Mapping Woody Plant Encroachment in Grassland Using Multiple Source Remote Sensing images: Case Study in Oklahoma

Abstract:

Woody plant encroachment (mainly Juniperus virginiana, a coniferous evergreen tree) in the native grassland has been rapidly increasing in the U.S. Southern Great Plains, largely triggered by overgrazing domestic livestock, fire suppression, and changing rainfall regimes. Increasing dense woody plants have significant implications for local grassland ecosystem dynamics, such as carbon storage, soil nutrient availability, herbaceous forage production, livestock, watershed hydrology and wildlife habitats. However, very limited data are available about the spatio-temporal dynamics of woody plant encroachment to the native grassland at regional scale. Data from remotes sensing could potentially provide relevant information and improve the conversion of native grassland to woody plant encroachment. Previous studies on woody detection in grassland mainly conducted at rangeland scale using airborne or high resolution images, which is sufficient to monitor the dynamics of woody plant encroachment in local grassland. This study examined the potential of medium resolution images to detect the woody encroachment in tallgrass prairie. We selected Cleveland county, Oklahoma, US. as case study area, where eastern area has higher woody coverage than does the western area. A 25-m Phased Array Type L-band Synthetic Aperture Radar (PALSAR, N36W98) image was used to map the trees distributed in the grassland. Then, maximum enhanced vegetation index (EVI) and normalized difference vegetation index (NDVI) in the winter calculated from time-series Landsat images was used to identify the invaded woody species (Juniperus virginiana) through phenology-based algorithm. The resulting woody plant encroachment map was compared with the results extracted from the high resolution images provided by the National Agriculture Imagery Program (NAIP). Field photos were also used to validate the accuracy. These results showed that integrating PALSAR and Landsat had good performance to identify the woody encroachment in the study area. This study demonstrates the potential to monitor the dynamics of dense woody plant encroachment at the region scale using PALSAR and Landsat images and improves our understanding about the spatio-temporal dynamics of woody plant encroachment to native grasslands.