but the warming at the end of the last ice age suggests a different outcome. However, by studying these observations, it is clear that large circulation changes took place and it seems unlikely that circulation changes of this magnitude could have happened without substantial changes in the wind forcing.