Use of Nested Flow Models and Interpolation Techniques for Science-Based Management of the Sheyenne National Grassland, North Dakota, USA

Authors


Corresponding author: Department of Hydrology, GNS Science, Lower Hutt 5010, New Zealand; (64) 4-570-4396; fax: (64) 4-570-4600; m.gusyev@gns.cri.nz

Abstract

Noxious weeds threaten the Sheyenne National Grassland (SNG) ecosystem and therefore herbicides have been used for control. To protect groundwater quality, the herbicide application is restricted to areas where the water table is less than 10 feet (3.05 m) below the ground surface in highly permeable soils, or less than 6 feet (1.83 m) below the ground surface in low permeable soils. A local MODFLOW model was extracted from a regional GFLOW analytic element model and used to develop depth-to-groundwater maps in the SNG that are representative for the particular time frame of herbicide applications. These maps are based on a modeled groundwater table and a digital elevation model (DEM). The accuracy of these depth-to-groundwater maps is enhanced by an artificial neural networks (ANNs) interpolation scheme that reduces residuals at 48 monitoring wells. The combination of groundwater modeling and ANN improved depth-to-groundwater maps, which in turn provided more informed decisions about where herbicides can or cannot be safely applied.

Ancillary