View article

The practise of climate modeling has evolved from study of individual subsystems to integrated coupled climate or earth system models. Coupled climate models comprise general circulation models (GCMs) for both the atmosphere and ocean, a land-surface model, and a sea-ice model. Rivers play an important role in the Earth’s hydrological cycle (Figure 1), and most climate system models now include continentalscale river transport models (RTMs) to complete the global water balance. The RTM takes as its input runoff calculated by the land-surface model, routes it through the river network and ultimately into the system’s ocean component.

Many continental-to-global scale river transport models (RTM’s) exist, but none are currently simultaneously able to achieve high performance and to advect tracers. We are developing a massively parallel dynamical core (dycore) for river transport modeling at the continental-to-global scale. This approach treats the world’s river networks as a directed graph G, whose vertices represent the centroids of grid cells or catchments, and whose edges represent the surface flow paths or river reaches between catchments. For RTMs using a linear reservior assumption, transport becomes a linear transformation, with the transport matrix T having identical structure to the adjacency matrix A of G.