Basal temperature of snow (BTS) data show characteristic spatial autocorrelations at distances typically less than 200 m, leading to non-independent regression residuals. Systematic temporal variations may also introduce model bias resulting in a shift in the predicted lower limit of permafrost. Both phenomena are analysed. The selection of an appropriate sampling design for a BTS measurement program appears critical in order to minimize problems typical of observational data in complex terrain. Copyright

BTS, Mountain permafrost, Sampling design, Spatiotemporal variability, Statistical analysis
dx.doi.org/10.1002/ppp.541
Permafrost and Periglacial Processes
Department of Geography and Environmental Studies

Brenning, A. (Alexander), Gruber, S, & Hoelzle, M. (Martin). (2005). Sampling and statistical analyses of BTS measurements. Permafrost and Periglacial Processes, 16(4), 383–393. doi:10.1002/ppp.541