Return-flow prediction and buffer designation in two rainforest headwaters.

Within tropical rainforests, most rainfall infiltrates the soil to then return to the surface within stream channels or from streamside soils. The wet zones created are very sensitive to disturbance and should be protected from forestry (or agricultural) operations. This study identifies the location of returning subsurface waters using measurements of the topsoil moisture content and first-order, stream-head location. The ability of a topographically based index to predict the location of both channel heads and wet soils is then evaluated to assess its utility for streamflow generation modelling and the designation of buffer or protection zones. Analysis of the location of perennial stream-heads in the 0.44 km2 Baru catchment (East Malaysia) indicated that these could be predicted by the Kirkby topographic index (λ spatial distribution) to within two terrain pixels (each 20 m × 20 m). Within this catchment, topography, therefore, played a key role in the generation of perennial streams, and the λ spatial distribution may be useful in locating the perennial stream network on digital maps for objective ‘hydrological buffer zoning’. The spatial structure of 468 soil moisture content measurements within the 0.21 km2 Huai Pacha catchment (Northern Thailand), was identified by variogram analysis and then used to interpolate the data. A patchy distribution of near-saturated soils was observed along the perennial streams. The λ spatial distribution was poorly correlated with the whole moisture data-set, showing the need to characterise soil and geological variability at this locality in addition to topographic factors. The wet patches did, however, coincide with areas of high topographic convergence and with values of the λ spatial distribution greater than 9 ln(m). Use of this threshold in the λ spatial distribution, rather than a fixed-width riparian buffer defined smaller protection zones with a higher proportion of the sensitive saturated or near-saturated soils. Consequently, the λ-based buffers would be more efficient, easier to justify scientifically, and allow access to more of the timber growing on less sensitive, drier soils. More studies on return-flow are needed in tropical rainforests to add to our understanding of runoff processes, validate distributed hydrological models and provide a hydrological basis for the objective definition of forestry buffer or protection zones within this critical environment