Hydrometeorological Conditions, Mass Balance and Runoff from Alpine Glaciers

Abstract

Discharge records for rivers draining from highly-glacierized basins in the upper Rhône catchment area, Switzerland, have been examined together with precipitation and air temperature data with a view to determining which variables from which stations can be used to represent interacting climatic elements that influence mass balance of and runoff from Alpine glaciers. Relationships between climate and mass balance and between climate and runoff were analysed at seasonal and annual timescales by correlation and multiple linear regression techniques. Variables were selected from summer air temperatures, fluctuations of which are similar over the catchment area, and precipitation records from valley stations with the intention of providing best levels of explanation of variances of runoff and mass balance. An attempt to estimate the relationship between precipitation and elevation was made using the catch of totalizing raingauges emptied at roughly half-yearly intervals, for which reasonable levels of correlation exist between valley and mountain gauges. Summer air temperatures averaged over the months May through September and June through August are highly positively correlated with summer totals of runoff but strongly negatively associated with mass balance. Relationships between precipitation, best expressed by totals from October through May, May through September and May alone, with mass balance are more strongly positive than are those with runoff negative. Between 75 and 91 per cent of the variances of total summer discharge (May through September) and of net mass balance are explained by multiple regression against mean summer air temperatures and either one or two precipitation variables. Significant fluctuations of climatic conditions, glacier mass balance and runoff occurred in the period 1922–1985. Generally-declining air temperatures led to runoff reducing substantially between the 1940s and the 1970s, at the same time as glacierized areas diminished. Multiple regression models, whilst exhibiting high degrees of fit, appear to have limited predictive power on account of non-stationarity of relationships between variables used.