Atlanta is hot. I made the mistake of biking in 30+ C (~90F) weather, and I thought I was about to die. Of course, I was probably closer to death when I left to to school one day and almost got hit by a car turning left only a few hundred meters from where I’m staying. Luckily, I just very nearly avoided the injury, and it prompted me to get temporary health insurance while I am here.
Oddly enough, I still prefer biking here than London, ON. The drivers here are rude, selfish, and dangerous, but there is still an awesome biking infrastructure. Before I moved to Atlanta, nearly 3 miles of the 6.5 mile ride had a dedicated pedestrian/bike path away from the street. That has now extended to ~4 miles. Its relaxing and a lot of fun.
Of course, all this is possible because I decided to buy a cheap used bike on Facebook market. That process was easy enough, but a used bike has issues that have been a real hassle to deal with. Nevertheless, I make it work. I really don’t like busing, and I didn’t budget to drive a car down and bring my bike (I probably should have). On top of biking, I got a gym membership at GA Techs Rec center. As an alumni, it wasn’t too expensive. I’m also having a great time catching up with friends, and the summer is still just beginning (even though its already June!).
Now onto research. For those who don’t remember, II am working directly with scientists at Georgia Tech over the summer. I am learning to use models they’ve developed and adapting these for use on Titan. To recap what I am doing: the goal of this project is to study how the organic molecules on Titan will react as liquid water in melt ponds produced during impacts refreezes. This will be a helpful step in optimizing future Titan mission science by understanding the environment they intend to target. I am modeling the evolution of the freezing crater melt-ponds with organic impurities using the one-dimensional, two-phase, reactive transport model that Buffo et al. (2019) wrote for Europa, which was adapted from their work tracking sea ice (Buffo et al., 2018). I am working with Jacob, Dr. Britney Schmidt, and the rest of the GA Tech Planetary Habitability & Technology Lab Group. Through my collaboration, I’ll be modifying parameters in Jacob’s model to meet the conditions I am studying, and adding the ability to handle more organic molecules.
I’m happy to say things are coming along nicely. Once I got here, Jacob was able to help me get familiar with his code very quickly. I can run example models of the liquid water refreezing over a given amount of time and make changes as needed. I modified it to allow it to run over a set depth rather than a set period of time since I am working with crater melt ponds of a set size. The model works by solving the physics of freezing for a very small sliver of the ice-liquid interface. These increments are customized for the model to run efficiently. It essentially is modeling a meter at the ice-liquid interface. I start it 10m below the surface because the solutions break down for the extreme thermal gradients we would see at the surface-water interface. This shouldn’t really be a problem, since we can extrapolate curves to the surface once we know what the compositional gradients are like in the upper ice shell (and this works really well for other examples).
The way I am using the model is I set it to start 10m below the surface. Then the model tracks evolution of the melt water for a specific amount of time, all within a 1m system. This system is split into 1cm increments, calculating the structure and composition for each 1 cm cross section within that 1m system. You can probably imagine how smaller increments or a larger system (than 1m) would be more computationally intensive. Luckily, the scales of problems we are interested in are meters to kilometers in size, so these results will easily resolve the top-level science we want to understand.
You may be thinking, why is a 1m system relevant for changes throughout the likely 100s of meters deep melt ponds? Well, the first meter is just the first step in the process. The model is set up to extract the results of each 1cm increment that is frozen over the set increment of time (100s) and then reruns the model for 1m below the now frozen section. We can do this because as it freezes, most impurities are rejected, but it takes time for the rejected material to mix back into the system. Therefore, we can see how after a particular time step, the concentration in the liquid fraction is highest near the top of the multiphase layer. At the same time further away, the concentration is still its initial value. Right now, I have the model chemistry set to conditions for salt in seawater (~3.4%). The initial concentration is something I need to consider in my work this summer; I’ll probably look into the mass estimates we see on Titan, or we may do order of magnitude investigations (1%, 10%). That isn’t a major priority just yet. Because the composition of the liquid below the doesn’t change quickly until a lot of the material refreezes, the effect of increased concentration as it freezes likely won’t become significant until the last few meters of the system.
That brings us to another complication with the model. The system is a top down approach. It doesn’t consider freezing from the bottom. For salts, Jacob does not think it should effect the top ice to add a bottom boundary considering terrestrial examples and since salty brines are heavy. We have to consider whether this will hold true for organics. HCN, and an HCN-Water mix is less dense than water and ice, so if the bottom freezes I wonder whether buoyancy effects would drive more impurities near the top over time. We think we may have a solution for this. However, it does complicate the situation.
The buoyancy of salt is what drives it out of the mushy ice as it freezes. If the salt is replaced with a less buoyant material (HCN), most of that material will just be frozen into the ice. Remember back to my last blog post; this is similar to what the organic material was observed to do in terrestrial lake ice (Santibáñez et al., 2019). Things become interesting is when we consider the bottom layer. Basically, the bottom layer is going to act like the top layer of a salt water ocean. As it rejects material, that buoyant material will mix with the upper ocean. Therefore, as the pond freezes, the upper ice will see increased levels of HCN frozen within it. Jacob’s model does not do well with negative buoyancy effects, but this does not undermine the effectiveness of using this model because it becomes applicable to the bottom layer where it can be thought to act in the same way as mushy layer in a salt ocean. We model the lower boundary, thereby finding the amount of material rejected. Buoyancy will mix it with the ocean. As it does so, it increases the net concentration and leads to increased freezing in the top layer that can be predicted using the results from the bottom model. The beauty of using this approach is that its based in terrestrial examples of volcanism which, unlike water, is less buoyant when liquid (Worster et al., 1990). I’ll make a point to come back to that to discuss it more detail later.
That covers the major assumptions and structure of the model, and progress I’ve made so far. Regarding parameters, I’ve successfully imported most of the expected chemistry values for HCN. I had some trouble finding its diffusivity in water, but Jacob helped direct me to a source that had it. I need to put together some figures that lay out the values, their sources, and any other relevant info. I’ll be presenting an Oral presentation next week at TEPS workshop in Toronto, so I’ll try and write that stuff up in a new post the week after. As I type, I’m running a 300m model using all these. It is taking a long time, but it should produce enough information to put together a presentation for next week and set up a plan for moving forward.
In other work, I’ve been making some updates to my glacier work. It is coming along nicely. We are basically preparing the results for a final write up of all the work we have been doing.