Automated Geomorphic Landform Extraction Using a LiDAR Derived Dataset in a Low Relief Alluvial Landscape
Academic Level at Time of Presentation
Graduate
Major
Geosciences/Watershed Studies
List all Project Mentors & Advisor(s)
Haluk Cetin, Ph.D.; Gary Stinchcomb, Ph.D.
Presentation Format
Poster Presentation
Abstract/Description
Digital terrain models derived from high-resolution light-detection and ranging (LiDAR) point cloud data are powerful tools capable of geomorphic landscape identification and quantification. Early geomorphic landform studies were conducted using “on the ground” measurement techniques where accuracy and precision were determined by worker experience level. More recent geomorphic landform studies incorporated the use of remotely sensed pixel-based data to quantify ground surface dimensions. However, human interaction continues to play a large role in most decision-based models. This study examined the efficacy of using an Object-based Image Analysis (OBIA) to extract discrete geomorphic landforms, and landform dimensions, in a low relief alluvial landscape on the Tennessee River in North Alabama. OBIA offers a quick and easily repeatable workflow where pixel-based data (elevation and spatial relationships) are automatically grouped into vector objects, thereby limiting human decision making and associated error. Results indicated that OBIA produces reliable and reproducible output in low relief landscapes and offers more precise and accurate estimates of cross-sectional information compared to digital elevation models for this study area. Furthermore, the data showed that geomorphic landform identification varies as a function of spatial sampling extent. Lower relief landscapes may require a larger areal sample to accurately identify subtle geomorphic landforms despite utilization of high-resolution LiDAR.
Fall Scholars Week 2018 Event
Earth and Environmental Sciences Poster Session
Automated Geomorphic Landform Extraction Using a LiDAR Derived Dataset in a Low Relief Alluvial Landscape
Digital terrain models derived from high-resolution light-detection and ranging (LiDAR) point cloud data are powerful tools capable of geomorphic landscape identification and quantification. Early geomorphic landform studies were conducted using “on the ground” measurement techniques where accuracy and precision were determined by worker experience level. More recent geomorphic landform studies incorporated the use of remotely sensed pixel-based data to quantify ground surface dimensions. However, human interaction continues to play a large role in most decision-based models. This study examined the efficacy of using an Object-based Image Analysis (OBIA) to extract discrete geomorphic landforms, and landform dimensions, in a low relief alluvial landscape on the Tennessee River in North Alabama. OBIA offers a quick and easily repeatable workflow where pixel-based data (elevation and spatial relationships) are automatically grouped into vector objects, thereby limiting human decision making and associated error. Results indicated that OBIA produces reliable and reproducible output in low relief landscapes and offers more precise and accurate estimates of cross-sectional information compared to digital elevation models for this study area. Furthermore, the data showed that geomorphic landform identification varies as a function of spatial sampling extent. Lower relief landscapes may require a larger areal sample to accurately identify subtle geomorphic landforms despite utilization of high-resolution LiDAR.