Is representative elementary area defined by a simple mixing of variable small streams in headwater catchments?

The spatial variability of hydrology may decrease with an increase in catchment area as a result of mixing of numerous small-scale hydrological conditions. At some point, it is possible that a threshold area, the representative elementary area (REA), can be identified beyond which an average hydrologic response occurs. This hypothesis has been tested mainly via numerical simulations, with only a few field studies involving simple mixing. We tested this premise quantitatively using dissolved silica (SiO(2)) concentrations at 96 locations that included zero-order hollow discharges through sixth-order streams, collected under low-flow conditions within the 4.27-km(2) Fudoji catchment. The catchment possesses a simple topography consisting almost solely of hillslopes and stream channels, uniform bedrock geology, soil type and land use in the Tanakami Mountains in central Japan. Dissolved SiO(2) provides a useful tracer in hydrological studies insofar as it is responsive to flowpath depth on hillslopes of uniform geology. Our results demonstrate that even in a catchment with an almost homogeneous geology and simple topography, dissolved SiO(2) concentrations in zero-order hollow discharges largely varied in space and they became similar among sampling locations with area of more than 10-1-100 km(2). Relationships between stream order and standard deviation of SiO(2) concentration closely matched the theoretical predictions from simple mixing of random fields. That is, our field data supported the existence of the REA and showed that the REA was produced by the simple mixing of numerous small-scale hydrological conditions. The study emphasizes the need to consider both the heterogeneous nature of small-scale hydrology and the landscape structure when assessing the characteristics of catchment runoff. Copyright (C) 2010 John Wiley & Sons, Ltd.