Combining spatially balanced sampling, route optimisation and remote sensing to assess biodiversity response to reclamation practices on semi-arid well pads

ABSTRACT Biodiversity decline is widely considered a critical global environmental threat. In the western United States, land surface disturbance associated with oil and gas development is considered a top driver of habitat fragmentation and biodiversity decline. Land reclamation and ecosystem restoration activities help mitigate biodiversity loss, though monitoring practices to track these efforts are inconsistent, often lack measures of biodiversity response and are labour-intensive. Digital image analysis has been shown to reduce labour requirements and can provide robust, statistically valid reports on vegetation cover. We compare handheld image analysis to unmanned aerial system (UAS) image analysis to measure vegetation and ground cover on reclaimed oil and gas well pads. We utilise a spatially balanced sample design called balanced acceptance sampling along with a travelling salesperson algorithm to optimise walking and flight paths to obtain imagery in our study design. Images are then analysed with a free software program, ‘SamplePoint’, to classify vegetation on reclaimed well pads into functional groups. We conclude image acquisition is significantly faster with the UAS than with the handheld approach. We found that UAS image analysis and handheld analysis produced similar results in assessing vegetation and ground cover and we discuss pros and cons of each method. Key policy insights Rapid monitoring techniques which are statistically sound and provide robust datasets should help enhance knowledge of land reclamation practices in oil and gas fields. Unmanned aerial systems can cover well pads significantly faster than a human walking with a camera, and images gathered by each have similar results when vegetation is analyzed at the functional group level. Although slower, hand-held images may provide finer detail than UAS images flown 7.6 m above ground level, which may make hand-held images more useful for classifying vegetation to species-specific levels. Utilizing GPS technology along with spatially balanced sampling, route optimization, and digitial images increases speed of data collection and spatial accuracy of data compared to traditional line point intercept techniques.