Modelling near-surface temperature conditions in high mountain environments: an appraisal

In response to the paucity of temperature data for high mountain environments at spatial and temporal resolutions reasonable for ecological modelling, this study evaluated both physically and statistically based modelling (or interpolation) of near-surface temperatures in a high mountain environment at sub-daily time scales and differing spatial scales. We placed emphasis on the selection of guiding environmental variables for temperature interpolation at a macro-scale, on a comparison of physically and statistically based modelling at meso- and micro-scale, and on a discussion of scaling issues. Geostatistical interpolation was found to perform very well at a macro-scale if additional topographic and atmospheric covariables-the latter provided by remote sensing-are considered. Physically based modelling performed best at a nano-scale, but revealed limits of spatial applicability due to limited input data. Statistical modelling based on multiple linear regression, however, showed results of intermediate accuracy throughout the spatial scales, providing encouraging evidence that we can find a simple approach to estimate near-surface temperature fields in high mountain environments. Attention to scaling issues proved to be important to achieve accurate results, though this was hampered by the starting point of the study: the lack of observational data at finer spatial scales.