These commits add a steady-state flux routing hydrology scheme. It routes basal water conservatively down the hydraulic gradient from source cells (where bmlt > 0) to the ice sheet boundary. It is fully parallel, adapted from a similar serial scheme that was written some years ago for Glimmer. By conserving basal water and allowing the water to flow through cells with basal temperature below the pressure melting point, this scheme improves on the current local till model used in many Greenland simulations. To use the new scheme with Glissade, the user sets which_ho_bwat = 3 = HO_BWAT_FLUX_ROUTING in the config file. The driver is subroutine glissade_bwat_flux_routing, in module glissade_basal_water.F90.
The main computational difficulties arise from interior depressions that can be sinks for meltwater. These regions are handled using the algorithm of Planchon and Darboux (2001). In depressions, the value of the hydraulic potential is raised just enough to give every cell a downhill route to the ice sheet boundary, and no farther. This algorithm parallelizes fairly efficiently.
The scheme computes the steady-state water flux through each cell. Given this flux and an expression for water velocity, it is easy to compute basal water depth. However, these depths are small (~ 1 mm or less). It is unclear whether effective pressure should depend on the flux or the nominal depth.
From a given cell, water can be routed downhill to one down-gradient neighbor, two neighbors, or up to eight neighbors depending on the value of ho_flux_routing_scheme, a new config option. More testing is needed to determine which routing scheme is most realistic.
The scheme has been tested by Sarah Bradley and colleagues in LGM runs with a North American Ice Sheet, and shown to give rise to ice streams where none were present before. It will next be tested for Greenland and Antarctica.
Answers are BFB, apart from runs that use the new hydrology scheme
These commits add a steady-state flux routing hydrology scheme. It routes basal water conservatively down the hydraulic gradient from source cells (where bmlt > 0) to the ice sheet boundary. It is fully parallel, adapted from a similar serial scheme that was written some years ago for Glimmer. By conserving basal water and allowing the water to flow through cells with basal temperature below the pressure melting point, this scheme improves on the current local till model used in many Greenland simulations. To use the new scheme with Glissade, the user sets which_ho_bwat = 3 = HO_BWAT_FLUX_ROUTING in the config file. The driver is subroutine glissade_bwat_flux_routing, in module glissade_basal_water.F90.
The main computational difficulties arise from interior depressions that can be sinks for meltwater. These regions are handled using the algorithm of Planchon and Darboux (2001). In depressions, the value of the hydraulic potential is raised just enough to give every cell a downhill route to the ice sheet boundary, and no farther. This algorithm parallelizes fairly efficiently.
The scheme computes the steady-state water flux through each cell. Given this flux and an expression for water velocity, it is easy to compute basal water depth. However, these depths are small (~ 1 mm or less). It is unclear whether effective pressure should depend on the flux or the nominal depth.
From a given cell, water can be routed downhill to one down-gradient neighbor, two neighbors, or up to eight neighbors depending on the value of ho_flux_routing_scheme, a new config option. More testing is needed to determine which routing scheme is most realistic.
The scheme has been tested by Sarah Bradley and colleagues in LGM runs with a North American Ice Sheet, and shown to give rise to ice streams where none were present before. It will next be tested for Greenland and Antarctica.
Answers are BFB, apart from runs that use the new hydrology scheme