Characterizing Woody Encroachment in the Konza Prairie Using Object-Based Analysis of Aerial Photographs

Abstract

Woody encroachment is a threat to the ecological integrity of tallgrass prairie in Kansas. Encroachment data that covers a large spatial and temporal scale would be valuable to managers of tallgrass prairie, but no such dataset exists. The objective of this research is to develop a replicable technique for creating woody vegetation maps from aerial photographs. Rather than using a traditional pixel-by-pixel approach to classification, this project uses an object-based approach, wherein individual pixels are grouped into meaningful image objects according to user-defined parameters and then classified. I created woody vegetation maps of eight watersheds in the Konza Prairie using imagery from 1978, 1991, 2003, 2006, and 2010. I also determined the efficacy of LIDAR data in classifying the image from 2006. Ground-based vegetation survey data exist for two of the watersheds included in the remote sensing portion of this study. I analyzed the data from the available years nearest to the imagery dates (1983, 1992, 2003, and 2007) in order to provide a measure of validation for the woody vegetation maps. The results of this research were used to determine the applicability of this mapping technique and to draw preliminary conclusions about the landscape-level factors associated with woody encroachment.