COMBINING NATIONAL WETLAND INVENTORY, LANDSAT, AND LIDAR TO MODEL THE WETLAND WATER STORAGE IN THE PRAIRIE POTHOLE REGION OF THE UNITED STATES

With an area around 715,000 km², the Prairie Pothole Region (PPR) of North America extends from north-central Iowa to central Alberta. The landscape of PPR is dotted with many small wetlands created during the last glacial retreat approximately 12,000 years ago. The region supports many ecosystem services including carbon sequestration, floodwater retention, waterfowl production, and pollution reduction. However, cultivated agriculture results in wetland drainage. Concern over the reduction of flood mitigation services historically provided by PPR wetlands has stimulated interest in developing spatially distributed hydrological models to simulate the effects of wetland water storage.

Many attempts have been made to model wetland water storage services of PPR wetlands; however, two main obstacles have reduced the value of these efforts. First, the availability of high resolution elevation data is usually lacking, and researchers have had to rely on Digital Elevation Models (DEMs) with resolutions up to 10 meters to model water storage; these coarse resolutions are usually inadequate to capture the relief of the region. Second, Nation Wetland Inventory (NWI) datasets, which are often used for wetland identification and water storage modeling in PPR, were derived principally from 1970–1980 photography; these NWI datasets are temporally static and do not reflect land cover changes caused by human management and climate fluctuations over the past two decades.

In our research, we used a decision tree model to classify a series of Landsat images (1989, 1991, 1997, 2001, 2003, 2004, 2005, and 2008) into water and no-water to capture the interannual dynamics of wetland surface water. Together with the Conservation Reserve Program (CRP) datasets, these Landsat products were composited to update the NWI dataset and compile a dataset of “current wetland distribution.” We also developed a bare earth DEM from Light Detection And Ranging (LIDAR) at a resolution of 0.5 meters. This DEM was used to delineate each wetland catchment area as well as the position and elevation of spill points. From each catchment and its spill point, we modeled wetland water storage. The maximum water storage of an area considered each individual wetland, wetland connectivity, and surrounding land cover. The model output is being compared to field survey data from the USGS’s Cottonwood Lake Study Area, ND, for accuracy assessment.