Troubling the Pixel
Date
2025-01-09
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Monitoring vegetation recovery in forested environments can be done effectively with satellite time series data. However, most research in this space has relied on Landsat imagery, which has a 30-meter pixel size. Emerging CubeSat technology has much finer spatial resolution and offers new possibilities for monitoring vegetation on small industrial disturbances such as well pads and seismic lines. These features are associated with hydrocarbon exploration and are some of the most extensive disturbances in Alberta’s forests. We used PlanetScope satellite imagery to classify vegetation on small industrial disturbance features across a 309-km2 study area in northern Alberta as either growing or declining, and validated our results with LiDAR data. We found that recovery classification on medium-sized well pad features was significantly better than for medium and narrow linear features. We also found that disturbances in some ecosite types (low-density treed wetland, upland mesic, and upland dry sites) were easier to classify than others (transitional and wetland treed) and that linear features oriented in a north-south direction were easier to classify than those oriented east-west. Our research demonstrated that 58% of pixels on disturbed linear features in our study area displayed some degree of spectral regeneration, alongside 76% of the evaluated pixels on wellpads. We believe that the privilege of working with this emergent imagery comes with a responsibility to think deeply on how we use, transform, and interpret the data. To investigate the intersection between landscape, technology, and people, we employed arts-based research methods to trouble the pixel – the fundamental data unit in our remote-sensing research. By creating a textile pixel and documenting it in our study area, we were able to reflect on the parallel methodologies at play in our work, and to identify practical interventions into our scientific practice that allows us to more deeply engage with our data. This includes providing care instructions for our scientific products, slowing down our processes with manual tasks, and dedicating time to witness our data in-situ. Our hope is that through sharing our experiments in investigating our positionality to remote sensing, others are encouraged to reflect on their own practices.
Description
Keywords
boreal forest, fiber art, site-specific art, satellites
Citation
Yeomans, T. M. (2025). Troubling the pixel (Master's thesis, University of Calgary, Calgary, Canada). Retrieved from https://prism.ucalgary.ca.