A field experiment was conducted on a sloping Tokomaru silt loam (Typic Fragiaqualf). This soil has a permeable topsoil, but is underlain with an impermeable fragipan at 500 - 700 mm. Paired runoff plots (100 m2
in area) were constructed, and over a four-year period, rainfall, water table levels and runoff were monitored. Infiltration and saturated hydraulic conductivity were also measured. During the study period, forty surface runoff events occurred with most of these being relatively minor events where less than 1 mm of runoff was generated. Subsurface flow rather than overland flow removed most of the excess rain over winter and early spring. The dominant surface-runoff-generating process was found to be saturation overland flow. Hortonian runoff only occurred when the infiltration capacity was reduced by pugging of the soil surface. When such a treatment was applied to one of the plots, the saturated hydraulic conductivity of the topsoil was reduced from 500 - 1200 mm/day to only 0.8 mm/day. For the damaged plot, 25% (1.8 mm) of the 7.2 mm of rain which fell over a seven hour period became overland flow. This compared to 18% (1.3 mm) of the rain becoming surface runoff on the undamaged plot. Aspects of modelling and simulation are discussed and several rainfall-runoff models are reviewed. A simple, physically-based, finite-difference model for predicting water table behaviour and runoff generation is described. The model uses the Dupuit-Forchheimer assumptions for flow below the water table, and it assumes a constant hydraulic potential between the water table and the surface. Water table behaviour during and following a rainstorm was predicted reasonably accurately. Using hourly rainfall data, the occurrence and magnitude of runoff surface events over a winter/spring period were simulated. The model was also used to illustrate the importance of slope angle in subsurface flow and runoff generation. Over a winter/spring period, a 100 m2
plot with a 10% slope was predicted to have nine days on which overland flow occurred, and subsurface flow rates which sometimes exceeded 30 mm/day. For the same period, a plot with a 2% slope was shown to have much less subsurface flow (with rates not exceeding 10 mm/day), and nineteen days of surface runoff.