Piper Steffen
MSc Candidate in Geography, University of Victoria
BASc Engineering Physics, Queen's University
Contact
Spectral Lab Research:
I am part of the Satellite-based Kelp Mapping project (SKeMa) which will develop a software framework for First Nation communities to leverage satellite imagery (Sentinel-2) and machine learning for management of kelp forests in their traditional territories. Detecting kelp forests in imagery often relies on accurate spatial and temporal delineation of the coastline to improve accuracy of kelp mapping by understanding where kelp cannot grow, above the low tide line. However, a comprehensive map of British Columbia’s coast at low tide does not exist at sufficient resolution and accuracy for kelp mapping with Sentinel-2 satellite data.
My research aims to develop an automated software method for coastline extraction from Sentinel-2 satellite imagery by leveraging global tide models and image processing algorithms including deep learning. Although numerous studies have evaluated methods for coastline extraction, the vast majority of work has focused on uniform, sandy beaches at local and regional scales. How these methods perform when applied to heterogeneous and geomorphologically complex coastlines such as those found in BC, is unknown. This work will improve the accuracy of the SKeMa tool by reducing false-positive kelp detections over spectrally similar land cover types such as intertidal green algae. Further, the coastline extraction method developed can be applied broadly to other coastal ecosystem mapping projects such as eelgrass restoration or monitoring sea level rise.
Background:
My professional background is in the Canadian space industry, most recently as a software developer at Rocket Lab in Toronto, ON. My role focused on designing test facilities and ground software applications for satellite ADCS components (reactions wheels and star trackers) in a production and R&D context. I also completed an internship at GHGSat in Montreal, QC, where I focused on optimizing software processes to convert raw multispectral satellite and aircraft data into observations of point-source methane emissions worldwide.
In my free time I am most likely train running, knitting, or generally spending time outside with my dog.
Expertise and Experience:
satellite remote sensing, software development, engineering, physics, Sentinel-2, tide modelling
