By combining seasonal forecasts from the CEPMMT, vegetation models and machine-learning-based predictions of burnt areas, researchers are able to identify areas at risk before the event occurs.
With a potentially extreme El Niño expected in 2026, an international team — including Philippe Ciais and Pierre Friedlingstein, figures at the Global Carbon Project — is proposing an early warning framework for the Earth's carbon cycle. Rather than seeing the damage after the fact, it runs global vegetation models fed by a set of ECMWF seasonal forecasts, coupled with predictions of burned areas from machine learning.
The results predict a decrease in the net productivity of terrestrial biomes of 0.23 ± 0.21 billion tonnes of carbon (PgC) in the second half of 2026. Two hot spots stand out: a weakening of the carbon sink in eastern South America, and an increased risk of fires in Southeast Asia. The interest of the approach is to give forest and fire managers a few months in advance to prepare.
Read the full article on the original source
Go to the source →