The thermal conductivity of seasonal snow

Abstract Twenty-seven studies on the thermal conductivity of snow ( K eff ) have been published since 1886. Combined, they comprise 354 values of K eff , and have been used to derive over 13 regression equation and predicting K eff vs density. Due to large (and largely undocumented) differences in m...

Full description

Bibliographic Details
Published in:Journal of Glaciology
Main Authors: Sturm, Matthew, Holmgren, Jon, König, Max, Morris, Kim
Format: Article in Journal/Newspaper
Language:English
Published: Cambridge University Press (CUP) 1997
Subjects:
Online Access:http://dx.doi.org/10.1017/s0022143000002781
https://www.cambridge.org/core/services/aop-cambridge-core/content/view/S0022143000002781
Description
Summary:Abstract Twenty-seven studies on the thermal conductivity of snow ( K eff ) have been published since 1886. Combined, they comprise 354 values of K eff , and have been used to derive over 13 regression equation and predicting K eff vs density. Due to large (and largely undocumented) differences in measurement methods and accuracy, sample temperature and snow type, it is not possible to know what part of the variability in this data set is the result of snow microstructure. We present a new data set containing 488 measurements for which the temperature, type and measurement accuracy are known. A quadratic equation, where ρ is in g cm −3 , and K eff is in W m −1 K −1 , can be fit to the new data ( R 2 = 0.79). A logarithmic expression, can also be used. The first regression is better when estimating values beyond the limits of the data; the second when estimating values for low-density snow. Within the data set, snow types resulting from kinetic growth show density-independent behavior. Rounded-grain and wind-blown snow show strong density dependence. The new data set has a higher mean value of density but a lower mean value of thermal conductivity than the old set. This shift is attributed to differences in snow types and sample temperatures in the sets. Using both data sets, we show that there are well-defined limits to the geometric configurations that natural seasonal snow can take.