Author
HULET, APRIL | |
ROUNDY, BRUCE - BRIGHAM YOUNG UNIVERSITY | |
PETERSEN, STEVEN - BRIGHAM YOUNG UNIVERSITY | |
JENSEN, RYAN - BRIGHAM YOUNG UNIVERSITY | |
BUNTING, STEPHEN - UNIVERSITY OF IDAHO |
Submitted to: Rangeland Ecology and Management
Publication Type: Peer Reviewed Journal Publication Acceptance Date: 3/10/2014 Publication Date: 5/1/2014 Citation: Hulet, A., Roundy, B.A., Petersen, S.L., Jensen, R.R., Bunting, S.C. 2014. Cover estimations using object-based image analysis rule sets developed across multiple scales in pinyon-juniper woodlands. Rangeland Ecology and Management. 67(3):318-327. DOI: 10.2111/REM-D-12-00154.1. Interpretive Summary: Numerous studies have been conducted that evaluate the utility of remote sensing for monitoring and assessing vegetation and ground cover to support land management decisions and complement ground-measurements however, few land cover comparisons have been made using high-resolution imagery and object-based image analysis (OBIA) to evaluate rule-sets (models) across multiple spatial scales. Our primary objective was to test the accuracy of OBIA rule-sets developed using eCognition Developer that estimate cover measurements from high-spatial resolution imagery (0.06-m pixel), relative to ground based measurements on P-J expansion woodlands at four spatial scales: 1) individual 30 X 33-m subplots, 2) individual sites ranging from 5-24 hectares, 3) regions or western juniper sites versus Utah juniper sites, and 4) all P-J woodlands sites that span across the Great Basin. Correlations between OBIA and ground measurements were relatively high for individual subplots and site spatial scales (ranging from r = 0.52 to r = 0.98); correlations for regional and Great Basin spatial scales were lower (ranging from r = 0.24 to r = 0.63) which was expected due to reflectance differences within the imagery as well as vegetation differences found at each site. The trade-off for decreased accuracy over a larger area (region and network scale) may be useful to prioritize fuel-management strategies but will unlikely capture subtle shifts in understory plant communities that site and subplot spatial scales often capture. Technical Abstract: Numerous studies have been conducted that evaluate the utility of remote sensing for monitoring and assessing vegetation and ground cover to support land management decisions and complement ground-measurements. However, few land cover comparisons have been made using high-resolution imagery and object-based image analysis (OBIA) to evaluate rule-sets (models) across multiple spatial scales. In this study, we investigate the accuracy of OBIA rule-sets developed using eCognition Developer that estimate cover measurements from high-spatial resolution imagery (0.06-m pixel), relative to ground based measurements on Pinus L. (pinyon) and Juniperus L. (juniper) woodlands. Rule-sets were evaluated at four spatial scales: 1) individual 30 X 33-m subplots, 2) individual sites (5-20 hectares), 3) regions (western juniper vs. Utah juniper), and 4) pinyon-juniper woodland across the Great Basin. Color-infrared imagery was acquired over five sites in Oregon, California, Nevada, and Utah with a Vexcel UltraCamX digital camera in June 2009. Ground cover measurements were also collected at study sites in 2009 on 80, 0.1-hectare subplots. Correlations between OBIA and ground measurements were relatively high for individual subplots and site spatial scales (ranging from r = 0.52 to r = 0.98). Correlations for regional and Great Basin spatial scales were lower (ranging from r = 0.24 to r = 0.63) which was expected due to reflectance differences within the imagery as well as vegetation differences found at each site. All site and subplot OBIA average cover estimates for live trees, shrubs, perennial herbaceous vegetation, litter, and bare ground were within 5% of the ground measurements, and all region and Great Basin OBIA average cover estimates were within 10%. The trade-off for decreased accuracy over a larger area (region and network scale) may be useful to prioritize fuel-management strategies but will unlikely capture subtle shifts in understory plant communities that site and subplot spatial scales often capture. |