January – July 2022
My placement as part of the DurhamARCTIC programme took place at the International Arctic Research Center (IARC) at the University of Alaska Fairbanks (UAF). I spent a total of six months in Fairbanks from January to July 2022. During this time, I carried out both fieldwork and placement, dedicating three months overall to the placement.
During my placement, I collaborated with Dr Donna Hauser, a marine biologist at IARC, whose work involves analysing vessel traffic in the Arctic for assessing marine mammal vulnerability. My placement consisted of working with a pan-Arctic dataset of vessel Automated Identification System (AIS) data. AIS is a navigational safety feature that allows vessels to see each other’s location at sea. While not all types of vessels are required by the International Maritime Organisation (IMO) to use AIS transponders, harvesting AIS signals picked up by satellite and terrestrial stations makes it possible to get a good sense of major vessel traffic within a given region.
Dr Hauser has access to a database of AIS signals for the Arctic region since 2012. It consists of individual pings by ships as they move along their voyage tracks. During my placement, I worked independently to create an automated process for cleaning up the dataset in order to make it useable for further analysis. I did this through the use of the coding language Python, by creating a script that cleans up the dataset from any errors (for example, incorrect GPS signals that made ships appear to jump to the North Pole and back in just a few minutes!). This gave me the opportunity to brush up on my Python skills, which I am continuing to use as I carry out further geospatial analysis as part of my PhD thesis.
In addition, I expanded my existing skills in the use of the geospatial analysis software ArcGIS Pro. I used ArcGIS Pro to check the outputs of the Python script, identify cases where the script was not picking up errors in the AIS database, so that I could go back to the script to create commands for handling these specific cases. As I found myself repeating the same consecutive steps over and over to import the script’s output into ArcGIS Pro and connecting ship location points into ship tracks, I also created an automated multi-step process using ArcGIS Pro’s “Model Builder” function. This allowed me to run several geospatial tools consecutively with one click, instead of having to launch them manually and having to wait each time for the previous one to complete. My skills in using Model Builder were previously quite limited, so this gave me the opportunity to become proficient in using a tool that I had been meaning to engage with for some time. This will undoubtedly become a fundamental component of any of my future geospatial analysis work.
Historical AIS datasets are not widely available and when they are, they require extensive processing of the raw data to turn them into clean and ready-to-use geospatial datasets. To my knowledge, the script I developed during my placement is the only existing workflow for obtaining a clean and ready-to-use vessel traffic geospatial dataset from this database. As such, I felt that I made a valuable contribution to Dr Hauser’s work, and anyone else who has access to this dataset.
Finally, the placement offered me the opportunity to establish meaningful research collaborations with IARC. This led to developing connections with people that became fruitful for my fieldwork and data collection as part of my PhD thesis and opened up conversations for future joint research grants and postdoctoral opportunities.
The raw AIS database for the Bering Sea region. The individual vessel AIS pings (left) clearly contain erroneous vessel locations (for example, notice the cluster of points near the North Pole – ships do not go that far north, not even during the summer!) When the points are linked to create vessel tracks (right), the incorrect pings create an illusion that ships are jumping to the North Pole and back in just a few minutes!
The cleaned AIS database for the Bering Sea region after processing using the Python script. Vessel voyages are now more clearly visible and impossible jumps to the North Pole have been removed.
An office with a view.
Winter views of the IARC car park.