Remote sensing of broad-scale controls on large river anabranching

Abstract

River patterns reflect complex geomorphological processes and affect ecosystems and human development along floodplains. Physical controls on anabranching development have been studied primarily at local scales on relatively small rivers (i.e., discharge <1000 m3 s−1). However, there has been little systematic quantification of river anabranching (development of stable multiple channels) for large rivers globally. Here, we use remote sensing, cloud computing, and geospatial analysis to explore the importance of water surface slope, floodplain topography, sediment supply, substrate lithology, and permafrost to anabranching throughout 20 of the world's largest river basins. We use Landsat-derived surface water extent to compute an anabranching index (Ai) for ∼1 M km of river reaches, and compare it with global datasets of water surface slope, available floodplain extent, substrate lithology, and permafrost. We find that ∼49 ± 19% of large-river mainstems are anabranching, ranging from at least ∼17% (Mekong) to as much as ∼84% (Ob’). At the basin scale, anabranching channels comprise at least ∼17% (Yenisey) to as much as ∼55% (Kolyma) of all Landsat-observable river reaches, with a mean global value of 35 ± 11%. Anabranching channel patterns are most commonly associated with low water surface slope (normalized slope < 0.2; absolute slope < 0.2 m km−1) accompanying wide floodplains (normalized floodplain width > 0.6; absolute width > 42 km). Cross-sectionally averaged channel width increases in the most intensely anabranching reaches, suggesting net sediment storage due to reduced stream power. Unconsolidated sedimentary substrates promote both the prevalence and intensity of anabranching, whereas the presence of permafrost enhances intensity only. Overall, our results identify pervasive channel anabranching and some important control mechanisms throughout the world's large river basins, with strong implications for satellite-based discharge retrievals. We conclude that global analyses afforded by remote sensing and cloud computing offer additional, complementary approaches to traditional field-based approaches for understanding large river geomorphic processes and channel forms.