In this thesis, two new approaches are proposed for the inference of two distributed glaciological variables of difficult access: ice-thickness and accumulation distribution. A first method is proposed for inferring the ice-thickness distribution based on the principle of mass conservation and considerations on the ice-flow mechanics. The method is applied to a set of glaciers in the Swiss Alps and the results are used to calibrate a volume-area scaling relation. A second approach is proposed for inferring the distribution of snow accumulation by combining time-lapse photography and simple modeling. Although their simplicity, both methods perform surprisingly well when compared to direct measurements and show therefor the potential for a large number of applications.