In relatively well-studied western Central Asia (Hindu Kush, Karakorum) it has long been recognized that MIS 2 glaciers were smaller than earlier ones, although the template for global glaciations inferred from the marine oxygen isotope records deferred acceptance for many years until new cosmic-ray exposure ages shifted the balance of the argument towards asynchronism. The less-studied ranges of northern Central Asia (Tien Shan, Altai, Sayan, Khangay Nuruu, Gobi-Altai), north of any monsoonal effects, appear to be characterized by asynchronism on a fine (~250 km) scale. In the Kyrgyz Tien Shan, the most extensive preserved moraines date to MIS4/5, and in several locations the MIS 2 “LGM” moraines are confined to the cirques. Yet in the Xinjiang Tien Shan, 300 km to the east, MIS moraines were more extensive (Daxigou) or only slightly less extensive (Muzart River) than MIS 3 moraines. Farther east, in Mongolia, in the northern (Sayan) and central (Khangay) ranges the pattern is similar to Daxigou. However, in the Gobi-Altai it is unique for this area: the highest peaks were covered with ice caps that shed cirque glaciers during the Holocene (an interglaciation), whereas during MIS 2 and earlier the glaciers were smaller.

What explains this pattern of variable asynchronism? One factor seems to predominate, and that is the winter precipitation during MIS 2. Rupper at al. (2008) explored the role of sublimation as the dominant control on ELA in arid/hyperarid (<125 mm/yr) regions. In such regions, temperature plays a subordinate role to precipitation in controlling ELA. If a region undergoes a transition over time from semi- to hyper-aridity, ELAs will rise dramatically (Kyrgyz Tien Shan). If a region becomes less arid, ELAs will lower. In the Gobi, it is paradoxically possible that increased spring/fall air temperatures during the Holocene insolation maximum reduced the effect of the Siberian winter highs and increased water advection into the desert and lowering ELAs, whereas in the Pleistocene the ELAs were over the tops of the peaks. Although it is unlikely that sublimation/precipitation can explain all the spatial complexities in ELA over Central Asia, it is a model that can account for the fine scale of variability. It directs attention to locally increased MIS 2 aridity rather than increased precipitation during MIS3/4.