Protocols for vegetation and habitat monitoring with unmanned aerial vehicles: Linking research to management on US public lands

Background/Question/Methods

Monitoring of the condition and trend of natural resources is critical for determining effectiveness of management actions and understanding ecosystem responses to broad-scale processes like climate change. While broad-scale remote sensing has generally improved the ability to monitoring ecosystems, many important fine-scale indicators must still be measured in situ through direct observation or via very-high-resolution (i.e., < 10cm ground-sampling distance [GSD]) imaging. Recent research has shown that many field-derived indicators (e.g., cover, density, vegetation structure, canopy gaps) of ecosystems can be obtained via very-high-resolution (VHR) aerial photography or digital photogrammetric techniques. The relaxing of regulations on operating unmanned aerial systems (UAS) in national airspace in the United States has turned UAS into a viable platform for acquiring VHR imagery for natural resource monitoring programs. However, while much research has looked at using VHR imagery to estimate ecosystem indicators, the available technologies and techniques have not previously been synthesized into a protocol for integrating UAS-based image collection and analysis into a national monitoring program. Our objective was to create and test a multiple-indicator sampling protocol for UAS-based data collection that could be integrated into the Bureau of Land Management’s Assessment, Inventory, and Monitoring program.

Results/Conclusions

Our proposed protocol uses a double-sampling approach where statistical relationships between UAS-derived and field-measured indicators collected at the same locations are used to make indicator estimates to areas where only UAS images are collected. The protocol takes advantage of the UAS platform’s ability to collect data quickly over field plot areas at GSD < 1cm. Our protocol calls for use of a multi-rotor UAS (e.g., a “quadcopter”) to collect downward-looking (i.e., nadir) images with 60% forward overlap and 30% sidelap augmented by oblique (i.e., 45º camera angle) images. The nadir images are used to assess, via image interpretation, vegetation cover by functional group, bare ground amount, and canopy gaps. The overlapping nadir and oblique images are used to create three-dimensional representations of the plot area to assess soil erosion, hydrologic connectivity, and vegetation height. We tested this protocol in Chihuahuan Desert shrubland sites and found a high degree of correspondence between field measurements and UAS-derived indicator estimates. The ability to realize the potential of UAS technologies for advancing monitoring of natural resources will depend not only on research verifying VHR imagery can be used to derive meaningful ecosystem indicators but also on being able to work these technologies into implementable protocols for monitoring programs.