Bayesian Reconstruction of Past Land Cover From Pollen Data: Model Robustness and Sensitivity to Auxiliary Variables

• Main Authors: Behnaz Pirzamanbein, Anneli Poska, Johan Lindström
• Format: Article
• Language: English
• Published: American Geophysical Union (AGU) 2020-01-01
• Series: Earth and Space Science
• Online Access: https://doi.org/10.1029/2018EA000547

Abstract Realistic depictions of past land cover are needed to investigate prehistoric environmental changes, effects of anthropogenic deforestation, and long‐term land cover‐climate feedbacks. Observation‐based reconstructions of past land cover are rare, and commonly used model‐based reconstructions exhibit considerable differences. Recently, Pirzamanbein et al. (2018, 10.1016/j.spasta.2018.03.005, Spatial Statistics, 24:14–31) developed a statistical interpolation method that produces spatially complete reconstructions of past land cover from pollen assemblage. These reconstructions incorporate a number of auxiliary data sets raising questions regarding the method's sensitivity to different auxiliary data sets. Here the sensitivity of the method is examined by performing spatial reconstructions for northern Europe during three time periods (1900 CE, 1725 CE, and 4000 BCE). The auxiliary data sets considered include the most commonly utilized sources of past land cover data—for example, estimates produced by a dynamic vegetation model and anthropogenic land cover change models. Five different auxiliary data sets were considered, including different climate data driving the dynamic vegetation model and different anthropogenic land cover change models. The resulting reconstructions were evaluated using cross validation for all the time periods. For the recent time period, 1900 CE, the different land cover reconstructions were also compared against a present day forest map. The validation confirms that the statistical model provides a robust spatial interpolation tool with low sensitivity to differences in auxiliary data and high capacity to capture information in the pollen‐based proxy data. Further auxiliary data with high spatial detail improves model performance for areas with complex topography or few observations.