NSF Award
2425845
Forest ecosystems play a critical role in the Earth system as major carbon sinks, which are essential for carbon neutralization and climate change mitigation. However, significant deforestation and forest degradation could push the Earth to climate change tipping points. As such, there is a growing interest in forest carbon sequestration through afforestation and reforestation initiatives from local to global scales. These developments have led to a strong demand in advancing the scientific understanding of the impact of forest carbon dynamics in the carbon cycle. Specifically, a new generation of models has emerged to connect individual plant level processes with the global carbon cycle. In addition, advancements in remote sensing have generated unprecedented new high-resolution measurements at the global scale. Despite these opportunities, several key challenges persist in understanding forest carbon dynamics, including the lack of understanding of fine-scale but widespread disturbances such as tree mortality in existing remote sensing products, the computational bottlenecks of the theory-based models for global scale analysis, and the limited flexibility of the models in enhancing the prediction quality using new observations. This project aims to develop new capabilities to bridge these research gaps and significantly advance the monitoring and understanding of forest carbon dynamics in the Earth system. The enhanced understanding can provide necessary information for estimating carbon budgets and realizing carbon neuralization goals. The research results will be used to develop materials for both undergraduate and graduate courses in AI and geosciences. The project will also engage students from underrepresented groups in the research activities and partner with K-12 schools to promote education on topics intersecting AI and geosciences.<br/><br/>This project will result in several advances of artificial intelligence techniques with the potential to further the understanding of how forest carbon influences the Earth system’s carbon cycle under climate change and what terrestrial ecosystems’ capacity is in climate change mitigation. First, the project team will develop cross-platform and cross-region learning frameworks to enable fine-scale carbon dynamics monitoring at large geographic scales. Second, the team will create high-fidelity fast approximations of the theory-based carbon forecasting model by developing new theory-guided meta-learning and invertible frameworks to enable global-scale capabilities under diverse climate change scenarios. Finally, the team will develop new theory-guided diffusion methods to significantly enhance the ability of theory-based models in improving predictions by leveraging observations enabled by new sensing platforms.<br/><br/>This award by the Division of Research, Innovation, Synergies, and Education within the Directorate for Geosciences is jointly supported by the National Discovery Cloud for Climate initiative within the Directorate for Computer and Information Science and Engineering.<br/><br/>This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.
