Stream Restoration: Project Evaluation and Site Selection in the Cacapon River Watershed, West Virginia

Stream restoration is being conducted throughout the world at unprecedented rates to address stream channel degradation and water quality concerns. Natural Channel Design (NCD) is a common method used for restoration and has received governmental endorsement; however, the effects of NCD on channel stability and ecosystem functioning are poorly studied. We examined the effects of a reach-scale NCD project on channel stability, riparian vegetation, and water quality along the Cacapon River, West Virginia using a before-after-control-impact design and determined that restoration increased the abundance and diversity of woody vegetation, but had minimal effects on streambank stability and water quality. Increased erosion rates in some portions of the restored reach were attributed to differences in pre-restoration stability, vegetation removal, and soil composition among sub-reaches. No differences in in-stream concentrations of total phosphorus, nitrates, ammonia, or total suspended solids were detected following restoration; however, in-stream turbidity was drastically increased during construction. This study is a clear example of the value of monitoring streambank migration, vegetation communities, and soils to evaluate the effects of stream restoration and to provide insight on potential reasons for treatment failure. Ideally, pre-restoration monitoring should be used to inform project design by determining restoration potential of areas selected for restoration. As a surrogate for process monitoring, we created a maximum entropy model of streambank erosion potential (SEP) in a Geographic Information System (GIS) framework to prioritize sites for management and to determine which variables in the watershed are associated with excessive rates of streambank erosion. Model development included measuring erosion rates throughout a central Appalachian watershed, application of a quantitative approach to locate target areas for management termed Target Eroding Areas (TEAs), and collection of environmental data throughout the study extent using high resolution, remotely sensed data. A likelihood distribution of TEAs from occurrence records and associated environmental variables over our study extent was constructed using the program Maxent. All model validation procedures indicated that the model was an excellent predictor of TEAs, and that the major environmental variables controlling these processes were streambank slope, soil characteristics, shear stress, underlying geology, and riparian vegetation. A classification scheme with low, moderate, and high levels of erosion potential derived from logistic model output was able to differentiate sites with low erosion potential from sites with moderate and high erosion potential. This type of modeling framework can be used in any watershed to address uncertainty in stream restoration planning and practice. To address the need for accurate, high resolution estimation of streambank erosion, we also explored the role of laser scanning for estimating streambank migration and volumetric sediment loss. This was accomplished by comparing estimates of streambank migration and volumetric sediment loss derived from repeated erosion pin, streambank profile, and combined airborne and terrestrial light detection and ranging (LiDAR) surveys. Results indicated that LiDAR derived estimates were larger and highly variable compared to estimates derived from erosion pin and streambank profile surveys, which more accurately represented change along the study reach. Inflated LiDAR estimates were most likely the result of combining high resolution terrestrial LiDAR with relatively low resolution airborne LiDAR that could not effectively capture topographic features such as undercut banks. Although cost-prohibitive in some cases, repeated terrestrial LiDAR scans would likely circumvent these issues with higher point densities and better scan angles facilitating more accurate representation of streambank geometry, ultimately providing more accurate estimates of channel change