Plants are removing carbon from the atmosphere – but it’s not easy to estimate the global uptake. Sophia Walther explains how climate scientists combine intelligent computer algorithms with measurements and satellite data to get a better picture.
An editorial view from Dr. Sophia Walther, Max Planck Institute for Biogeochemistry
Do you know how big your CO2 footprint is? That is, what trace you leave in the atmosphere through activities that cause CO2 emissions? Have you ever asked yourself whether your personal contribution might be bigger in winter, when you heat the house and prefer to use the car for commuting, than in summer, when you possibly ride your bicycle but use air conditioning or book a flight to your favorite holiday destination? I am sure that even if you’ve thought about it and even if you’ve done the math, you haven’t found an answer on how you impact the environment on the long term. Because you have to take into account that – and we can consider us lucky here – not all the carbon emitted into the atmosphere stays there. As we know from science: Plants need carbon to live and grow, so they take up carbon from the atmosphere in form of CO2, a process known as photosynthesis. Actually, the whole biosphere – that is all plants and organisms in the soil – is an active player in removing carbon from the atmosphere, currently about one third of anthropogenic CO2 emissions, thus actually reducing our footprint.
But plants help us even more: They also provide us with food, clothes, furniture, housing, heating or recreational walks in the woods. Just like our CO2 emitting activities change during the year and depend on the current weather, the uptake of CO2 by the land surface goes through a seasonal cycle and the plants’ activity is also affected by the actual meteorological conditions, like cold spells or dry periods.
In order to understand CO2 concentrations in the atmosphere we need to understand when, where and to what extent plants take up CO2 from the atmosphere. But there is a problem. We can only measure this natural CO2 exchange for very small areas, say the size of the nice green park in your neighborhood where you go for your weekend strolls. We can do this for many such points on Earth, but each point is rather small. It can never be measured for a whole city, let alone a continent or larger regions like the Siberian taiga. So, the challenge is to find indirect ways to come to evidence-based answers – one being artificial intelligence, based on the fact that computers are really good learners if you give them many and the right kind of data.
Remember that there are intelligent algorithms that even learned to recognize spoken words and that those are convenient helpers in everyday life for many smartphone users. Similarly, there are algorithms that can learn how the plants’ metabolism behaves and how healthy they are in times of different weather and growing conditions. They can learn this from the data obtained at the many small measurement points on Earth like in the nearby park or in the Siberian taiga. So, once the computer has sufficient information to learn, it can help us come up with a good estimate of how much carbon is sucked out from the atmosphere and taken up by the plants that grow in the places where we cannot directly measure it. Still, other relevant information is needed, for example how warm it is, whether it is a clear day or overcast and what type of vegetation grows in a specific place. This information can be obtained from satellites all around the world, also in very large and remote areas like the Siberian taiga or tropical rainforest. In that way, intelligent computer algorithms that combine point measurements and satellite data can do clever estimates of natural CO2 emissions and removals for the whole globe.
In our everyday scientific work, my colleagues and I try to improve those estimates by testing what kind of information the computer needs in order to deliver better results. Is it decisive to know how warm it is on a given day and how wet the upper soil is? Might the simulations become even better if the algorithm also had information on how wet it is deeper in the soil? What role does yesterday’s temperature play or the one from one week ago or from last winter? Is it important to know the age of the trees in the Siberian taiga? And how reliable are the estimates? For example, if we provide the algorithms with wrong data, their estimates will also be incorrect. The question is, of what quality the data fed into the algorithm need to be in order to still come up with estimates of the plants’ carbon uptake that are as reliable as possible.
Our work helps disentangling how measured CO2 concentrations in the atmosphere originate from the contributions of natural biospheric carbon exchanges and from anthropogenic emissions – including your footprint. This is crucial in order to monitor the service that our biosphere provides in taking up carbon from the atmosphere and mitigating climate change. This service results from complex interactions of biological processes of plants and soils with human activities. However, the balance of these interactions may change in the future. For example, deforestation and land degradation by humans diminish the beneficial compensation of anthropogenic CO2 emissions by our biosphere, resulting in disadvantageous effects.
This text was created during a workshop by DKK Managing Director Marie-Luise Beck at the Max Planck Institute for Biogeochemistry, in which PhD researchers spent two days working on climate communication.
About the author
Dr. Sophia Walther is working as Postdoctoral Researcher in the Global Diagnostic Modelling Group at Max Planck Institute for Biogeochemistry (MPI-BGC) in Jena.
1. september 2020
Picture credits: Sophia Walther