Incorporating Affects of Raised Roads into Hydrology Model to Improve Simulation of Low-Relief Watershed

For a watershed with low topographic relief, its inclusive raised roads can considerably affect the hydrology by altering runoff paths, facilitating runoff flows in some areas but impeding flows in others, and modifying water balance components of overland runoff, infiltration, and evapotranspiration. However, these affects cannot be considered when using the commonly-used deterministic eight-neighbor (D8) algorithm to determine overland flow directions, and thus can rarely be taken into account by the currently available simulation models. As a result, the overland runoff could be erroneously predicted to overtop raised roads such as interstate highways. Because raised roads are usually scattered across the watershed with a large total length, a simulation without taking these affects into consideration is likely to generate misleading information of design discharge and nonpoint pollution sources. A decision based on this information probably results in failure of engineering projects for water resources and water quality improvement. The objectives of this study were to: 1) develop an approach to correct D8 overland flow directions using geometric data of raised roads and location data of the associated culverts; and 2) verify this approach by evaluating the simulation performance of Soil and Water Assessment Tool (SWAT) as influenced by raised roads. The evaluation was conducted for the Wild Rice River watershed, which has a very low topographic relief and is located in northwestern Minnesota. For this purpose, one SWAT model was set up based on the D8 overland flow directions and another SWAT model was set up based on the flow directions corrected using the correction approach developed in this study. These two models were calibrated to have a compatible performance in terms of reproducing the daily streamflows observed at two U.S. Geological Survey gauging stations. The performance differences were examined in terms of reproducing the daily streamflows observed at these two same stations for the validation period as well as the water balance components and runoff contribution spatial patterns. The results indicated a better simulation performance of the SWAT model that took into account the affects of raised roads. The model based on the corrected overland flow directions can more accurately mimic the physical hydrologic processes within the watershed.