Multivariate controls on the incidence of glacier surging in the Karakoram Himalaya

Surge-type glaciers experience cyclic flow instabilities characterized by alternating periods of slow and fast flow. The geographical distribution of surge-type glaciers has been shown to be distinctly non-random in that they are clustered in some regions yet are completely absent from others. In order to identify factors that influence glacier surging, a number of environmental and glacial attributes were examined for an area in the Karakoram Himalaya, Central Asia. A new GIS-based glacier inventory was produced using a combination of ASTER and Landsat remotely sensed images and paper maps. A total of 150 glaciers were digitized, with 19 of these (12.6%) being classified as of surge-type. Attribute data for 10 glacial and environmental attributes were recorded either during the digitization phase or extracted automatically from the GIS. Simple data visualization techniques revealed a positive correlation between glacier surging and glacier length, area, perimeter size, average width, debris cover, and orientation. The use of univariate logit regression analysis showed that length, area, perimeter, average width, and the heaviest debris cover class showed significant correlation with surging. Multivariate logit regression techniques were employed to show that length, area, average width, and debris cover were all multicollinear, with the strongest statistically modeled relationship using the variable perimeter size. The significance of glacier perimeter on surging may be explained by an increased availability of avalanche-fed snow and debris material which may act as a mass balance proxy. The findings that glacier size (in particular length and perimeter) is most strongly related to surging are consistent with the findings of studies in a number of different regions.