Responding effectively to the impacts of climate change will rely increasingly on predictive ecological models. A major challenge of using spatially-explicit models is spatial mismatch: the area represented by each grid square is typically much larger than both the scale at which microclimate conditions vary and at which most organisms experience their environments. This ongoing project is combining emerging methods in remote sensing with mechanistic modeling to map the 3D microclimate landscape and quantify how spatial resolution affects our ability to detect microclimate-scale variation. We are using terrain layers derived from unmanned aerial systems (UAS) to drive a series of mechanistic microclimate models, predicting hourly, above and below-ground temperatures at extremely fine spatial resolutions, from about 6.25-625 cm2. Then, we are comparing predictions with those generated using coarser-resolution (10m2-1km2), freely downloadable terrain data to determine how spatial resolution affects the accuracy, strength, and direction of model predictions.
Results/Conclusions
Increasing the spatial resolution of our microclimate model increased our ability to detect heterogeneity in thermal conditions and, to a point, model accuracy. Most spatially-explicit ecological models, such as species distribution models, are driven by relatively coarse-resolution data. However, our results so far suggest that lower-resolution models can mask variation in key microclimate components that determine the physiological outcomes of organisms’ interactions with their environments. Coarse-resolution models may be unable to detect microrefugia, the patches of microclimatic space in which species can persist during periods of otherwise unsuitable climatic conditions, and microclimatic range edges, the dynamic boundaries of microclimatic space that could indicate the potential for range shifts or act as barriers to dispersal. Although coarse-resolution spatial data are free and widely available, they may be insufficient for building realistic models and should be used with caution. Collection of spatial data that more realistically represent the microclimate landscape with which organisms interact should be a high research priority.