B53J-04 – Continental asymmetry in climate-induced tropical drought: Driving mechanisms and ecosystem response

Authors

James Randerson
University of California Irvine
Abigail Swann
University of Washington Seattle
Charles Koven
Lawrence Berkeley National Laboratory
Forrest Hoffman (forrest at climatemodeling dot org)
Oak Ridge National Laboratory
Yang Chen
University of California Irvine

Session

Climatic Controls on Vegetation Dynamics in Tropical Forests II
Friday, December 18, 2015 14:25–14:40
Moscone West 2008

Abstract

Current theory does not adequately explain diverging patterns of future drought stress predicted by Earth system models (ESMs) across tropical South America, Africa, and equatorial Asia. By 2100 for the Representative Concentration Pathway 8.5 (RCP8.5) many models predict significant decreases in precipitation across northeastern South America and Central America. In contrast, most models predict increasing levels of precipitation across tropical Africa and equatorial Asia. Using the Community Earth System Model v1.0 with RCP8.5 simulations to 2300, we found that this longitudinal precipitation asymmetry intensified over time and as a consequence, terrestrial carbon losses from the neotropics were considerably higher than those in Africa and Asia. Carbon losses in some areas of the Amazon in a fully coupled simulation exceeded 15 kg C/m2 by 2300, relative to estimates from a biogeochemically-forced simulation in which atmospheric carbon dioxide and other greenhouse gases did not influence the atmospheric radiation budget. Variations in the amount of neotropical drying varied considerably among CMIP5 ESMs, and we used several types of analysis to identify driving mechanisms and to reduce uncertainties associated with these projections. CMIP5 models in general underestimated North Atlantic sea surface temperatures and the strength of the Atlantic meridional overturning circulation (AMOC). Models that more accurately simulated North Atlantic SSTs during the historical era had smaller mean precipitation biases and predicted greater neotropical forest drying than other models. This suggests that future drought stress in northern South America and Central America may be larger than estimates derived from the multi-model mean. Analysis of idealized radiatively coupled, biogeochemically coupled and fully coupled CMIP5 model simulations indicated that the direct effects of atmospheric carbon dioxide on plant physiology also was an important factor driving asymmetric precipitation change across the tropics, and had a similar pattern as changes induced solely from greenhouse gas effects on atmospheric radiation. We conclude by discussing the implications of the continental drought asymmetry for the vulnerability of tropical forests to fire, agriculture, and tree mortality.


Forrest M. Hoffman (forrest at climatemodeling dot org)