The dataset contains all files to reproduce the figures in the paper North American boreal forests are a large carbon source due to wildfires from 1986 to 2016. These figures are created by Matlab, Python and ArcGIS. For Python, a environment of Python 2.7 or Python 3.7 with packages (pandas, numpy, scipy, matplotlib) pre-installed is required. The files with the extension of *.sglburnemit are essentially text files.
Wildfires are a major disturbance to influence forest carbon balance through both immediate combustion emissions and post-fire ecosystem carbon dynamics. Here we use a process-based biogeochemistry model, the Terrestrial Ecosystem Model, to simulate carbon budget in Alaska and Canada during 1986-2016 considering fire disturbances. The difference Normalized Burn Ratio (dNBR) data for fires are extracted from Landsat TM/ETM imagery, and used to estimate the proportion of vegetation and soil carbon combustion. We find that the region is a carbon source of 2.74 Pg C during the 31-year period. The loss is attributed to fire emissions at 57.1 Tg C/yr, overwhelming the net ecosystem production at 1.9 Tg C/yr in the region. Our during-fire emission for Alaska and Canada are lower than some field measurements and model estimations (for Alaska: 1.4 Tg C/yr versus 1.6-3.3 Tg C/yr; for Canada: 2.1 Tg C/yr versus 1.3-4.3 Tg C/yr). Fire severity complicates after-fire carbon dynamics, with low severity fires increase soil temperature and decrease soil moisture, stimulating soil respiration. However, the opposite trend is found under moderate or high fire severity. Net nitrogen mineralization rates gradually recovered after fire, enhancing net primary production. Net ecosystem production recovers quicker under higher burn severities. Overall, our carbon budget analysis might be biased mainly due to the burn severity uncertainty.
Cite this work
Researchers should cite this work as follows:
- Zhao, B., Zhuang, Q., Pumpanen, J., Shurpali, N. (2020). North American boreal forests are a large carbon source due to wildfires from 1986 to 2016. Purdue University Research Repository. doi:10.4231/TSK3-1733