Sensitivity of simulated surface runoff to mesoscale meteorological model resolution in a linked-model experiment

A mesoscale meteorological model (MM5) is linked to a hydrologic model to simulate river-basin response to single-storm events. MM5 uses a nested-domain configuration, with grid increments of 36, 12 and 4 km, to produce high-resolution precipitation fields for input to the hydrologic model. A problem that will arise when performing longer-term simulations is the extraordinary computational demands of the nested MM5. To evaluate the effect that the MM5 resolution has on the simulation of direct surface runoff in the linked-model experiments, and with the goal of decreasing the computational intensity of these experiments, 3 single-storm events and their basin response were simulated with MM5 using 3 domain set-ups: 36-12-4, 36-12, and 36 km. The results show that the 36-12 km set-up generates similar patterns of precipitation and direct surface runoff to those of the 36-12-4 km domain set-up. The 36 km domain set-up produces unrepresentative precipitation distributions in time and space. It is concluded that 12 km precipitation fields may be a suitable compromise, providing sufficient resolution for simulating the basin response to climate variation and change.