Reconstruction of Past Human land-use from Pollen Data and Anthropogenic land-cover Changes Scenarios

Accurate maps of past land cover and human land-use are necessary when studying the impact of anthropogenic land-cover changes on climate. Ideally the maps of past land cover would be separated into naturally occurring vegetation and human induced changes, allowing us to quantify the effect of human land-use on past climate. Here we investigate the possibility of combining regional, fossil pollen based, land-cover reconstructions with, population based, estimates of past human land-use. By merging these two datasets and interpolating the pollen based land-cover reconstructions we aim at obtaining maps that provide both past natural land-cover and the anthropogenic land-cover changes. We develop a Bayesian hierarchical model to handle the complex data, using a latent Gaussian Markov random fields (GMRF) for the interpolation. Estimation of the model is based on a block updated Markov chain Monte Carlo (MCMC) algorithm. The sparse precision matrix of the GMRF together with an adaptive Metropolis adjusted Langevin step allows for fast inference. Uncertainties in the land-use predictions are computed from the MCMC posterior samples. The model uses the pollen based observations to reconstruct three composition of land cover; Coniferous forest, Broadleaved forest and Unforested/Open land. The unforested land is then further decomposed into natural and human induced openness by inclusion of the estimates of past human land-use. The model is applied to five time periods - centred around 1900 CE, 1725 CE, 1425 CE, 1000 and, 4000 BCE over Europe. The results suggest pollen based observations can be used to recover past human land-use by adjusting the population based anthropogenic land-cover changes estimates.