Main content area

Exploring uncertainty of Amazon dieback in a perturbed parameter Earth system ensemble

Boulton, Chris A., Booth, Ben B. B., Good, Peter
Global change biology 2017 v.23 no.12 pp. 5032-5044
climate, climate change, dieback, dry season, emissions, forests, leaf area index, models, photosynthesis, temperature, uncertainty, Amazonia
The future of the Amazon rainforest is unknown due to uncertainties in projected climate change and the response of the forest to this change (forest resiliency). Here, we explore the effect of some uncertainties in climate and land surface processes on the future of the forest, using a perturbed physics ensemble of HadCM3C. This is the first time Amazon forest changes are presented using an ensemble exploring both land vegetation processes and physical climate feedbacks in a fully coupled modelling framework. Under three different emissions scenarios, we measure the change in the forest coverage by the end of the 21st century (the transient response) and make a novel adaptation to a previously used method known as “dry‐season resilience” to predict the long‐term committed response of the forest, should the state of the climate remain constant past 2100. Our analysis of this ensemble suggests that there will be a high chance of greater forest loss on longer timescales than is realized by 2100, especially for mid‐range and low emissions scenarios. In both the transient and predicted committed responses, there is an increasing uncertainty in the outcome of the forest as the strength of the emissions scenarios increases. It is important to note however, that very few of the simulations produce future forest loss of the magnitude previously shown under the standard model configuration. We find that low optimum temperatures for photosynthesis and a high minimum leaf area index needed for the forest to compete for space appear to be precursors for dieback. We then decompose the uncertainty into that associated with future climate change and that associated with forest resiliency, finding that it is important to reduce the uncertainty in both of these if we are to better determine the Amazon's outcome.