Reconstructing winter snow accumulation since AD 1150 in the Andes (30-37°S) using tree rings, historical documents and instrumental records

MASIOKAS, M.H.; VILLALBA, R.; BETMAN, E.; CHRISTIE, D.; LE QUESNE, C.; LUCKMAN, B.H.; MAUGET, S.; PRIETO, M.R.

Valdivia

Conferencia; II International Synposium “Reconstructing Climate Variations in South America and the Antarctic Peninsula over the past 2000 years”; 2010

PAGES y Centro de Estudios Científicos

The Andean snowpack between 30°-37°S is the main water source in the adjacent Chilean and Argentinean lowlands and has affected communication and transportation between these regions since at least Spanish colonial times. Here we present a series of sequential reconstructions of a regionally-averaged snowpack record based on different proxies with varying explanatory power and temporal coverage. First, the 1951-2010 snowpack record is extended back to 1909 using a strongly correlated regional streamflow series. Instrumental values for winter precipitation in central Chile, also strongly related to snow accumulation levels in the adjacent Andes, are then used to reconstruct regional snowpack variations since 1866. Snow accumulation levels are then extended back to AD 1535 using two selected tree-ring width chronologies from drought-sensitive Austrocedrus chilensis trees plus historical and instrumental information on extreme wet years in central Chile. A very simple tree-ring based model that explains almost 50% of the variance over the 1951-2000 calibration period is finally used to reconstruct winter snowpack variations back to AD 1150. Additional information on mountain snow accumulation since the 18th century has recently been compiled from historical reports and could also be incorporated into future models to provide better estimates of past snowpack changes in this portion of the Andes. This is the first attempt to reconstruct Andean snowpack variations from different proxies with different inherent limitations (e.g. tree rings are more capable of recording below-average snowpack conditions, historical documents focus mainly on extreme wet/snowy years), and the approach appears quite promising. An innovative combination of time series analysis techniques will be used to identify the main intra- to multi-decadal patterns in the snowpack reconstructions. Such analyses will allow the objective evaluation of the variations observed in the instrumental series within a multi-century perspective.