Murray State Theses and Dissertations

Abstract

An up-to-date and detailed vegetation map provides critical information for habitat management. In addition, a habitat model is necessary for Park’s Fire Management, for classification of fuel types, and for delineation of fire management units. Several attempts to map the vegetation at the Mammoth Cave National Park were conducted in 1934, 1975, 1997 and 2011. This essential goal of this study was to produce a new vegetation habitat model and update the vegetation map for the Park. Landsat-8 Operational Land Imager (OLI) imagery, LiDAR and bedrock dataset were used for habitat model configuration and vegetation mapping at Mammoth Cave National Park.

Vegetation habitat types were determined by a combination of slope, aspect and bedrock. The habitat model indicated that Acid and Calcareous were the two dominant habitats within the park, accounting for 46.24% and 49.74% of the total park area respectively. Among the ten habitat types, Acid Mesic and Calcareous Sub-Mesic occupied the largest areas, which accounted for 29.26% and 21.03% respectively. The habitat was observed and described at 29 ground reference sites due to limited accessibility. The habitat types of 22 sites (76%) predicted by the model were consistent with field observations. The discrepancy between model result and field observation at three sites was likely due to previous human disturbance. And the model needs further improvement to accurately predict Acid Xeric habitat locations.

Principal Component Analysis (PCA), enhanced vegetation index (EVI), and unsupervised classification, were applied to map the vegetation types. Five classes were mapped: barren land/ man-made structure, evergreen, deciduous, mixed forests and water. In the resultant map, deciduous trees accounted for the largest area in the park and most of the evergreen and mixed trees were found in the southern part of the park. The classification results were evaluated by 398 deciduous, 76 evergreen and 65 mixed field plots data. The overall accuracy of PCA technique and EVI Index was 85%, 7% higher than using PCA technique alone and 13% higher than NLCD 2011.

The influence of historic disturbance which occurred before the establishment of the Park can still be seen today. Approximately 70% of the evergreen forests, dominated by eastern red cedar (Juniperus virginiana) are found in previously cropland and pasture fields. They are the first successional forest in the area. While 40% of coniferous trees are currently in Xeric or Sub-Xeric habitat types with favorable conditions to support coniferous species, the remaining 60% will likely to be replaced by deciduous trees in the future.

Year manuscript completed

2017

Year degree awarded

2017

Degree Awarded

Master of Science

Department

Geosciences

College/School

Jesse D. Jones College of Science, Engineering and Technology

Thesis Advisor

Robin Q. Zhang

Committee Chair

Robin Q. Zhang

Committee Member

Bassil El Masri

Committee Member

Jane Benson

Committee Member

Kate S. He

Document Type

Thesis

Download

Share

COinS