Precipitation Variation over Eastern China and Arid Central Asia During the Past Millennium and Its Possible Mechanism: Perspectives from PMIP3 Experiments

Jian Shi,Qing Yan,Dabang Jiang,Jinzhong Min,Ying Jiang
DOI: https://doi.org/10.1002/2016jd025126
2016-01-01
Journal of Geophysical Research Atmospheres
Abstract:Multiproxies suggest a tripole humidity pattern in Asia in the Medieval Climate Anomaly (MCA, 950–1250 A.D.) and Little Ice Age (LIA, 1500–1800 A.D.), with drier (wetter) conditions in arid central Asia (ACA), wetter (drier) conditions in North China, and drier (wetter) conditions in South China. However, the mechanisms behind this reconstructed humidity variation remain unclear. In this study, we investigate Asian humidity changes by using the last millennium simulations of the Paleoclimate Modelling Intercomparison Project Phase III (PMIP3). The results indicate that only one out of nine PMIP3 models (Meteorological Research Institute Coupled ocean‐atmosphere General Circulation Model version 3) can well reproduce the reconstructed humidity pattern. This model indicates that the tripole humidity pattern is mainly caused by precipitation changes in spring and summer and is prominent in the past millennium on a multidecadal time scale. In spring, the reduction (increase) of precipitation in ACA and South China is attributed to the northward (southward) shift of the westerlies and a weakened (strengthened) western Pacific subtropical high in the MCA (LIA). In summer, precipitation over ACA decreases (increases) due to a local descending (ascending) motion, while abundant (deficient) precipitation over eastern China results from the enhanced (depressed) summer monsoon. Moreover, we suggest that a La Niña (El Niño)‐like condition may be the primary reason the tripole precipitation pattern was maintained in the MCA (LIA), although a warmer (colder) North Pacific and North Atlantic also play a role. The mechanisms must be further validated since most simulations fail to reproduce the reconstructed humidity condition in the MCA/LIA, making model‐model comparisons difficult.
What problem does this paper attempt to address?