Chances are, you’ve grown up hearing that trees are always good for the environment – but that isn’t always the case. Afforestation – the creation of new forests, as distinguished from reforestation (which deals with replenishing existing forests) – can be a hotly debated topic in environmental circles. Now, researchers from a wide range of European institutions have teamed up to apply the power of supercomputing to demonstrate why going green doesn’t necessarily mean growing trees.

While trees do sequester carbon dioxide when they photosynthesize, that carbon largely stays in the physical body of the tree – meaning that if wood from a tree burns or decomposes, much of the carbon can return to the atmosphere. But trees themselves also affect the physics of climate change, perhaps most crucially by changing the albedo (reflectivity) of a surface, which affects how much warming occurs in a given area. Even other, smaller changes – like shifts in the coarseness of an area or changes in the distribution of water – can have subtle, but important effects on the local climate.

Edouard Davin, a senior scientist at the Institute for Atmosphere and Climate at ETH Zürich, worked with his colleagues to model these deeply intertwined effects using a supercomputer at the Swiss National Supercomputing Centre (CSCS): Piz Daint. The Swiss system is equipped with 5,704 Cray XC50 nodes (each hosting one Intel Xeon E5-2690 CPU and one Nvidia Tesla P100 GPU) and 1,813 Cray XC40 nodes (each hosting two Intel Xeon E5-2695 CPUs). The main section of Piz Daint delivers 21.2 Linpack petaflops, placing it tenth on the most recent Top500 list of the world’s most powerful publicly ranked supercomputers.

Using Piz Daint, the researchers used a suite of nine regional climate models (RCMs) to examine the effects of afforestation across Europe in terms of warming. All models showed a decrease in albedo during the winter, leading to more heat absorbed during snowy months when trees absorb more light than bright white snow would. These warmer winters were, however, complicated by discordant results from the nine models during the other seasons, with some models predicting aggressive cooling from afforestation and some predicting aggressive warming. The results were published in Earth System Dynamics.

The researchers are continuing to improve and build upon these models in an attempt to bring them into closer harmony. One element, Davin explained, will be variation among tree species.

“Another essential aspect to consider in future research is the type of trees used for afforestation,” Davin said in an interview with CSCS’ Simone Ulmer. “Foresters tend to favor coniferous trees for economic reasons. But in many regions, broadleaf trees could be better adapted to a warmer climate and provide more cooling effect.”