How do ecosystems respond to changing weather patterns, rising temperatures and increasing carbon dioxide concentrations? Is the effect of precipitation more important than that of temperature? Or are ecosystem dynamics more strongly affected by nutrient availability? What is the role of extreme events in shaping biogeochemical cycles? To find out the answers we need to understand the interactions among three complex systems: climate, vegetation, and soil. Thus, we combine experiments and in-situ long-term observation with Earth Observations gathered by aircraft and satellites across a range of spatial scales, and embrace data-driven machine learning and theory-driven mechanistic modelling. With our research, we try to understand how the terrestrial biosphere reacts to and exerts feedbacks on ongoing environmental change and variation in atmospheric conditions.
Latest publications
Hodač, L., Karbstein, K., Kösters, L., Rzanny, M., Wittich, H. C., Boho, D., et al. (2024). Deep learning to capture leaf shape in plant images: Validation by geometric morphometrics. The Plant Journal. doi:10.1111/tpj.17053. // Figure 1: Workflow of this study from data acquisition to research questions. (a) A flowering plant of Ranunculus auricomus with developed taxonomically informative basal leaves. Image data were acquired from identical leaves as smartphone images (= in situ leaf images) and by leaf scanning after herbarization (= herbarized leaf images). The in situ leaf images had a natural background, while the herbarized leaf images had a uniform background. From both in situ and herbarized images, we automatically extracted DL features (using a CNN) and manually extracted GM shapes (using a landmark digitization). (b) Our research questions regarding clustering of DL features (RQ1), covariation of DL features and GM shapes (RQ2), and automatic classification of DL fea- tures into morpho-phylo-groups (RQ3)
Hodač, L., Karbstein, K., Kösters, L., Rzanny, M., Wittich, H. C., Boho, D., et al. (2024). Deep learning to capture leaf shape in plant images: Validation by geometric morphometrics. The Plant Journal. doi:10.1111/tpj.17053. // Figure 1: Workflow of this study from data acquisition to research questions. (a) A flowering plant of Ranunculus auricomus with developed taxonomically informative basal leaves. Image data were acquired from identical leaves as smartphone images (= in situ leaf images) and by leaf scanning after herbarization (= herbarized leaf images). The in situ leaf images had a natural background, while the herbarized leaf images had a uniform background. From both in situ and herbarized images, we automatically extracted DL features (using a CNN) and manually extracted GM shapes (using a landmark digitization). (b) Our research questions regarding clustering of DL features (RQ1), covariation of DL features and GM shapes (RQ2), and automatic classification of DL fea- tures into morpho-phylo-groups (RQ3)
1.
Rzanny, M.; Bebber, A.; Wittich, H. C.; Fritz, A.; Boho, D.; Mäder, P.; Wäldchen, J.: More than rapid identification—Free plant identification apps can also be highly accurate. People and Nature 6 (6), pp. 2178 - 2181 (2024)
Schatke, M.; Ulber, L.; Musavi, T.; Wäldchen, J.; von Redwitz, C.: From unwanted to wanted: Blending functional weed traits into weed distribution maps. Weed Research 64 (6), pp. 407 - 417 (2024)
Romanou, A.; Hegerl, G. C.; Seneviratne, S. I.; Abis, B.; Bastos, A.; Conversi, A.; Landolfi, A.; Kim, H.; Lerner, P. E.; Mekus, J.et al.; Otto-Bliesner, B. L.; Pausata, F. S. R.; Pinto, I.; Suarez-Guiterrez, L.: Extreme events contributing to tipping elements and tipping points. Surveys in Geophysics (2024)
Mauder, M.; Jung, M.; Stoy, P.; Nelson, J. A.; Wanner, L.: Energy balance closure at FLUXNET sites revisited. Agricultural and Forest Meteorology 358, 110235 (2024)
