Pattern scaling of simulated vegetation change in North Africa during glacial cycles

Abstract. Over the last hundreds of millennia natural rhythms in Earth's astronomical motions triggered large-scale climate changes and led periodically to humid conditions in much of North Africa. Known as African humid periods (AHPs), such times sustained river networks, vegetation, wildlife and prehistoric settlements. Mechanisms, extent and timing of the changes still cannot be completely outlined. Although AHPs along glacial cycles are recognizable in long marine sediment records, the related land cover changes are difficult to reconstruct due to scarcity of proxy data over the continent. Moreover, most available information covers only the latest AHP during the Holocene. Here we use a comprehensive Earth system model to look at additional, much earlier, possible cases of AHPs. We simulate the full last glacial cycle, aiming to reproduce the last four AHPs as seen in available proxies. The simulated AHPs seem in broad agreement with geological records, especially in terms of timing and relative strength. We focus on the simulated vegetation coverage in North Africa and we detect a dominant change pattern that seems to scale linearly with known climate forcing variables. We use such scaling to approximate North African vegetation fractions over the last eight glacial cycles. Although the simple linear estimation is based on a single mode of vegetation variability (that explains about 70 % of the variance), it helps to discuss some broad-scale spatial features that had been only considered for the Holocene AHP. Extending the climate simulation several millennia into the future reveals that such pattern scaling breaks when greenhouse gases become a stronger climate change driver.