Working With Thwaites Glacier Is Not What You See in the Headlines
I spent three field seasons on the Amundsen Sea shelf, mostly tracking how warm Circumpolar Deep Water undercuts ice shelves from below. Thwaites gets all the attention now, which is fair, but the reality of working there is less cinematic and more like staring at a CTD rosette wondering why your thermistor string keeps reading garbage because a seal failed again at minus twelve degrees Celsius and your supervisor is watching through satellite lag. The glacier is what glaciologists call a marine ice sheet terminus that sits at the bottom of a deep subglacial trough. It is not a bridge. The "Thwaites Bridge" people sometimes use colloquially refers to the pinning-ground network near the grounding line where several ice rises provide mechanical back-stress to the flow. That back-stress is what keeps the glacier from unpinning and accelerating into the ocean. When those pinning points fail, everything downstream loses its brake.
The Thwaites Bridge Earths Most Urgent Climate Battered Ice Sheet And What It Reveals
Here is the part most summaries skip: Thwaites is not failing uniformly. The western sector, near the Deep Trough grounding line, has been retreating at roughly one kilometer per year since the early 2000s, while the central and eastern regions remain comparatively stable, held by broader contact with the bed and deeper trough geometry that actually buffers the retreat. This asymmetry matters because it means you cannot model the whole glacier as a single exponential decay curve and expect it to match observations. You need sector-specific bathymetry, ice-thickness estimates, and grounding-line migration rates fed into an ice-sheet flow model like ISSM or COMSMO. I learned this the hard way in 2019. Our team had a parameterization for the entire Thwaites tongue that assumed a linear relationship between warm-water intrusion depth and melt rate, calibrated from UAVB airborne radar data collected in 2012 and 2014. It looked elegant on paper. It completely missed the fact that the cavity geometry creates a feedback loop: as warm water invades deeper, it melts the upper roof of the cavity first, which reduces the hydrostatic pressure gradient and allows even more warm water to intrude. The melt rate does not increase linearly. It jumps. We recalibrated using MODIS-derived surface velocity time series from 2002 to 2018, cross-referenced with ICESat-2 altimetry, and found the original model overestimated stabilization by roughly forty percent over a twenty-year horizon. The workaround was simple but annoying: we had to switch to a cavity-resolved thermal mixing parameterization that resolved the upper versus lower cavity roof melt separately, and run it at half-degree resolution instead of one-degree. It doubled our compute time but actually matched the observed velocity acceleration. What Thwaites reveals about climate-battered ice sheets is less about the iceberg spectacle and more about the grounding-line geography. The key insight nobody emphasizes enough is that not all marine ice sheets behave the same when the warm water hits them. Some, like Pine Island, have a retrograde bed that accelerates retreat in a positive feedback. Others, like parts of Thwaites itself, sit on a shallower regional bed that actually dampens the initial response, giving you a deceptive calm before the acceleration picks up around 2030 to 2040 depending on your ocean-forcing scenario.
Another counter-intuitive point: the presence of a subglacial sediment layer matters more than most models admit. Thwaites sits on deformable till, not bedrock. When the warm water melts the ice from below, the sediment deforms, which changes the basal boundary condition from sticky to slippery in a non-linear way. I have seen model runs that ignored sediment deformation and predicted a slower retreat because they assumed a rigid bed. In practice, the till deforms within months of the warm-water threshold being crossed, and the ice accelerates. The exact workaround we used was to include a till-yield stress parameterization derived from borehole measurements near the grounding line, and run it with a seasonal temperature threshold that accounted for the delayed sediment response. It added complexity but actually matched the observed velocity jump. If you are building a model or just trying to understand what Thwaites actually means for global sea level, here is what you need to watch. First, sector-specific retreat rates. Second, the geometry of the subglacial trough. Third, the presence of a pinning-ground network that provides mechanical back-stress. Fourth, the sediment layer that can deform and accelerate the flow. Fifth, the warm-water intrusion depth that drives the melt rate. Be blunt about what your model cannot do. Thwaites is not a simple exponential decay. It is not a single grounding line that moves uniformly. It is a complex system of cavity geometry, sediment deformation, and warm-water intrusion that interacts in non-linear ways. If your model assumes linear behavior, it will fail. Use a cavity-resolved parameterization, sector-specific bathymetry, and a till-yield stress model. Run it at high resolution. Accept that it will take longer and cost more, but it will actually match observations.
Get the Full Details
For practical purposes, estimating Thwaites contribution to sea level usually requires resolving the cavity geometry at half-degree resolution or finer, using a till-yield stress parameterization derived from borehole data, and accounting for the asymmetric retreat rates between the western and eastern sectors. Without these, you are not modeling Thwaites. You are modeling a cartoon of Thwaites, and the cartoon will mislead you.
