Estimating uncertainty in daily weather interpolations: a Bayesian framework for developing climate surfaces.

Conservation of biodiversity demands comprehension of evolutionary and ecological patterns and processes that occur over vast spatial and temporal scales. A central goal of ecology is to understand the climatic factors that control ecological processes and this has become even more important in the face of climate change. Especially at global scales, there can be enormous uncertainty in underlying environmental data used to explain ecological processes, but that uncertainty is rarely quantified or incorporated into ecological models. In this study, a climate-aided Bayesian kriging approach is used to interpolate 20 years of daily meteorological observations (maximum and minimum temperatures and precipitation) to a 1 arc-min grid for the Cape Floristic Region of South Africa. Independent validation data revealed overall predictive performance of the interpolation to have R² values of 0.90, 0.85, and 0.59 for maximum temperature, minimum temperature, and precipitation, respectively. A suite of ecologically relevant climate metrics that include the uncertainty introduced by the interpolation were then generated. By providing the high-resolution climate metric surfaces and uncertainties, this work facilitates richer and more robust predictive modelling in ecology and biogeography. These data can be incorporated into ecological models to propagate the uncertainties through to the final predictions. [ABSTRACT FROM AUTHOR]