Watershed-scale Phosphorus Inputs from Streambanks

Issue

Streambanks were identified in the Iowa Nutrient Reduction Strategy (INRS) as a potentially important but poorly quantified source of phosphorus (P) to surface waters. New research from Iowa and other Midwestern states has quantified streambank contributions, with results varying widely. Most of these efforts have focused on a small number of eroding banks in relatively small watersheds. Few of these studies have extended for more than five years and may therefore not be representative of the full range of inter-annual variability in hydrology and P contributions. Recent projects funded by INRC and other sources have supported long-term monitoring of the extent and magnitude of eroding banks on Onion Creek in central Iowa, which has been identified as a potentially large contributor to the sediment and P loads of Squaw Creek. Results from the Onion Creek studies found major variation over time, highlighting the importance of long-term monitoring to understand the causes of variability in P sources, dominant P forms and rates of P export. A related INRC-funded effort that is nearing completion addresses the need to make estimates of streambank-derived P scalable to larger areas of the state, as the field-intensive methods that are most often used for quantifying bank erosion rates and bank-derived P are impractical in watersheds much larger than 100 square kilometers. Methods that make use of high-resolution geospatial data are more efficient at this scale. A widely-used technique at this scale is to digitize streambanks from aerial photos in sequential years to estimate bank migration rates. When combined with high resolution digital elevation data or a sample of bank heights from field measurements and estimates of P concentrations in bank materials, P contributions can be estimated on these larger scales. This approach still takes significant time and involves some subjectivity in bank delineation.

To address these limitations, we have developed the Aerial Imagery Migration Model (AIMM), which uses a band index from high-resolution multispectral aerial imagery to automate detection of streambanks and acquires bank heights from LiDAR-derived elevation data. In initial testing, AIMM compares favorably with manual bank digitization in accuracy, is extremely efficient and eliminates the subjectivity of manual methods. However, testing of AIMM has been very limited. Exploration of the potential of AIMM to address additional reach- or basin-scale management problems also remains to be addressed.

Objective

This research will extend recent work identified above to explore questions related to the use of monitoring data to evaluate practice effectiveness. Since AIMM enables efficient detection of bank erosion and deposition over large areas, it has the potential to provide valuable insights into the spatial distribution of problem areas and to improve management decisions targeting areas of greater erosion risk.

The first objective is to continue and extend ongoing work in the Onion Creek watershed that began in 2011. Periodic eroding-length assessments and erosion pin measurements will continue along with continued manual and automated water sampling near the watershed outlet. The collection of additional data will enrich the existing dataset and address new questions that leverage the diversity of hydrological conditions encountered in the long-term study.  

The second objective is to extend the application of AIMM beyond the initial test watersheds to the entire state. In this effort, we will compare AIMM estimates of erosion and deposition to those obtained by the Iowa DNR using manual bank delineations in a subset of watersheds around Iowa. This comparison will allow a test of the sensitivity of AIMM to variables that differ among landform regions. AIMM will then be used in combination with soil P-concentrations estimated from GIS layers to estimate P contributions from streambanks for each HUC-10 watershed in the state. In select sites where the data is available, these estimates will be compared with annual total P export derived from continuous monitoring at gaging stations.

A third objective of this research is to use AIMM and manual bank delineation to evaluate the effectiveness of bank stabilization and flow-training structures in reducing streambank contributions of P at a multi-reach scale.

Approach

Field measurements in Onion Creek will continue to use the same methods established in previous work. A network of banks selected randomly from a complete set of eroding banks identified in a full watershed survey has been set up for erosion pin measurements at horizontal intervals along the bank. Water samples will be collected, sediment concentration determined by filtration and dissolved and total P concentrations determined by photometric and persulfate digestion methods, respectively.

For Objective 3, research will focus on a portion of southwest Iowa’s West Nishnabotna River, where eight streambank stabilization projects, totaling approximately 6,000 linear feet, have been completed since 2012 on a 16-mile stretch. We will use AIMM to estimate erosion and deposition in this reach, compared with 16-mile reaches upstream and downstream that have had comparatively little streambank stabilization. Geospatial analysis for the second and third objectives will be performed in the Applied Geomorphology lab at Iowa State University. Bank erosion estimates from manual digitization will employ an existing dataset available from partners at the Iowa DNR that is coincident with the 2007-2010 LiDAR data. Additional digitization will be completed for a second, more recent year where high resolution imagery is available.

Results of this research will be combined with those of a collaborative INRC-proposed project, Linking Agricultural Practices to Water Quality Improvement: The Importance of Scale in Accurately Characterizing N and P Loads Delivered to Iowa Streams (Crumpton et al.). The combined dataset will provide insight as to the effect of watershed scale on the relative contribution of upland versus in-stream sources of P.

Results of this research will be disseminated in multiple forms, including for scientific journals and Extension publications and presentations at appropriate regional or national academic conferences and workshops for watershed managers and engineers. The team also plans to make AIMM and supporting documentation available as an open-source ArcPy script and an ArcGIS plugin.