Vegetation Cover for Last Glacial Maximum

We use a simple statistical approach to reconstruct the global vegetation cover for the LGM (Last Glacial Maximum). Following the IGBP vegetation classification, we divide the land cover into 11 vegetation types plus bare soil. For each of them, we derive a statistical relationship between the probability of its occurrence and climate variables, referred to as the VC-models. The statistical relationships are then CO2 corrected. Using these VC-models without and with CO2 corrections, we reconstructed the vegetation densities under LGM climate conditions based on an ensemble average of 6 GCMs/ESMs of the 3rd Phase of the Paleoclimate Modeling Inter-comparison Project (PMIP3). In deriving the VC-models, the AVHRR vegetation data and the ERA-Interim reanalysis data are used. Nine different VC-models were tested by reconstructing today vegetation using the ERA-Interim data and comparing the reconstructions with the observations. Among the models tested, the VC-models using monthly mean rainfall and monthly mean temperature as climate drivers give the best overall performance. The PMIP3 data used for vegetation reconstruction and reconstructed vegetation of 1 deg resolution for the LGM are stored here