Most agree that there is room for improvement in the science and engineering basis for stream restoration. There is also a bewildering array of methods and models, the application of which requires a diverse range of expertise and effort with little clear indication of need, reliability, efficacy, and effort. There is a need for not only research and improved methods supporting stream restoration practice, but also for improved organization, distribution, and coordination of existing and emerging methods and training. The National Center for Earthsurface Dynamics (NCED), a Science and Technology Center funded by the National Science Foundation, has formed a Stream Restoration Group to organize and focus research relevant to stream restoration, to collaborate with agencies and practitioners in identifying knowledge gaps and developing improved tools for restoration practice, and to disseminate this knowledge to practitioners. Our goal is to move restoration practice to an analytical, process-based approach that will ultimately lead to better prediction and hence better design. Predictive understanding is needed in a number of key areas, including sediment routing at the reach to network scale, channel and floodplain response to watershed changes, and linkages between physical channel conditions and nutrient cycling, stream metabolism, primary production, and population dynamics. The broadest challenge facing restoration is placing projects in a watershed context. A central part of the Restoration Group’s efforts involves support and interaction with a partners group consisting of agency and industry professionals. This group helps to define research needs, to identify and contribute useful models for restoration design, to evaluate restoration practice, and to coordinate training in restoration practice. This paper outlines some efforts of the NCED Stream Restoration Group to support improved stream restoration practice, including our research priorities and examples of knowledge transfer. Research efforts focus on laboratory and field experiments on sediment transport and stream channel change. With its partners, NCED is conducting short courses and developing training materials and a web-based “toolbox” that provides numerical models and supporting information to improve evaluation of stream channels, design of restoration projects, and linkages between geomorphic design and ecological outcomes. The toolbox is intended to address the need for useful models of tested reliability and documented limitations that more immediately link research to practice. Toolbox programs are open source and address a wide range of practical problems, such as hydraulic geometry, transport and sorting in coarse-bedded streams, bed evolution below dams, and the delineation between threshold and alluvial channels. Although it is widely acknowledged that most predictions and design choices associated with stream restoration have large uncertainty, estimates of uncertainty are rarely incorporated into restoration design. We propose approaches by which uncertainty can be incorporated into stream restoration design and decision-making. Example toolbox applications are discussed in this paper. RESTORATION PRACTICE To be effective, research, models, and training must be developed in the context of current practice and understanding. Current stream restoration practice is based on analogy – a template is sought in a nearby channel, stream type, or hydraulic geometry relation that the designer judges to be suitable. The template channel is then scaled to the design site, typically using estimates of bankfull discharge (FISRWG, 1998). When scaling is accomplished via an estimated bankfull discharge, an implicit assumption of equilibrium is introduced: it is assumed that the disturbed channel is adjusting toward some ‘stable’ state and that evidence of this future condition can be found. But if a disturbed stream is adjusting to changes in essential controlling factors, particularly water and sediment supply, its future steady-state condition may not yet be evident. An appropriate template is unlikely to exist and would, in any case, be difficult to reliably demonstrate. The experience of designing stream channels when no suitable reference reach or identifiable bankfull discharge are available appears to be common (e.g. Sortman, 2004). Beyond its deficiencies in supporting channel design, an analogy approach does not efficiently support learning and cannot lead to true prediction because it provides no basis for linking cause and effect in a logically complete and testable framework. An analogy approach may have some practical use in particular cases for which the channel disturbance is not driven by essential changes in forcing, but by changes internal to the channel and reversible. The primary examples are exclusion of livestock from the riparian corridor and restoring a natural geometry to artificially straightened stream channels. If no substantial changes in water and sediment supply are anticipated, a basis exists for transplanting the geometry of a similar stable channel. What is the alternative to an analogy approach to stream restoration design? It must begin with specification of the materials and configuration of the stream valley and the water and sediment supply, including the variability and uncertainty in these quantities. In the essential next step, the specified conditions must be connected via predictive relations sufficient to link cause and effect. The predictive relations must satisfy general physical principles of mass, momentum, and energy conservation and will include empirical relations of demonstrated generality. As all seem to acknowledge, streams adjust to the water and sediment supplied to them. To this must be added essential feedbacks between the physical channel and the chemistry and biology of its waters, flora, and fauna which can influence details of transport processes and broad expression of channel geometry. A predictive, science-based approach to stream restoration must be built on these essential inputs. An ability to predict water and sediment supply is needed, as is an ability to predict the patterns of erosion and deposition within a design reach. Although existing methods invoke sediment transport in their designs, even calculate transport at some stages of the design process, only recently has a logically complete structure for predicting inputs and outcomes in stream channels emerged (Shields et al., 2003). There is good reason that an analogy approach dominates stream restoration practice: in most practical cases, current approaches do not permit predictions of sediment transport of sufficient accuracy to support channel design. This is particularly the case in gravel-bed streams, for which neither empirical nor theoretical approaches can provide suitable accuracy on a routine basis (Wilcock, 2001). A predictive approach requires that the future supply of sediment can be adequately forecast. With current technology, this is possible only for threshold channels. In this case, a precise estimate of sediment supply is not needed, it is only necessary to determine that the supply will be smaller than a critical threshold amount. One NCED Stream Restoration tool provides a basis for estimating this threshold. The primary challenges facing development of a predictive stream restoration science are: 1. Forecasting water and sediment supply. Stream channel change is driven by changes in water and sediment supply. Catastrophic failure of restoration projects can usually be attributed to a poor (or missing) estimate of the water and sediment supply. A reliable estimate of sediment supply is the essential threshold between analog and predictive design. 2. Variability and uncertainty. There is enormous uncertainty in virtually every aspect of channel design: historical trends, future forcing, and calculated water and sediment fluxes. Estimates of uncertainty in water and sediment supply are rarely made and incorporation of uncertainty in channel design is virtually absent. Ignoring uncertainty does not make its consequences disappear. 3. Watershed context. An adequate forecast of future water and sediment supply can only be done in a watershed context. NCED RESTORATION ACTIVITIES Research: NCED research addresses a wide range of erosion and sediment transport topics supporting improved restoration science and practice. Routing and supply of sediment: Channels develop in response to their water and sediment supply. An inability to predict sediment supply (mean and variability) is the primary technical barrier to predicting future channel configuration and composition. Development of predictive sediment supply relations will require a means of determining sediment storage throughout the stream network and a reliable treatment of sediment storage in reach-scale transport models. We are currently developing models for routing sand and fine gravel through coarse immobile beds (Dietrich, et al., 2005; Grams, et al., 2005). Our goals are to develop reach-averaged transport models incorporating storage dynamics and to test these models at the watershed scale in different transport environments. Transport, sorting, and morphodynamics of mixed-size sediment: The transport of streambed material drives channel change and the composition and configuration of the bed, which determines the essential, organism-scale template for the stream ecosystem. We now have in place surface-based transport models (Wilcock and Crowe, 2003) and a general framework for bed scour and aggradation (Parker, et al., 2000).
