Downscaling climate in complex terrain with temperature sensor networks: deployment and data analysis challenges

Abstract:

The spatial complexities of mountain climate provide ecological resiliency in the face of rapid climate change. Species may not have to move far to track macroclimatic change, as the range of topoclimates over hundreds of meters distance often exceed the degree of macroclimatic changes. Two key topoclimatic gradients - insolation differences producing variability in maximum temperatures, and cold-air pooling producing variability in minimum temperatures – are readily measured using stratified deployments of small temperature sensors. By holding the microclimatic conditions – radiation shelters, shade, and height above ground – as constant as possible, the temperature differentials across terrain features can be quantified. Those temperature differentials can be projected across complex terrain in GIS using multiple regression and machine-learning models. Weather conditions measured at a base station provide a temporal stratification, so that radiation differences are muted under cloudy skies, and cold air pools are disrupted by wind, clouds, and precipitation, and these can be modeled on daily or even hourly time-steps using machine learning.

These analyses are illustrated with data acquired in a coastal California mountain range for frost frequency analysis, and at treeline sites in the Great Basin for dendrochronology research. Three major conclusions are: 1) spatial variability at the topoclimate scale can easily exceed climate change projections, so the velocity of climate change is actually quite slow in complex terrain; and 2) local maximum and minimum temperatures are weakly correlated if at all; and 3) use of a base station to incorporate weather variability can enhance the spatial projections. The link between macroclimatic and topoclimatic conditions produces yet another component of variability driven by climate change.