The hundredfold speedup in glacier motion in a surge of the kind the kind that took place in Variegated Glacier in 1982-1983 is caused by the buildup of high water pressure in the basal passageway system, which is made possible by a fundamental and pervasive change in the geometry and water-transport characteristics of this system. The behavior of the glacier in surge has many remarkable features, which can provide clues to a detailed theory of the surging process. The surge mechanism is akin to a proposed mechanism of overthrust faulting.

The quasi-periodic oscillations between normal and fast motion exhibited by surge-type glaciers provide the best observational oppommity to determine limiting conditions that allow fast motion. The measurements from Variegated Glacier prove that its surge motion is caused by rapid sliding induced by high water pressure. This arises from a major restructuring of the basal hydraulic system, which impedes water discharge prior to and during surge. Although the evolving glacier geometry and stress distribution play a principal enabling role, the seasonal timing of two distinct surge pulses, each initiated in winter and terminated in summer, indicates a major influence from variable external water inputs. This influence is not considered in existing surge models and should promote caution in the use of data from temperate and subpolar surge-type glaciers to deduce surge potential in polar ice masses. The spatial spreading of surge motion from a zone of local initiation occurs by stress redistribution, which may spread the surging zone rapidly upglacier or downglacier inside a region of active ice, and by mass redistribution with compressional thickening at the surge front, which enables down-glacier propagation into less active ice. The data from other surge-type glaciers, including the extensive data from Medvezhiy Glacier, are not inconsistent with the above processes but are inadequate to establish whether completely different mechanisms operate in some surges. The way by which water accumulates and produces fast sliding is not established in detail for any surge-type glacier and may be different on different glaciers depending, for example, on the presence or absence of unconsolidated debris between the ice and rock. 1. INTRODUCTION multiyear, periodic, pulselike increases of speed that are not Meier and Post [1969] asked the question, "What are glacier large enough to produce the large ice displacements usually associated with glacier surges and therefore appear to be intersurges?" Their answer has provided the definition of surge behavior. Briefly stated, it is a behavior characterized by a mul-mediate between surge-type glaciers and normal glaciers [Mayo, flyear, quasi-periodic oscillation between extended periods of 1978]. Many normally flowing glaciers, including surge-type normal motion and brief periods of comparatively fast motion. glaciers in their quiescent phase, show seasonal variation of Glaciers showing this behavior, here called "surge-type" gla-velocity [Hodge, 1974; Aellen and Iken, 1979]. At yet shorter ciers, have been identified in various mountain ranges of the time scales, complex variations with time have been found world [Post, 1969; Dolgushin and Osipova, 1975] and sections [Iken, 1978; Iken et al., 1983]. Some of these variations, termed of ice caps [Thorarinsson, 1969; LiestOl, 1969]. They represent minisurges, occur as short (--1 day long) pulses of increased only a small percentage of all glaciers and are highly concen-speed that recur repeatedly at multiday intervals; thus they have t...

ABSTRACT. Numerical calculations by finite elements show that the variation of horizontal velocity with depth in the vicinity of a symmetric, isothermal, non-slipping ice ridge deforming on a flat bed is approximately consistent with prediction from laminar flow theory except in a zone within about four ice thicknesses of the divide. Within this near-divide zone horizontal shear strain-rate is less concentrated near the bottom and downward velocity is less rapid in comparison to the flanks. The profiles over depth of horizontal and vertical velocity approach being linear and parabolic respectively. Calculations for various surface elevation profiles show these velocity profile shapes are insensitive to the ice-sheet geometry. RESUME. Deformation au voisinage des difJluences glaciaires. Des calculs numeriques aux elements finis montrent que la variation de la vitesse horizontale avec le profondeur au voisinage d'une diflluence de glace symetrique isotherme et sans glissement sur un lit plat est a peu pres coherente avec les previsions de la theorie de I'eco ulement laminaire sauf dans une zone i:loignee de la diffluence de moins de quatre fois I'epaisseur de la glace. A I'interieur de ceUe zone la deformation visqueuse horizontale est moins concentree vers le fond et la vitesse vers le bas est moin s rapide que vers les rives. Les profils selon la profondeur des vitesses horizontales et verticales sont approximativement I'une lineaire, I'autre parabolique. Les calculs pour differents profils d' altitude superficielle montrent que les formes des profils de vitesse sont independants de la forme geometrique de I'appareil glaciaire.

The length of time T M over which a glacier responds to a prior change in climate is investigated with reference to the linearized theory of kinematic waves and to results from numerical models. We show the following: T M may in general be estimated by a volume time-scale describing the time required for a step change in mass balance to supply the volume difference between the initial and final steady states. The factor 1 in the classical estimate of T M = Ill u, where I is glacier length and u is terminus velocity, has a simple geometrical interpretation . It is the ratio of thickness change averaged over the full length I to the change at the terminus. Although both u and 1 relate to dynamic processes local to the terminus zone, the ratio I l u and, therefore , T M are insensitive to details of the terminus dynamics , in contrast to conclusions derived from some simplified kinematic wave models. A more robust estimate of T M independent of terminus dynamics is given by T M = hl (-b) where h is a thickness scale for the glacier and -b is the mass-balance rate (negative) at the terminus.We suggest that T M for mountain glaciers can be substantially less than the 10 2 -10 3 years commonly considered to be theoretically expected. ,/ /

ABSTRACT. The length of time T M over which a glacier responds to a prior change in climate is investigated with reference to the linearized theory of kinematic waves and to results from numerical models. We show the following: T M may in general be estimated by a volume time-scale describing the time required for a step change in mass balance to supply the volume difference between the initial and final steady states. The factor 1 in the classical estimate of T M = Ill u, where I is glacier length and u is terminus velocity, has a simple geometrical interpretation . It is the ratio of thickness change averaged over the full length I to the change at the terminus. Although both u and 1 relate to dynamic processes local to the terminus zone, the ratio I l u and, therefore , T M are insensitive to details of the terminus dynamics , in contrast to conclusions derived from some simplified kinematic wave models. A more robust estimate of T M independent of terminus dynamics is given by T M = hl (-b) where h is a thickness scale for the glacier and -b is the mass-balance rate (negative) at the terminus. We suggest that T M for mountain glaciers can be substantially less than the 10 2 -10 3 years commonly considered to be theoretically expected.