This paper reviews three major areas of development in the system-scale study of braided river behaviour: (i) computational modelling; (ii) spatial and dynamic scaling analyses; and (iii) remote sensing of bed morphology and morphological change. With each, the paper explains the principles behind them and addresses the importance of their contribution to the system-scale understanding of braided river form. It is argued that reduced complexity approaches to understanding braiding behaviour have resulted in a step change in the kinds of questions that can be asked about the braiding process. In particular, the work has shown that braiding appears to be the fundamental instability that arises when water moves over non-cohesive material in the absence of lateral confinement. The classic continuum of channel pattern is recast, with braided channels being the default channel pattern that a river reverts to in the absence of lateral confinement or channel resistance effects due to geomorphological (i.e. valley width), sedimentological (presence of cohesive sediment) or vegetation effects (root binding and sediment resistance to entrainment). Whilst it is possible to assess the conceptual basis of these reduced complexity models critically, as well as empirically, this step change in understanding is likely to be robust. The more complex models involving fewer process descriptions and proper numerical solution offer additional prospects. However, both reduced and increased complexity modelling approaches require different ways of approaching model validation: it is not possible to compare a time-series of predictions of braided river morphology with measured data. Severe non-linearities and boundary conditions mean that all models will diverge rapidly from the system they are designed to represent. For this reason, synthetic descriptors of river braiding have been developed as a means of comparing models, as well as increasing our fundamental understanding of the braiding process. Commonly, these are based upon spatial and dynamic scaling analyses and they have provided considerable insight into the fundamental nature of the braiding process. The key finding has been that braided rivers may well be self-organized, and this has been shown for a number of rivers, but that this state may be perturbed by the geological, geomorphological and probably ecological setting through which the river passes. The work has also shown that the evolution of channel pattern in smaller areas is forced by the evolution of channel pattern in larger areas. Research questions remain, notably in relation to the role of lateral constraints in causing a switch from self-organization to self-affinity, as well as over the extent to which these methods can be extended to remotely sensed measurements of braided river channel change. In turn, answering the latter question is being facilitated by developments in remote sensing, and notably laser altimetry and photogrammetry. These developments have allowed us to produce high-resolution digital elevation models of braided river morphology for areas with considerable spatial extent (rivers > 1 km wide and 3 km in length). This morphological evidence shows that:
1 braiding, as is conventionally produced in laboratory scaled models, appears to occur at all flows, regardless of magnitude, but is confined to a weakly meandering braid bar belt, whose location is fixed in the short term by the location of larger scales of active braidplain morphology;