Viewshed uncertainty in a forested, mountainous landscape
MetadataShow full item record
Visibility analysis is critical in a variety of applications, and many geographic information systems (GIS) are equipped with analysis functions, such as the viewshed, that provide visibility results based on digital elevation data. These GIS analyses are affected by uncertainty associated with input data and software systems, which limit the reliability of results. This thesis investigates numerous factors (different software and algorithms, elevation data models and spatial resolutions, observer positional errors, visualization approaches, and earth curvature) that can impact GIS-based visibility analysis in mountainous terrain. The results indicate that a raster digital elevation model (DEM) is a better data structure for visibility analyses than a triangulated irregular network (TIN), and higher resolution data tend to generate less variability and better agreement. The algorithms tested normally have high levels of agreement, but discrepancies can exist. There is a need for incorporating surface features into analyses and for field validation of results.