Project Implementation Steps

Step 1: Developing the Model

In this milestone, the team collected, processed, and analyzed multiple datasets. The datasets describe current and projected precipitation, streamflow from specific watersheds, culverts, terrain, landcover, soils, flood risk, and much more.

Forest Service engineering personnel from the White Mountain National Forest in New Hampshire, where Hubbard Brook Experimental Forest is located, helped select U.S. Geological Survey stream gauge locations within and near the White Mountain National Forest boundary, and the Hubbard Brook Experimental Forest provided culvert data and field surveyed conditions of selected culverts. Similarly, Forest Service engineering personnel for the Francis Marion National Forest helped provide details of culvert data and assisted in field surveys of selected culverts for the Santee Experimental Forest.

Photo Credit

Rick Wrenn, USDA Forest Service (retired)

Flooded road and bridge after an extreme precipitation event in 2015 at the Santee Experimental Forest, SC. Agency roads are important to nearby community residents, visitors to national forests, partners, and employees. 

Phase 1 is completed. The team’s Phase 1 activities included:

• Collect and process culvert data, geospatial and high-resolution hydro-meteorological data, and LiDAR for both sites.
• With geospatial technology and LiDAR, identify additional potential culverts and other road-stream crossings on sites where culvert datasets are not available.
• Analyze current precipitation and downscaled future precipitation projections for both sites. Current/historic data come from Forest Service long-term on-site gauges as well as the National Oceanic and Atmospheric Administration and the future projections from Environmental Protection Agency.
• Assess risks of culvert failure due to flooding and erosion using multiple hydrology and geospatial models. Using hydro-met data and site characteristics in the models, the team assessed the range of flood discharge and sediment export potentials causing overtopping (undersized culverts) and siltation of culverts, respectively. The modeling assessment of vulnerable culverts were also evaluated with on-site data, where available, and validated with expert judgement.
• The modeling procedure included non-stationary regional frequency analysis with various scenario assessments, uncertainty analysis, SCS Graphical Peak Discharge method, Rational method, Stream-Bank-Erosion Vulnerability Assessment (SBEVA), Revised Universal Soil Loss Equation (RUSLE), Watershed Debris Flow Model (WDFM), and watershed erosion prediction project (WEPP).
• Used future climate model projections based on downscaled high resolution precipitation intensity information, as an option for engineers/managers for climate resilient culvert sizing, provided by the Environmental Protection Agency under a memorandum of understanding with the USDA Forest Service.
• Assess culvert failure risk due to flooding by comparing their estimated hydraulic capacities to projected design discharges for a flood frequency (25-yr, 50-yr or 100-yr etc.) of interest and propose safe sizings of culverts.
• Assess culverts for their risk of failure due to stream-bank erosion, siltation, and other hydro-geomorphological phenomena driven by extreme precipitation and flooding.
• Develop new and/or modify/apply existing erosion prediction models to assess pre-fire flooding and associated debris flow.
• Publish the papers on hydrological and hydro-geomorphological risk assessment model applications.
• Create a hands-on Toolbox (Geospatial Models) for easy dissemination and replication.

Photo Credit

Karl Buchholz, USDA Forest Service

A damaged culvert in a forest. 

Step 2: Developing the Decision Support System (DSS)

This phase built on an open access geographic information system (GIS) and is complete. The project team built the two types of web applications that comprise the tool: a static interactive dashboard that showcases results of culvert vulnerability assessments stored in the server, and a dynamic web-application. Both are available at culvert-at-risk.org. An alternative using an ArcGIS based DSS tool is under construction.

The version 1 of the dynamic web-application will let users generate culvert vulnerability assessments whether or not users have their own culvert geodatabase. When culvert data is unavailable the application will dynamically identify potential culvert locations from road and stream intersections. The web application will also allow users to save the location of these potential culvert locations to their own geodatabase.

The team’s activities in Phase 2 included: 

• Design and develop the DSS user interface with the open-source libraries.
• Integrate multiple vulnerability assessment and flood discharge models into the DSS.
• Develop functionalities for data input, visualization, and output reporting.
• Conduct beta testing and incorporate user feedback for refinement.
• Finalize, test and deploy the tool on University of Idaho server.
• Ensure the tool is secure, fast, and runs smoothly for all users.
• Develop a user guide/manual and tutorials. 

Phase 3: Training and Evaluation

In this phase, the team is monitoring the effectiveness of the tool in real-world applications and preparing a final report summarizing project outcomes and recommendations for future improvements. 

The team is also conducting workshops on how to use the web application and interpret the results. Recent workshops and demonstrations introduced the tool to land managers, engineers, and hydrologists on the Chattahoochee-Oconee National Forest in Georgia and Francis Marion & Sumter National Forests in South Carolina. 

Additional workshops will be designed for experts at the Forest Service, Department of Transportation, Environmental Protection Agency, The Department of Interior’s National Park Service and U.S. Geological Survey, and other interested collaborators and stakeholders including Weyerhaeuser.