Predictive modeling of the potential natural vegetation pattern in northeast China

Based on the characteristics of natural vegetation distribution in northeast China, using multivariate analysis and geographical information system technology, we established a regional 'vegetation-environment' model to simulate geographical distribution of 16 natural vegetation types under present environmental conditions, representing the potential natural vegetation (PNV) of northeast China, on the basis of digital maps of seven environmental variables including climate and topography. Comparison of simulated PNVs distributions with the actual natural vegetation distribution indicated a good agreement, with overall predictive accuracy of 66.9% and overall Kappa value of 0.67. The predictions of model, however, were poor, for only 0.62 of AUC value was yielded. The current resolution and accuracy of the model can be applied to simulate and map the natural vegetation pattern at the regional scale and also used to analyze the effect of climatic changes on natural vegetation.