Reproducibility in computational geoscience has improved substantially through open data policies and shared source code; however, structured reuse of analytical workflows across domains remains challenging. While data and scripts may be available, they are often difficult to re-execute due to missing documentation, unstandardised environments, or insufficient workflow orchestration. The Data-to-Knowledge Package (D2KP) concept addresses this limitation by integrating FAIR data, modular toolboxes, executable workflows, and virtual research environments into reusable research metaobjects. This contribution presents three heterogeneous D2KPs implemented within the AquaINFRA research infrastructure for marine and freshwater science: (1) dasymetric population refinement for the Elbe river basin, (2) ensemble environmental outlier detection using the specleanr R package, and (3) reproducible spatiotemporal trend detection for water transparency analysis in the Gulf of Riga. The poster visualises the AquaINFRA infrastructure architecture, the internal workflow structures of each use case, and the resulting analytical outputs. By embedding domain-specific analyses into a shared infrastructure backbone, the D2KP approach demonstrates how reproducible research can be transformed into interoperable and reusable geospatial services.