Watershed Scale Impacts of Floodplain Restoration on Nitrate Removal and the Practical Applications of Modeling Cumulative Floodplain Restoration Hydraulics

Human land use practices such as urbanization and agriculture contribute excess nutrients (nitrogen and phosphorus) and runoff volumes to rivers that degrade aquatic ecosystems and cause a loss of river functions such as nutrient processing and flood attenuation. Floodplain restoration increases floodplain exchange and is commonly implemented to improve water quality and reduce flood impacts at watershed scales. However, the effect of multiple restoration projects at the watershed scale is not well studied. We addressed this knowledge gap by two studies. The first study evaluated the impact of cumulative and spatially varying Stage-0 and bankfull floodplain restoration on nitrate removal in a generic 4th-order Virginia Piedmont watershed for small and sub-annual storm sizes (i.e. 2-year, 1-year, half-year, and monthly recurrence intervals). We used HEC-RAS hydraulics results from a prior study together with a nitrate removal model coded in R. Results indicated that watershed nitrate removal varied depending on the location of restoration in the watershed and where removal was evaluated. The greatest reductions in nitrate loads were observed in the same part of the river network where restoration occurred, with diminished impacts downstream. Removal also increased with increasing stream order/river size. However, removal was generally of small magnitude, with up to 1% or 19% of the watershed load removed for median or 90th-percentile removal rates, respectively. We estimated removal for our restoration scenarios under the Chesapeake Bay Program Protocols and found the removal rate to also be a critical factor in determining the efficiency of restoration project. Other controlling factors for nitrate removal were the amount of restoration and storm size. The second study entailed modeling cumulative restoration in a case study watershed to assess the impacts on nutrient removal and flood attenuation. We built a 1D HEC-RAS model of the 4th-order Gwynns Falls watershed near Baltimore MD using georeferenced HEC-RAS model geometries from the Maryland Department of the Environment and simulated unsteady stormflow hydraulics due to cumulative Stage-0 floodplain restoration for small and sub-annual storms. Restoration actually increased peak flow on the main channel (up to 0.9%) due to slowing of the flood wave on the main channel which was then better synchronized with tributary inflows. Restoration increased nitrate removal but at low levels (up to 0.12% or 2.6% removal for a median and 90th-percentile removal rate respectively) due to the small footprint of restoration in the watershed (up to 21.4% of the main channel was restored). These small and sometimes adverse outcomes occurred in response to what would be expensive restoration. Therefore, we argue for large-scale solutions to address watershed-scale water quality and flooding issues yet acknowledge re-evaluation of restoration goals against other societal priorities may be necessary. Overall, our results highlight the potential value and limitations of floodplain restoration in reducing flooding and nitrate exports at the channel network scale and provide practical insight for application of floodplain modeling at the watershed scale.