Spatial variability of rainfall governs river flow and reduces effects of land use change at landscape scale:

Abstract Empirical data show an abundance of evidence for effects of land use change on streamflow (total quantity, sediment load, amplitude of fluctuations) for spatial scales up to 100 km2 but little hard evidence beyond that scale. Is that simply based on lack of research, or are other factors, such as low spatial correlation of rainfall events plus differentiation in routing times starting to dominate beyond this scale? Are changes in local buffering that are linked to land use change swamped by these other effects? The GenRiver model is a distributed model based on basic water balance at subcatchment level, linked to a rainfall generator SpatRain that can generate a wide range of space/time patterns of rainfall. The impacts of routing time differentiation and a simple length scale of the rainfall pattern can now be compared to impacts of differences in intercep-tion, infiltration capacity and storage linked to land use change. Model parameterization for the Sumber Jaya area in Lampung (Indonesia) can generate patterns of daily river flow that are similar to the observed fre-quency distributions, if a strongly disaggregated rainfall pattern is used for the input, and not for more homo-geneous rainfall patterns. Although a more ‘patchy ’ rainfall may induce more surface quickflow at field sca-le, it tends to a more regular pattern of riverflow at landscape scale. We conclude that the factors dominating river flow patterns at landscape scale (approximately 5th order rivers) are thus essentially different from those dominant at field scale, and that effects of land use change are likely to become less important with increasing scale of consideration