Efficiently Resolving the Terrestrial-Aquatic Interface in E3SM with Sub-Grid Methods to Improve Coastal Simulations
During ordinary tidal cycles or storm-forced surge events, ocean waters flood overland areas and interact with rivers, complex streamflow networks, wetlands, marshes, and upland runoff. Correctly resolving the terrestrial-aquatic interface (TAI) is thus essential to accurately model coastal hydrodynamics. Current capability of MPAS-Ocean, the ocean component of the Energy Exascale Earth System Model (E3SM), resolves the ocean and its coast with unstructured mesh elements no finer than 6 km and is not anticipated to be finer than 2 km in the near future. While 6 km to 2 km resolution is a step forward for earth system models (ESMs), it remains insufficient to properly capture the TAI connecting the upland hydrology and the ocean through the complex dendritic hydraulic conveyances that penetrate the coastal floodplain. Resolving this in MPAS-Ocean would require coastal mesh resolution a hundred times finer than currently employed. While the use of MPAS-Ocean’s unstructured meshes allows flexibility in placing local resolution, resolving the TAI down to tens of meters over large regions will lead to a significant increase in the number of spatial unknowns to be solved for and will require reductions in the time step, increasing computational costs over current MPAS-Ocean models by on the order of ten thousand to a million times. This is infeasible given current and expected computing capabilities for climate applications requiring century-long simulations.
An alternative lies with sub-grid models that integrate fine-scale properties of bathymetry, topography, and land cover to generate corrections to coarse-scale models such as MPAS-Ocean. Recent developments by the team in sub-grid theory and algorithms have been successfully implemented into storm surge models and current tests show that TAI sub-grid solutions will yield results comparable to standard solutions at least ten times coarser. Their tests demonstrate that the additional computational cost of such sub-grid methods in coastal regions is only a factor of 1.2 times more than the coarse mesh standard solution irrespective of the resolution of the underlying topographic data. This demonstrated a significant advantage of the sub-grid approach compared to the ten thousand to million times greater computational cost associated with directly resolving the TAI.
In this project, the team will adapt and incorporate a sub-grid method into MPAS-Ocean to improve the effective resolution of the TAI. The sub-grid mode is a natural fit to the finite-volume type numerical solution used in the MPAS-Ocean model. Importantly, sub-grid methods once implemented into MPAS-Ocean will allow a much more accurate representation of coastal processes and the complex hydraulic connectivity between upland regions and the ocean than currently achieved, thus improving links in the water cycle. In doing so, the response of the fully coupled earth system at the TAI to short- and long-term perturbations can be better assessed. In particular, the team plan to evaluate the response of the system to short-term highly energetic land-falling hurricanes that have a dramatic effect on the TAI through tides, ocean surge, waves, intense rainfall, and wind damage to vegetation and property. The lasting effects of such a hurricane can extend out to time scales much longer than the duration of the meteorological event itself. Such an analysis would not be possible in the current E3SM model setup. The sub-grid-corrections that will be implemented in this project will make accurate information about coastal inundation and the TAI exchanges viable in MPAS-Ocean.