Erosion and sedimentation as multiscale, fractal processes: implications for model, experiments and the real world

Past research and existing models allow a fair prediction of the soil and water balance at plot level under different land-use types, but do not apply to the landscape mosaics of the real world. We propose a classification scheme for landscapes on the basis of the presence of `filter' strips in between cropped fields and explore what research is needed to `scale-up' from homogenous experimental plots to heterogeneous landscapes. Both slope length and slope width can affect the net soil loss per unit area from a mosaic element. Total water and sediment yields from landscapes with a partial tree cover (agroforestry) may respond differently to interventions. For sediment load of rivers the location of trees may be more important than the area covered by trees, as riparian strips can intercept substantial sediment movement from upslope. By contrast, total water yield and base flow of rivers may be determined by the percentage forest cover, virtually independent of tree location. Farm productivity effect of vegetative soil conservation measures are a net effect of the labour and other costs involved and the yield. Harvested yield per unit area depends on the area occupied by the soil conservation strips, the effects on crop yields of soil redistribution within the slope, effects of heterogeneous water infiltration, shading and competition for water and nutrients between vegetation and crop, and effects on the soil organic matter balance. The WaNuLCAS model for spatially zoned agroforestry systems can predict where the net physical yield effect can be positive. Erosion and sedimentation processes primarily affect the heterogeneity of soil fertility within a landscape. The main issues thus are whether or not fertility can be exploited on the place where it ends up, and whether or not this outweighs the yield opportunities lost at the sites of net soil loss

---

• View source