Skip to main content
ARS Home » Plains Area » Las Cruces, New Mexico » Range Management Research » Research » Publications at this Location » Publication #270832
Title: Using very-large-scale aerial imagery for rangeland monitoring and assessment: Some statistical considerations

Author
item Karl, Jason
item DUNIWAY, MICHAEL - Us Geological Survey (USGS)
item NUSSER, SARAH - Iowa State University
item OPSOMER, JEAN - Colorado State University
item UNNASCH, ROBERT - Nature Conservancy

Submitted to: Rangeland Ecology and Management
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 3/1/2012
Publication Date: 7/1/2012
Citation: Karl, J.W., Duniway, M.C., Nusser, S.M., Opsomer, J.D., Unnasch, R.S. 2012. Using very-large-scale aerial imagery for rangeland monitoring and assessment: Some statistical considerations. Rangeland Ecology and Management. 65(4):330-339.

Interpretive Summary: The availability of very-high-resolution (VHR) imagery and techniques for processing those data into indicators of ecosystem function has opened the door for VHR imagery to be used in rangeland monitoring and assessment. However, VHR imagery can be expensive and, like any survey measurement, studies that rely on VHR-image-based indicators will be subject survey errors that lead to biased results or inflated standard errors. To yield defensible monitoring and assessment data for managers, indicators derived from VHR imagery must be implemented with a statistically valid sampling design. We discuss the use of VHR imagery from the perspectives of achieving unbiased estimates with minimal survey errors in the context of rangeland monitoring and assessment, and propose practical ways to conduct statistically-sound rangeland monitoring and assessment using VHR imagery. Technologies and techniques for acquiring, processing, and extracting information from VHR images are rapidly evolving. For this reason, it is of paramount importance to understand the statistical implications of designing and implementing a rangeland monitoring and assessment program based on VHR imagery, and the aspects of designing VHR-imagery studies that are different than traditional field-based studies like image scale, sample design limitations, and the need for validation of estimates.

Technical Abstract: The availability of very-high-resolution (VHR) imagery and techniques for processing those data into indicators of ecosystem function has opened the door for VHR imagery to be used in rangeland monitoring and assessment. However, VHR imagery can be expensive and, like any survey measurement, studies that rely on VHR-image-based indicators will be subject survey errors that lead to biased results or inflated standard errors. To yield defensible monitoring and assessment data for managers, indicators derived from VHR imagery must be implemented with a statistically valid sampling design. We discuss the use of VHR imagery from the perspectives of achieving unbiased estimates with minimal survey errors in the context of rangeland monitoring and assessment. Our discussion includes study design considerations such as the importance of probability-based sampling for generating credible estimates; measurement errors in VHR-derived indicators; and model-assisted estimation to integrate data from multiple sources, improve precision and adjust for sources of errors. We demonstrate the effects of these design considerations using example datasets where VHR imagery and concurrent field data were collected. Finally, we propose practical ways to conduct statistically-sound rangeland monitoring and assessment using VHR imagery.