Model for Forecasting Transportation-Related Wetland Impacts

The South Carolina Department of Transportation (SCDOT) develops near (3-5 years) and long (15-20 years) range plans for road widening, alignment, bridge replacement, and new road construction. Each road/bridge project may impact wetlands or streams through overburden or impairment. Most wetlands and streams are protected in the United States under Federal regulations (i.e. Clean Water Act). Enforcement of this protection is through the U.S. Army Corp of Engineers (USACE) and Environmental Protection Agency (EPA). Destruction of wetlands or impacts of wetlands/streams is permitted by the USACE for road/bridge projects if the transportation project is least environmentally damaging of options and if mitigation actions are created. Mitigation actions are in the form of restoring wetlands or creating wetlands in the ‘amount’ within the same watershed-ecoregion where the transportation project occurs. Transportation projects cannot begin until the wetland/stream impacts are known, and the USACE has agreed to the estimated impacts with an approved action for restoring/creating wetlands in like amount within the watershed-ecoregion. These linked actions often result in a very long delay (often years or even cancelling of projects) in a transportation project until the SCDOT has an approved plan for wetland-stream mitigation action. The mitigation action can only be anticipated based on the prediction of future impacts within a watershed-ecoregion. Thus, the mitigation forecasting problem has large geographic scale dimensions, with impacts at the site scale (i.e. road/bridge location), mitigation actions at the meso-scale (i.e. watershed-ecoregion) but with a very large geographic scope (i.e. the entire state of South Carolina). Solutions for the prediction of transportation improvement projects require fine-grained spatial analysis with high quality (accuracy and spatial precision) geographic data. As the scope of analysis is large (state of South Carolina) the supporting database of wetlands and stream locations is complex. Thus, the problem is also one of a ‘big-data’ variant with large computational demands with fine-grained spatial data. In this research we began with a national database of high spatial resolution (NWI, SSURGO) but with accuracy levels unacceptable for the forecasts. Using highly accuracy spatial/attribute wetlands data (referred to as the ‘jurisdictional wetland determination’) from already permitted transportation projects the accuracy of National Wetands Inventory (NWI) and (SSURGO)-based wetlands data was evaluated. Omission errors for the NWI and SSURGO data were 50% and 10%, respectively with high rates of commission errors with the SSURGO data. Subsequently, a high spatial resolution database from LiDAR-derived elevation and products, hydrography, culverts, parcel-level zoning/use, and historical maps/imagery was used to model the likelihood of wetlands and streams for the state of South Carolina. A wetlands likelihood model was developed to predict areas that will likely be classified as jurisdictional wetlands by the USACE. A GIS-based road widening and bridge replacement tool was created to model the existing and new wetland/stream impacts for each of the more than 300 future transportation projects with likely unavoidable impacts. Aggregate impacts of wetlands and streams were summarized at the watershed-ecoregion scale for prediction of future mitigation needs.