Login / Signup

Inconsistent estimates of forest cover change in China between 2000 and 2013 from multiple datasets: differences in parameters, spatial resolution, and definitions.

Yan LiDamien Sulla-MenasheSafa MotesharreiXiao-Peng SongEugenia KalnayQing YingShuangcheng LiZongwen Ma
Published in: Scientific reports (2017)
The Chinese National Forest Inventory (NFI) has reported increased forest coverage in China since 2000, however, the new satellite-based dataset Global Forest Change (GFC) finds decreased forest coverage. In this study, four satellite datasets are used to investigate this discrepancy in forest cover change estimates in China between 2000 and 2013: forest cover change estimated from MODIS Normalized Burn Ratio (NBR), existing MODIS Land Cover (LC) and Vegetation Continuous Fields (VCF) products, and the Landsat-based GFC. Among these satellite datasets, forest loss shows much better agreement in terms of total change area and spatial pattern than do forest gain. The net changes in forest cover as a proportion of China's land area varied widely from increases of 1.56% in NBR, 1.93% in VCF, and 3.40% in LC to a decline of -0.40% in GFC. The magnitude of net forest increase derived from MODIS datasets (1.56-3.40%) is lower than that reported in NFI (3.41%). Algorithm parameters, different spatial resolutions, and inconsistent forest definitions could be important sources of the discrepancies. Although several MODIS datasets support an overall forest increase in China, the direction and magnitude of net forest change is still unknown due to the large uncertainties in satellite-derived estimates.
Keyphrases
  • climate change
  • machine learning
  • high resolution
  • deep learning
  • simultaneous determination
  • quality improvement
  • solid phase extraction