The PMIP3 Simulated Climate Changes over Arid Central Asia during the Mid-Holocene and Last Glacial Maximum

摘要

In this study, the climate changes over Arid Central Asia(ACA) during the mid-Holocene(approximately 6,000 calendar years ago, MH) and the Last Glacial Maximum(approximately 21,000 calendar years ago, LGM) were investigated using multimodel simulations derived from the Paleoclimate Modelling Intercomparison Project Phase 3(PMIP3). During the MH, the multimodel median(MMM) shows that in the core region of ACA, the regionally averaged annual surface air temperature(SAT) decreases by 0.13°C and annual precipitation decreases by 3.45%, compared with the preindustrial(PI) climate. The MMM of the SAT increases by 1.67/0.13°C in summer/autumn, whereas it decreases by 1.23/1.11°C in spring/winter. The amplitude of the seasonal cycles of the SAT increases over ACA due to different MH orbital parameters. For precipitation, the regionally averaged MMM decreases by 5.77%/5.69%/0.39%/5.24% in spring/summer/autumn/winter, respectively. Based on the analysis of the aridity index(AI), compared with the PI, a drier climate appears in southern Central Asia and western Xinjiang due to decreasing precipitation. During the LGM, the MMM shows that the regionally averaged SAT decreases by 5.04/4.36/4.70/5.12/5.88°C and precipitation decreases by 27.78%/28.16%/31.56%/27.74%/23.29% annually and in the spring, summer, autumn, and winter, respectively. Robust drying occurs throughout almost the whole core area. Decreasing precipitation plays a dominant role in shaping the drier conditions, whereas strong cooling plays a secondary but opposite role. In response to the LGM external forcings, over Central Asia and Xinjiang, the seasonal cycle of precipitation has a smaller amplitude compared with that under the PI climate. In the model-data comparison, the simulated MH moisture changes over ACA are to some extent consistent with the reconstructions, further confirming that drier conditions occurred during that period than during the PI.