Zheng, YW, Zheng, Z, Tarasov, P, Qian, LX, Huang, KY, Wei, JH, Luo, CX, Xu, QH, Lu, HY, Luo, YL (2010). Palynological and satellite-based MODIS observations of modern vegetational gradients in China. QUATERNARY INTERNATIONAL, 218(2-Jan), 190-201.
The MODIS-based estimates of woody, herbaceous and bare cover percentages within a 21 x 21, 11 x 11 and 0.5 x 0.5 km window around pollen sampling sites were attributed to the respective modern pollen spectra. Reconstructions of modern woody cover using the pollen-MODIS reference data set and the modern analogue approach matched well to the original MODIS-based estimates for three selected land-cover fractions. The best of all correspondence between the reconstructed and observed values was obtained for the 21 x 21 km window, including woody (r = 0.83), herbaceous (r = 0.80) and bare (r = 0.90) cover. Discrepancies in the pollen-MODIS cross-validation may be caused by long-distance pollen transport, patchy vegetation, particularly in the densely populated eastern areas and along the steep mountain slopes, under-representation of curtain woody taxa (e.g. Larix and Populus in pollen records), and possible uncertainties in the MODIS data. A series of tests performed with the reference data set including 1452 surface-pollen spectra and representing all vegetation and climatic units of China show the pollen-based reconstructions correspond well to the MODIS-based estimations of land-cover percentages within a 21 x 21 km window, supporting previously obtained results from northern Asia and North America. Results obtained for the smaller window sizes are less accurate, particularly when reconstructing woody cover. The pollen-MODIS-based modern analogue approach can reproduce present-day characteristics of land cover in China with a reasonably high degree of accuracy despite the complexity of the topography and regionally intensive land use. The generally strong correlations encourage the application of the approach to Chinese fossil pollen records. (C) 2009 Elsevier Ltd and INQUA. All rights reserved.