Wei Ji Leong
"When asked if I wanted to do a PhD at the Antarctic Research Centre, I literally jumped at the chance!", says Wei Ji Leong.
Self described as Malaysian but brought up in Brunei, Wei Ji still finds it crazy sometimes that someone from the tropics gets to work on studying ice.
"Antarctica's probably too cold for me, so I’ll settle for the next best thing - satellites".
Wei Ji comes from a Remote Sensing/Geographic Information Systems (GIS) background, having completed his undergraduate studies from Victoria University of Wellington back in 2014.
"I fondly remember my first 'ice' related project - automatically mapping the expansion of proglacial lakes in the Southern Alps over the last decade using Landsat imagery. Back then we called it 'supervised classification', whereas now it would have the fancy label of 'machine learning'", says Wei Ji.
That project was also how he got to know his current supervisor Huw Horgan, who actually supervised his BSc(Hons) back in 2015 on mapping potential subglacial lakes around the Antarctic grounding zone using ICESAT laser altimetry data. Wei Ji then worked in the GIS industry for about two years, sharpening up his computer programming and data management skills that is now proving useful for his next big challenge - 'deep learning'.
Wei Ji explains, "So you get all these different datasets from different satellites coming in, optical images, radar, laser, gravity, and so on. Each of them tells you something specific about what's happening to the ice, but what if there's a way to combine all of that information together and can tell you a different story, what if 1+1+1 doesn't equal 3?". Well, that's where the 'deep' in 'deep learning' comes in".
For his PhD, Wei Ji is interested in using deep learning to map deep hidden subglacial lakes around Antarctica.
"Say you have one satellite measuring the height of the ice going up or down, and another tracking how the ice is flowing faster or slower. Individually, you might just think of the ice as responding to some change in snowfall or melt, but if you put them together, you might realize that it could be something else. Water could be filling up or draining from subglacial lakes that are causing the height changes and the speedup or slowdown events! Add in more datasets into the mix, things like subglacial heat flux, ice penetrating radar observations, and we'll start to get a better sense of whether there might be water underlying that ice", says Wei Ji
It's still a lot harder than it looks because you can't directly 'see' the subglacial lakes, but armed with a burgeoning repository of data, an aptitude for programming and access to state of the art servers at the Antarctic Research Centre, Wei Ji believes he's up for the task.
Since starting his PhD, Wei Ji has attended the International Summer School in Glaciology at McCarthy, Alaska, and presented some of his early work at EGU General Assembly in Vienna, Austria as well as the IGS Radioglaciology Symposium in Stanford, California.
We Ji says, "It's one thing to work with the data, but it's another thing entirely to listen to the many glaciologists out there collecting the data, using them in ice sheet models, knowing what works and what doesn't. The Antarctic community is a really open and international, there's all sorts of enthusiastic people from so many different backgrounds, it's amazing how everyone contributes to make it all work!"
"I guess the one thing that ties all of us together is the sea. Water beneath the ice affects the rate at which glaciers slide into the sea, and my research into mapping subglacial water is part of the big picture question on how much and how fast Antarctica can contribute to sea level rise."