KENNICOTT GLACIER OUTBURST FLOODS: REAL TIME OBSERVATIONS AND SUBGLACIAL HYDROLOGIC MODELING
ANDERSON, SUZANNE P.. University of California, Santa Cruz.
Walder, Joseph S.. U.S. Geological Survey.
Anderson, Robert S.. University of California, Santa Cruz.
Kraal, Erin R.. University of California, Santa Cruz.
Fountain, Andrew G.. Portland State University.
Cunico, Michele . Portland State University.
Trabant, Dennis . U.S. Geological Survey.
Glacial outburst floods, which occur when bodies of water impounded by glaciers drain suddenly, produce catastrophic floods that can reshape river channels and pose a significant hazard downstream. Our understanding of the triggers for these floods is severely hampered by inaccessibility and relatively unpredictable timing. We have taken advantage of an ice-dammed lake that has produced annual outburst floods for the last 100 years at the Kennicott Glacier in south-central Alaska to monitor outburst floods in two consecutive years.
Hidden Creek Lake (HCL) forms each summer in a deglaciated tributary to the Kennicott Glacier, located in the Wrangell Mountains of south-central Alaska (Figure 1). Outburst floods from Hidden Creek Lake have been observed for nearly a century (Rickman and Rosenkrans, 1997). The date of the annual outburst has become progressively earlier in summer over this time, but there is considerable interannual scatter. This scatter does not reflect simple differences in the rate at which the lake reaches some critical level; the maximum lake level differed by 9 m between 1999 and 2000. Assuming that peak discharges are related to lake volume (Clague and Mathews, 1973), differences in peak flood stage measured in the Kennicott River over a number of years also suggest that the lake drains at different levels each year.
Our observations during the melt seasons of 1999 and 2000 included lake level history, lake basin hypsometry, ice dam deformation, water pressure in boreholes in the ice dam, and discharge, suspended sediment concentrations and water chemistry in the Kennicott River. In 2000, we also monitored flow in Hidden Creek (the major input into Hidden Creek Lake), and the filling and drainage history of Donoho Falls Lake, an ice-marginal basin down glacier from Hidden Creek Lake. Our stream gauging of the Kennicott River is augmented by daily stage measurements in the summer over the last five years, and we have acquired daily meteorological observations at cooperative station at McCarthy from the National Weather Service.
In 1999 and 2000, the HCL outburst occurred at the termination of a period of high discharge in the Kennicott River, and was followed within a few days by a period of very low discharge. Examination of Kennicott River stage records from 1997-2001 show that this pattern is typical. Long-period (10-20 day) oscillations of the river stage are evident in most summers. In 1999 and 2000, the same oscillations are also observed in our measurements of the electrical conductivity (EC) in the river, but the EC variations lag stage variations by several days. We infer that the discharge and chemistry both reflect variations in meltwater inputs, but with a lag imposed by the time required to expand or close subglacial conduits in response to these changing inputs (Anderson et al., submitted). As a result, the amount of water stored in the glacier oscillates throughout the melt season.
Lake drainage in 2000 was complete in 2.5 days. Drainage in 1999 appears to have followed the same pattern, although since lake drainage was underway at the beginning of our observations we cannot be certain about the onset. In both years, Kennicott River discharge began to rise 1-1.5 days after HCL drainage began. In 2000, our observations in Donoho Falls Lake show that this previously emptied basin began to fill at about the same time that the Kennicott River began to rise. Peak suspended sediment concentrations in the Kennicott River in both years occurred about 12 hours before peak water discharge. Donoho Falls Lake rapidly reached a steady high level in 2000, and drained equally rapidly a few hours after the peak flow in the Kennicott River. The flood hydrographs in both years were more symmetrical than is generally associated with glacial outburst floods (Tweed and Russell, 1999). The integrated flood discharge was within error equivalent to the volume of the lake in each year.
Passage of the flood wave through the glacier disrupts "normal" subglacial drainage. The chemistry of the water that filled Donoho Falls Lake basin differed from HCL water, indicating that Donoho Falls Lake filled with subglacial water, perhaps from the Root Glacier (Fig. 1), that is temporarily backed-up during the passage of the flood wave. This interruption to subglacial flow occurred only after the discharge in the Kennicott River began to exceed the maximum it had reached earlier in the summer. That Donoho Falls Lake maintained a steady high level for about a day suggests that subglacial conduits grew at a rate sufficient to accommodate both increasing discharge from HCL and background melt water from the remainder of the glacier. After HCL drainage was completed, the conduit system, at least from the terminus up to vicinity of the lake, is presumably larger than is necessary for transmission of melt water production. This promotes drainage from the distributed flow system. Discharge from the glacier falls to a summer low at this time.
We have constructed a lumped-element model of the subglacial hydrologic system (Clarke, 1996). The quasi-two dimensional model allows for flow between a series of nodes using equations that represent expansion and contraction of Rothlisberger conduits. Lateral flow into the nodes is also allowed using physics that mimic flow within the linked-cavity distributed flow system. We initiate the model with small (cm scale) conduits; mass balance on melt inputs and flow out of each node, combined with an assumed glacier porosity, determine the head at each node. Conduits grow up from the terminus in our model runs; once established, the conduits provide a low-head gradient fast flow system, fed by the high-head distributed flow system. The entire system responds to variations in melt water inputs.
A key hypothesis is that hydrologic head within the basal drainage system near the lake may control drainage of the ice-dammed lake. We hypothesize that catastrophic drainage occurs when the head gradient between the lake and subglacial drainage system near the lake is sufficient to open a Rothlisberger conduit, and that this condition is met only when head in the distributed flow system near the lake falls below the head in the lake. We are beginning to explore these hypotheses with the lumped-element model of subglacial hydrology.
Anderson, S.P., Longacre, S.A., and Kraal, E.R. (submitted): Patterns of water chemistry and discharge in the glacier-fed Kennicott River, Alaska: Evidence for subglacial water storage cycles, Chemical Geology.
Clague, J.J., and Mathews, W.H. (1973): The magnitude of jokulhlaups, Journal of Glaciology 12 (6): 501-504.
Clarke, G.K.C. (1996): Lumped-element analysis of subglacial hydraulic circuits, Journal of Geophysical Research 101 (B8): 17,547-17,559.
Rickman, R.L. and Rosenkrans, D.S. (1997): Hydrologic conditions and hazards in the Kennicott River basin, Wrangell-St. Elias National Park and Preserve, Alaska, U.S. Geological Survey Water-Resources Investigations Report 96-4296.
Tweed, F. S., and Russell, A.J. (1999): Controls on the formation and sudden drainage of glacier-impounded lakes: implications for jokulhlaup characteristics, Progress in Physical Geography 23 (1): 79-110.
Figure 1. Map of ice-dammed Hidden Creek Lake and terminus region of the Kennicott Glacier. Approximate maximum lake areas in 1959 and in 2000 shown. Elevations in feet.
Previous | Abstract Index | Next
Copyright © 2001 INSTAAR, Univ. of Colorado