PP23A-2285
Moisture Changes in the Mid-Holocene: Do PMIP3 Simulations Reflect the Proxy Record for the Western United States?
Tuesday, 15 December 2015
Poster Hall (Moscone South)
Nicholas Wayne Hermann, Vanderbilt University, Nashville, TN, United States, Jessica Leigh Oster, Vanderbilt University, Earth & Environmental Sciences, Nashville, TN, United States and Daniel E Ibarra, Stanford University, Earth System Science, Stanford, CA, United States
Abstract:
Paleoclimate proxy-model comparisons for past warm periods provide insight into the driving mechanisms of persistent drought, especially pertinent information given the ongoing drought in the western US. A network of 110 hydrologically sensitive proxy records from lake sediments, speleothems, and packrat middens shows increased aridity in the western and northwestern U.S. and increased moisture in the southwest monsoon region during the mid-Holocene, a time of higher summer and lower winter insolation than present. We compare this proxy network to output from 12 climate models from the Paleoclimate Modeling Intercomparison Project Phase III (PMIP3) at the 6ka time slice to determine which models best reflect paleo-precipitation patterns, and therefore, the atmospheric drivers of precipitation change. We carry out a quantitative proxy-model comparison for precipitation (P) and effective moisture (EM = P – evapotranspiration) at 6ka relative to preindustrial simulations. Using a weighted Cohen’s Kappa test, we evaluate model agreement at different thresholds of change for annual and seasonal anomalies. Overall, simulated P anomalies show better agreement with the proxy network than EM anomalies. The FGOALS-g2 (FG) model shows the best agreement with the proxy network for annual P anomalies, with increased annual P in the monsoonal southwestern US and northern Mexico, and decreased annual P in northern California, the northern Great Basin, and the Pacific Northwest at 6 ka relative to modern. FG simulates decreased winter P across the entire western US, increased autumn P in the monsoon region and decreased autumn P in the non-monsoon region. While other models simulate increased P from July-September in the monsoon region and more arid conditions at non-monsoon sites, they often do not simulate increased aridity in non-monsoon regions during other seasons. These differences may be due to a less defined autumn Aleutian Low in FG compared with most other models.