Geoscientific simulations are becoming more complex due to the inclusion of more detailed physics and intricate process interactions. This increasing complexity has also been steered by the growing availability of and advances in high-performance computing (HPC). However, future growth of computational power is decreasing, raising the question how new technical progress can contribute to progress numerical solutions.

In this session, we welcome contributions demonstrating concepts to overcome this bottleneck with a focus on the application fields of Solid Earth Sciences and Terrestrial Systems modeling, as in these research areas HPC is emerging, but not yet widely established. These concepts could include, for instance, the portability to GPUs (in particular domain-specific languages), algorithms for parallelization, and (physics-based) machine learning methods. As applications, we seek contributions that, among others, develop algorithms to efficiently enable modeling the interaction of multiple, non-linear processes, or bridge several orders of magnitude in temporal and spatial scales.