Holocene glacier variability: three case studies using an intermediate-complexity climate model

Synthetic glacier length records are generated for the Holocene epoch using a process-based glacier model coupled to the intermediate-complexity climate model ECBilt. The glacier model consists of a mass-balance component and an ice-flow component. The climate model is forced by the insolation change due to variations in the Earth's orbital parameters. We consider three glaciers, ranging from maritime to continental. At Nigardsbreen (southern Norway), the simulated long-term trend in the annual mass-balance is primarily determined by summer temperature, with a smaller contribution from winter precipitation. In the early Holocene, summers were warmer and winters wetter than today in the ECBilt simulation. Both signals seem consistent with proxy data. The simulated glacier length shows a phase of rapid expansion during the mid-Holocene, followed by more gradual growth. At Rhonegletscher (the Swiss Alps), ECBilt simulates warmer and wetter summers in the early Holocene. The temperature signal seems realistic, but proxy data and earlier modelling results are not conclusive with respect to the precipitation signal. The implied glacier length shows a maximum extent at 3-5 kyr BP, which seems unlikely. This suggests that the simulated precipitation response is not realistic. The simulated early-Holocene climate at Abramov glacier (Kirghizia) is characterized by high summer precipitation, associated with a northward extension of the Asian monsoon, The precipitation signal reaches its maximum around 6 kyr BP, which is consistent with the timing of the maximum in lake-level data, The simulated glacier length shows a pronounced postglacial maximum at the time of maximum monsoon intensity. There is considerable centennial-timescale variability in the simulated glacier length records. These length variations are generated by internal climatic variability. They are typically asynchronous among the three different glaciers. ECBilt has reasonable skill in simulating the relative importance of temperature and precipitation as well as the seasonality of the forcing, although the overall level of variability is underestimated. Length variations are shown to behave as a tagged moving-average process, with a glacier-specific memory.