Predicting glacier accumulation area distributions
A mass balance model based on energy balance at the terrain surface was developed and used to predict glacier accumulation areas in the Jotunheimen, Norway. Spatially distributed melt modelling used local climate and energy balance surfaces to drive predictions, derived from regional climate and topographic data. Predictions had a temporal resolution of 1 month and a spatial resolution of 100 m, which were able to simulate observed glacier accumulation area distributions. Data were stored and manipulated within a GIS and spatial trends and patterns within the data were explored. These trends guided the design of a suite of geomorphologically and climatologically significant variables which were used to simulate the observed spatial organisation of climatic variables, specifically temperature, precipitation and wind speed and direction. DEM quality was found as a critical factor in minimising error propagation. A new method of removing spatially and spectrally organised DEM error is presented using a fast Fourier transformation. This was successfully employed to remove error within the DEM minimising error propagation into model predictions. With no parameter fitting the modeled spatial distribution of snowcover showed good agreement with observed distributions. Topographic maps and a Landsat ETM+ image are used to validate the predictions and identify areas of over or under prediction. Topographically constrained glaciers are most effectively simulated, where aspect, gradient and altitude impose dominant controls on accumulation. Reflections on the causes of over or under prediction are presented and future research directions to address these are outlined. Sensitivity of snow accumulation to climatic and radiative variables was assessed. Results showed the mass balance of accumulation areas is most sensitive to air temperature and cloud cover parameterisations. The model was applied to reconstruct snow accumulation at the last glacial maximum and under IPCC warming scenarios to assess the sensitivity of melt to changing environmental conditions, which showed pronounced sensitivity to summer temperatures Low data requirements: regional climate and elevation data identify the model as a powerful tool for predicting the onset, duration and rate of melt for any geographical area.