Download (PDF 233 KB): https://research.fs.usda.gov/download/treesearch/43319.pdf
Abstract
The GIMMS NDVI dataset has been widely used to document a “browning trend” in North American boreal forests (Goetz et al. 2005, Bunn et al. 2007, Beck and Goetz 2011). However, there has been speculation (Alcaraz-Segura et al. 2010) that this trend may be an artifact due to processing algorithms rather than an actual decline in vegetation activity. This conclusion was based primarily on the fact that GIMMS NDVI did not capture NDVI recovery within most burned areas in boreal Canada, while another dataset consistently showed post-fire increasing NDVI. I believe that the results of Alcaraz-Segura et al (2010) were due simply to different pixel sizes of the two datasets (64 km
2 versus 1 km
2 pixels). Similar results have been obtained from tundra areas greening in Alaska, with the results simply due to these pixel size differences (Stow et al. 2007). Furthermore, recent studies have documented boreal browning trends based on NDVI from other sensors.
Citation
Verbyla, David. 2011. Browning boreal forests of western North America. Environmental Research Letters. 6: 041003. 3 p. doi:10.1088/1748-9326/6/4/041003.