Education and Outreach  

Award-winning Science Fair Project:
Manipulating Snow Depth and Ice Thickness

We know that snow is a good insulator. This is exemplified by the role that snow plays in affecting the growth rate and thickness of ice on lakes and the ocean.

The graph (right) shows ice thickness as a function of depth of snow along sampling transects on two lakes near Barrow, NW Alaska, in April 1997. There is some scatter in the data, mainly because of wind redistribution of the snow and perhaps variable snow density values, but one can clearly see that the ice thickness decreases as the snow depth increases.

The deeper snow offers more insulation. Consequently, the ice growth rates are lower and the ice does not grow as thick as those locations with less snow. The thinner snow offers less insulation. Consequently, the ice growth rates are higher and the ice thickness is greater than those locations with more snow. Another, more recent, example of the effect of snow depth variability on lake ice thickness was observed at the ALISON observatory on Lucille Lake, Wasilla, in late January 2003.

Snow manipulation at three gauges
on Aurora Pond.

In Fairbanks, we enjoy much less windy weather than Barrow and Wasilla. Consequently, the snow on the lakes and ponds remains fairly uniform in depth and density because it is not redistributed by the wind. If we want to demonstrate the effectiveness of snow as an insulator through its role in determining ice growth rates and thickness, we must manipulate the snow and ice.

This is what has been done at Aurora Pond, Fairbanks, since November 2002 when the ALISON observatory went into operation. We installed three hot-wire ice thickness gauges at Aurora Pond in order to investigate the effect of the snow on ice thickness. As the figure to the left shows, the snow is left undisturbed as it accumulates naturally at Gauge#1. At Gauge#2, the snow is cleared away each time the observatory is visited (once per week) so that the ice is bare for as much time as possible. At Gauge#3, the snow depth is artificially increased by adding the snow that has been cleared from Gauge#2.

When this experiment was initiated, Martin was unaware that there would be a home-school science fair in February in Fairbanks. He was impressed when Bonnie and Kate, two of the home-school students who run the Aurora Pond observatory, chose the experiment as the subject of their science fair project. Their hypothesis was that [1] the ice at Gauge#2 would be the most thick, [2] the ice at Gauge#3 would be the least thick, and [3] the ice thickness at Gauge#1 would fall somewhere between the values at gauges 2 and 3.

The next graph shows the results obtained for the period 25 November 2002 to 10 February 2003. Although there is some ambiguity in the early and later data (for which an explanation is being sought), the results indicate that the ice did not behave entirely as hypothesized, particularly between late December and late January.

The ice at Gauge#2, where the snow was cleared, is thicker than at the other two gauges. So hypothesis [1] proved to be correct. But look at the ice thickness at Gauge#3, where the snow depth was artificially increased. The ice is thicker than that at Gauge#1 (where the snow was left undisturbed to accumulate naturally) but not as thick as that at Gauge #2. So, hypotheses [2] and [3] proved to be incorrect. The ice thickness at Gauge#3 is not the least thick, and the ice thickness at Gauge#1 does not fall between the values for gauges 2 and 3.

The results of this manipulation experiment beg the question: why did the ice at Gauge#3, where the snow was deepest and thus apparently had the most insulation, grow thicker than expected?

In addition to measuring ice thickness and snow depth each week at each gauge, the snow density was measured at Gauge#3 beginning 13 January 2003. The next graph shows that the snow density at Gauge#3 was significantly higher than the undisturbed snow that accumulated naturally. This lead to the formulation of another hypothesis. That is, because the snow is more dense at Gauge#3, and thus has a lower air content, it provides less insulation than the undisturbed snow at Gauge#1. Consequently, the ice is thicker at Gauge#3 than at Gauge #1.

Bonnie and Kate received a First Place award at the Home-school Science Fair on Saturday 8 February in Fairbanks for their presentation on the results of manipulating the snow depth and thus ice thickness at Aurora Pond, and for suggesting that the higher snow density at Gauge#3 plays a role in the unexpected ice thickness results for that gauge. Congratulations Bonnie and Kate!

The plan now is to continue the snow and ice manipulation experiment: to measure the snow depth and density, and to add measurements of temperature at the top and bottom of the snow cover, at Gauge#3. This will allow the conductive heat flow through the snow cover to be calculated and compared to that at Gauge#1.

Can you formulate a hypothesis about the relative magnitudes of the heat flow at gauges 1 and 3 that would account for the ice thickness differences? That is, does the heat flow at Gauge#3 have to be higher or lower than the heat flow at Gauge#1?