You are using an old browser with security vulnerabilities and can not use the features of this website.
Gunnar Lischeid is a researcher who likes to keep track of things. He sorts huge amounts of data, reveals patterns that open up new knowledge, and detects interrelations between processes in a landscape. At the Leibniz Centre for Agricultural Landscape Research (ZALF) in Müncheberg he heads the “Data” research platform in addition to being Professor of Landscape Hydrology at the University of Potsdam.
The rare moor frog and the common spadefoot live there as well as the marsh harrier and the water rail: Water-filled depressions in fields, often surrounded by thick belts of plants, are tiny biotopes – and valuable ecosystems in the agricultural landscape. But Professor Gunnar Lischeid is interested not only in the species of animals, plants, or microorganisms living there. First and foremost, he would like to find out more about nutrients and contaminants in small bodies of water known as kettle holes.
These processes are surprisingly intensive and complex, despite the fact that most kettle holes cover much less than one hectare. And what is more, their characteristics vary from hole to hole. Some emit trace gases such as methane that contribute to climate change – methane is a 25 times more potent greenhouse gas than carbon dioxide – whereas others don’t. Researchers have yet to find an explanation for this. What is known about nutrient decomposition and material cycles in these kettle holes explains only a fraction of the measured data. So researchers are scratching their heads as to why the data do not fall in line with their expectations. “So far, 95% is yet to be understood,” Lischeid states.
Researchers are analyzing water samples and sediment cores, studying pH levels and chemical compositions, and testing various hypotheses in laboratory experiments. But Lischeid wants to understand the remaining 95% as well, and by other means than measuring or counting. He and his team are less interested in individual processes: “We want to study the interaction of many processes and understand them on a larger scale, in the landscape.”
Long-term data collected over decades and stored in large databases, then, are his greatest treasure and mathematics and computer science his most powerful tools. With modern data mining methods, Lischeid, who is also an agricultural engineer, searches for previously unknown patterns and interrelations. Large data pools are analyzed with complex statistical methods to find new knowledge and results.
“These methods are not as common yet in environmental sciences, so it is high time to do something about it,” Lischeid explains and gives an example of what knowledge can be gained when traditional methods are combined with data mining. In a forest in eastern Bavaria, Lischeid analyzed the turnover of nitrogen entering via the air and being discharged via the groundwater. Measurements in brooks of the drainage area revealed: “A major percentage of the nitrogen just disappeared.” The researchers assumed that the wetlands were responsible for the depletion. However, measurements indicated that the wetlands removed only a fraction of the nitrogen.
Statistical methods brought Lischeid closer to an explanation: It turned out that the by far largest amount of nitrogen was being depleted in the forest soil and in deeper layers of the soil above groundwater level. It had previously been assumed that this process would take place only under oxygen deficiency. But there was sufficient oxygen in the forest soil! “My colleagues couldn’t believe it at first,” Lischeid remembers. Additional measurements confirmed the result. It was found that the nitrogen was being depleted by microorganisms in tiny, anaerobic soil aggregates – a process completely underrated at the time. “In retrospect, everything is conclusive and logical, but someone needs to spell it out for you.”
Today, such thought-provoking impetuses often come from statistical data analyses. At the ZALF, the management of large research data sets, their analyses and documentation is being further intensified. The “Data” research platform headed by Lischeid is one of six new structural units set up here earlier this year. The combination of huge data volumes by statistical methods offers great potential for identifying complex interrelations, Lischeid is sure. Data mining could help environmental authorities, for instance, who have to decide on the basis of data which developments are harmful and which are not in order to arrive at new, more reliable estimates. Researchers can come to better understand why some of their models often go wrong. And in very special biotopes like kettle holes, many secrets could be revealed if the right data strings are pulled.
Critical thresholds, non-linear processes, or complex interactions in environments are some of the phenomena Lischeid is fascinated by. As a researcher, he wants to get to the bottom of them. “There is no way to find out more using classical methods in this case,” he explains. In the end, it was a theoretical physicist who inspired him to think out of the box.
He is now dealing with apparently strange methods like artificial neuronal networks, self-organizing maps, dimensionality reduction, and Sammon's Mapping – all enormously effective analytical tools for precisely and quickly processing high volumes of measurements. He enjoys twisting and turning data in all directions to find patterns which may become visible with one of the methods, he references various charts and, much like a detective, looks for clues to solve problems. As the head of a team of 22 researchers, there is little time to do this during the day, Lischeid admits. But when the Institute goes quiet in the evening, he broods over diagrams, scatterplots, and curves, or drills himself in new methods.
Often, what Lischeid finds when “playing around” with data is not surprising at all, and the interrelations opening up before him are logical and simple. In this case, the initial reflex is “We could have worked this out for ourselves”. Nevertheless, the existing knowledge can be revalued and reclassified based on his findings.
Data science is a booming research field, yet it has not been able to attract enough young talent. “Of course, businesses can pay young researchers much higher salaries.” After all, data mining techniques are also profitable in business consulting, marketing, and industrial enterprises.
Incidentally, Lischeid gets his best ideas while travelling on a train, as he tells us. “I love business trips by train.” There he finds time to sort ideas and develop them further. So it is no big deal for him to take the train to Müncheberg every day. As the train passes by kettle holes in the fields, it is the most fertile working atmosphere for the agronomist.
The Leibniz Centre for Agricultural Landscape Research (ZALF) studies the sustainable utilization of agricultural landscapes. Its research focuses on social challenges such as climate change, food security, and the protection of biodiversity. Set up in the 1920s in Müncheberg, the Institute initially specialized in plant breeding. Today, it explores the natural- and social-scientific fundamentals of processes in agricultural landscapes, the effects of various uses, and resulting conflicts.
The pearls – Potsdam Research Network connects the University of Potsdam and 21 non-university research institutions in the science region of Potsdam/Berlin. The network focuses on joint research projects, developing young researchers, and joint research marketing of the science region Potsdam.
Gunnar Lischeid studied agriculture and geology, holds a PhD in forestry, and habilitated in hydrology. At the ZALF, he heads the research platform “Data” set up in 2018, and he is Professor of Landscape Hydrology at the University of Potsdam.
In our series “Pearls of Science” we regularly introduce researchers from institutions connected with the University of Potsdam in the “pearls – Potsdam Research Network”.
Text: Heike Kampe
Translation: Monika Wilke
Published online by: Alina Grünky
Contact to the online editorial office: email@example.com