Scale effects on headwater catchment runoff timing, flow sources, and groundwater-streamflow relations

[1] The effects of catchment size and landscape organization on runoff generation are poorly understood. Little research has integrated hillslope and riparian runoff investigation across catchments of different sizes to decipher first-order controls on runoff generation. We investigated the role of catchment sizes on riparian and hillslope dynamics based on hydrometric and tracer data observed at five scales ranging from trenched hillslope sections (55–285 m) to a 280-ha catchment at Maimai on the west coast of the South Island, New Zealand. The highly organized landscape is comprised of similar headwater catchments, regular geology, steep highly dissected topography, relatively consistent soil depths, and topographically controlled shallow through flow. We found a strong correlation between riparian zone groundwater levels and runoff for the headwaters, whereas the water tables in the valley bottom of the larger catchments were uncorrelated to runoff for 14 months of record. While there was no clear relationship between catchment size and new water contribution to runoff in the two storms analyzed in detail, lag times of tracer responses increased systematically with catchment size. The combination of hydrometric and tracer data allowed assessment of the runoff contributions from different parts of the landscape. Runoff was generated consistently in headwater riparian zones. This agreed also with the observed variations of tracer (O and silica) responses for the different catchments. During wetter antecedent conditions or during larger events (>30 mm under dry antecedent conditions) hillslope and valley bottom floodplains did contribute to event runoff directly. We propose that analysis of landscape-scale organization and the distribution of dominant landscape features provide a structure for investigation of runoff production and solute transport, especially as catchment-scale increases from headwaters to the mesoscale.