We develop a new data assimilative (DA) system by combining two parallel frameworks: the Parallel DA Framework (PDAF) and the Earth System Modeling Framework (ESMF). The new DA system is built on ESMF at the top level driving PDAF and any combination of Earth system model (ESM) components, thereby allowing maximum flexibility and ease of implementation of DA in fully coupled ESM applications, with minimal intrusive code changes to the original model components. We demonstrate the new DA system using the 3D unstructured-grid ocean model SCHISM as the ESM component. The new system is validated using both simple benchmarks and a realistic simulation of the Kuroshio current system around Taiwan. The new system improves the model skill for temperature and surface elevation before, during and after typhoon events. The flexibility and ease of implementation make the new system widely applicable for other coupled ESMs, thus paving the way for a new generation of data assimilative cross-scale ESMs.