Glacial-interglacial shifts in global and regional precipitation δ 18 O

Abstract. Previous analyses of past climate changes have often been based on site-specific isotope records from speleothems, ice cores, sediments and groundwaters. However, in most studies these dispersed records have not been integrated and synthesized in a comprehensive manner to explore the spatial patterns of precipitation isotope changes from the last ice age to more recent times. Here we synthesize 88 globally-distributed groundwater, cave calcite, and ice core isotope records spanning the last ice age to the late-Holocene. Our data-driven review shows that reconstructed precipitation δ18O changes from the last ice age to the late-Holocene range from −7.1‰ (ice age δ18O < late-Holocene δ18O) to +1.8‰ (ice age δ18O > late-Holocene δ18O) with wide regional variability. The majority (75%) of reconstructions have lower ice age δ18O values than late-Holocene δ18O values. High-magnitude, negative glacial–interglacial precipitation δ18O shifts (ice age δ18O < late-Holocene δ18O by more than 3‰) are common at high latitudes, high altitudes and continental interiors. Conversely, lower-magnitude, positive glacial–interglacial precipitation δ18O shifts (ice age δ18O > late-Holocene δ18O by less than 2‰) are most common along subtropical coasts. Broad, global patterns of glacial–interglacial precipitation δ18O shifts are consistent with stronger-than-modern isotopic distillation of air masses during the last ice age, likely impacted by larger global temperature differences between the tropics and the poles. Further, to complement our synthesis of proxy-record precipitation δ18O, we compiled isotope enabled general circulation model simulations of recent and last glacial maximum climate states. Simulated precipitation δ18O from five general circulation models show better inter-model and model-observation agreement in the sign of δ18O changes from the last ice age to present day in temperate and polar regions than in the tropics. Further model precipitation δ18O research is needed to better understand impacts of inter-model spread in simulated precipitation fluxes and parameterizations of convective rainout, seawater δ18O and glacial topography on simulated precipitation δ18O. Future paleo-precipitation proxy record δ18O research can use new global maps of glacial δ18O reconstructions to target and prioritize regional investigations of past climate states.