A fully implicit finite element model of conductive heat transfer, incorporating temperature-dependent apparent thermal diffusivity and fusion effects, was employed to model the thermal field in freezing and thawing soils. Hourly temperature time series at eight probes levels in the near-surface soil were available from three sites in Alaska. Using 3-hr averages, the simulated temperature was subtracted from the observed temperature for each level to display the relative magnitude and time-dependence of nonconductive heat transport.
Two sites are from the discontinuous permafrost region of central Alaska. Located within the taiga, one site is underlain by permafrost while the other experiences only seasonal frost due to water seepage. A third site is located at Barrow, in the continuous permafrost zone.
The temporal pattern of residuals indicate that the thermal field during mid-winter soil frost, at the time of the lowest annual temperatures, most closely fits the conductive model. At the taiga seasonal frost site, nonconductive effects are prominent during initiation of soil thaw immediately following snow melt. The strongly negative residuals early in summer appear to result from coupled-flow effects associated with upward migration of cold water from the thaw front. A similar pattern of observed temperatures cooler than simulated was also observed during the early stages of active layer thaw at both taiga and tundra sites underlain by permafrost, and appears to result from the combined effects of evaporative cooling and upward migration of cold water. At these sites, the residuals became more positive as summer progressed and the active layer deepened, suggesting that nonconductive cooling processes are more effective early in summer. The magnitude of coupled-flow effects is strongly site specific, with active layer residuals during summer consistently averaging around -3C at Barrow, but ranging between -1 and 0.5 C at the taiga site.
At the permafrost sites, both the period of the zero-curtain and the snowmelt period produced strong positive residuals, again indicating the thermal effects of nonconductive heat transport processes during these regimes. The liberation of latent heat during the zero-curtain regime keeps observed temperatures warmer than those simulated, while the downward transfer of sensible heat occurs during snowmelt as moisture migrates rapidly to depth. This methodology is useful for quantifying and displaying the relative thermal magnitude of nonconductive heat transfer throughout the annual cycle at dissimilar sites.
Kenneth M. Hinkel Department of Geography, ML 131 University of Cincinnati Cincinnati, OH 45221-0131 Phone: (513) 556-3421 Fax: (513) 556-3370 E-mail: 71042.2643@compuserve.com