In this thesis, a methodology using general circulation models (GCMs) for the quantification of natural climate variability as recorded in paleoclimatic proxy data is proposed. A systematic comparison between paleoclimatic proxy data and GCM integrations is made possible by developing a method for a direct and process-based simulation of proxy records using statistically downscaled GCM output. This means that "synthetic" records are generated from GCM experiments for comparison with actual in situ proxy data. The approach is applicable to various paleoclimatic proxy indicators and is demonstrated for the simulation of valley glaciers, using process-oriented models for glacier mass balance and glacier length. Simulated glacier length changes are compared to observed or reconstructed glacier fluctuations, which represent proxy data for low frequency climate variations extending back over several centuries. The role of internal variability in the climate system as a major source of glacier fluctuations is investigated.
A reliable statistical downscaling approach for GCM output is developed based on large-scale predictors obtained from daily reanalyses of the European Centre for Medium-Range Weather Forecasts (ECMWF) and daily operational weather station data in the vicinity of the proxy site to be investigated. The method produces a good simulation of local temperature and precipitation for weather stations in Norway and the Alps. It is successfully applied to European Center/Hamburg (ECHAM4) GCM experiments.
In order to simulate fluctuations of specific valley glaciers using GCM integrations, a method is developed to relate glacier mass balance to meteorological data or to downscaled GCM output. The climate sensitivity of a glacier is quantified by a Seasonal Sensitivity Characteristic (SSC), which represents the dependence of glacier mass balance on monthly anomalies in temperature and precipitation and is calculated from a mass balance model of intermediate complexity. SSCs for a number of valley glaciers located in various parts of the world are calculated. With regard to the climate sensitivity of mass balance to temperature, the SSCs show that for glaciers in dry climates, the effect of temperature is mainly restricted to summer months, whereas for glaciers in wetter climates, spring and Autumn temperatures additionally contribute. The considerable overall sensitivity to precipitation varies significantly between the glaciers, the effect of summer precipitation is also strongly dependant on the altitudinal range of the glaciers. Mass balance simulations using ECMWF reanalyses for the period 1979-1993 are in good agreement with in situ measurements for Nigardsbreen glacier (Norway).
The downscaling methodology and the SSCs described above are used to simulate mass balance fluctuations for Nigardsbreen and Rhonegletscher (Switzerland) using multi-century coupled (ECHAM4/OPYC) and mixed-layer (ECHAM4/MLO) GCM integrations. External forcing is excluded in the GCM experiments. For both glaciers, a high correlation is found between decadal variations in the North Atlantic Oscillation (NAO) and glacier mass balance. The dominant factor for this relationship is the strong impact of winter precipitation associated with the NAO. For Nigardsbreen (Rhonegletscher), a high NAO phase means enhanced (reduced) winter precipitation, typically leading to a higher (lower) than normal annual mass balance. This mechanism, entirely due to internal variations in the climate system, can also explain observed strong positive mass balances for Nigardsbreen and possibly other maritime Norwegian glaciers within the period 1980-1995.
A dynamic ice flow model considering simple shearing flow and sliding is applied, in order to simulate glacier length changes due to internal variations in the climate system using the statistically downscaled GCM integrations described above. 10000 year records of glacier length fluctuations are generated using auto-regressive processes determined by the GCM experiments. Return periods and probabilities of specific glacier length changes, simulated excluding external forcings such as solar irradiation changes, volcanic or anthropogenic effects, are calculated and compared to historical glacier length records.
It is concluded that preindustrial fluctuations of the glaciers as far as observed or reconstructed, including their advance during the "Little Ice Age", can be explained entirely by internal variations in the climate system as represented by a GCM. External forcing is not required. The probability that these glacier fluctuations occur within a time period of 10000 years is 99.7% for Nigardsbreen and 80.4% for Rhonegletscher. However, fluctuations comparable to the observed present-day retreat of the glaciers (4 km for Nigardsbreen and 2.3 km for Rhonegletscher since the "Little Ice Age" maximum) do not occur in these GCM experiments, it is therefore very unlikely that this retreat is entirely due to internal climate variations. The probability of occurrence for such a retreat is 11.1% for Nigardsbreen and 9.9% for Rhonegletscher within a time period of 10000 years. The results imply that the present-day retreat is in all likelihood due to external forcing, greenhouse gas forcing being the most probable potential candidate.