Meneguzzo, Dacia2021-10-132021-10-132020-08https://hdl.handle.net/11299/224946University of Minnesota Ph.D. dissertation. August 2020. Major: Natural Resources Science and Management. Advisor: Joseph Knight. 1 computer file (PDF); xi, 202 pages.Trees are an important resource in the Great Plains region of the United States yet little information describing their extent and location is readily available in formats that are convenient for resource professionals and decision makers. National forest inventory and natural resource monitoring programs seldom account for these non-traditional forests in their official statistics. In addition, most satellite-derived datasets are too coarse to accurately depict small or narrow groupings of trees common in the Great Plains. As a result, there is a lack of scale-appropriate data for inventory and monitoring of these tree resources. Methods are needed to conduct large-scale comprehensive assessments of tree cover in the Great Plains. Remote sensing-based approaches offer several advantages over ground based inventories because they are often cost effective, they alleviate access issues, and they provide wall-to-wall spatial coverage. The research presented here will demonstrate that tree cover can be mapped at a statewide level using an object-based image analysis (OBIA) approach and high-resolution (i.e., 1 m) digital aerial photography from the National Agriculture Imagery Program (NAIP) as the sole data source. Initial results indicated that the OBIA method was more accurate in terms of describing the actual observed spatial pattern of tree cover and produced a more realistic output product compared to a pixel-based classification method. Next, technological improvements were made to the OBIA method to make it more robust for operational land cover mapping at a regional level. Lastly, a shape-based classification approach was developed for positively identifying various configurations of windbreaks (both single and multiple-leg) from the output land cover maps, which is an improvement over existing methods that only map single-leg windbreaks. This is important for management purposes since windbreaks provide many ecological and economic benefits on the landscape, from conserving topsoil to protecting crops, livestock, and farmsteads from the harsh effects of wind. The outcomes of this research are actual published (or in the process of) high-resolution geospatial data products that are publicly available for download. These datasets identify and provide detailed spatial information about mapped tree cover and windbreaks that can be summarized at a variety of scales, from individual farms to the state or regional level. In addition, they are valuable for many different types of research studies and on-the-ground management activities. In a region of climate extremes, the hope is that these datasets will support informed decision making for placing trees in the right place on the landscape to maximize the benefits they can provide. For example, one of the goals in this region is windbreak establishment in areas with highly erodible soils that lack trees arranged as windbreaks. These maps will assist with such planting efforts as stated by Darci Paull, a GIS technician with Kansas Forest Service, “If we know where windbreaks are, then we know where they aren’t. Combining this information with other spatial information, for example, highly erodible soils data, we can identify at-risk soils that would benefit from the protection of a windbreak.”enagroforestrygeographic object-based image analysisGISland cover mappingtree coverwindbreaksThesis or Dissertation