Days with an update link to the update below and are bolded.
Introduction – Restructuring my research update
If you’ve been following me, you may know that I have started reviewing books that I am reading. I am really enjoying this. I have spent less time watching TV and more time finding reasons to put on an audiobook. Then I spend so much time talking about things I loved or hate in a way that really helps me learn not only what I like but to understand why. Best of all, it isn’t that big of a time commitment because I only comment a bit here and there as I read and get things to say.
Now switch over to my research blogs, and these are often weeks (months lately) apart. Then when I do write them, I spend hours putting it together, collecting my thoughts and creating a narrative out of details that often allude me. Therefore, I am going to try a new approach. I will do daily (semi maybe) updates where I create a blurb about what I did that day. I am not sure how much I am going to like this. Especially since I have a serious fear of being found out that I don’t do enough work (spoilers). What better way to make that apparent than by giving a daily account. On brighter side, hopefully I can find I am working or at least improve upon the issue.
If this idea sticks, I will still try to do the occasional event write ups of big things. Of course, maybe I’ll find it less appealing in application than in theory.
9/18/19 – Month to date
Let’s get started! I should first recap my month to date. Late August I submit my revised manuscript. Right after I attend Dragon Con; in retrospect that might deserve a blog post. There were several interesting panels I attended. during Dragon Con, I set up multiple codes to run over the 5 days of fun. That ended up crashing, and I, idiotically, never planned any time for me to check in on the code.
I returned home on the third. I spent the following days troubleshooting the code. My goal was to streamline the process. The coding takes a lot of time. I am running over 7 different initial concentrations (10ppt to 1000ppt, or 1% to 100%). The process is speed up by doing modeling 10m at a range of depths (up to 300m) rather than modeling the entire ~300m (in a centimeter resolution). Another way to speed this up for deeper depths and higher concentrations is higher time steps.
The problem is the code breaks down. I set up a loop to reduce the time steps if it reaches a critical threshold and set it up to continue through the desired depths (1m to 300m). Initially, I think it failed with the time steps while I was at Dragon Con. I won’t get into the details. It runs now, but something is happening once it moves to the next depth where it calculates the initial thermal gradient then repeats over and over without end. Rather than get this to work, I have just resigned to manually moving it to the next depth each time. I keep fixing it, resetting all 7 models, then one inevitably reaches the problem first because the lower depths and concentrations go faster. At some point, I just have to say, it is doing what I want well enough, I can’t keep resetting the entire process, even if it isn’t 100% streamlined.
I did that between the 3rd and the 9th I think, but I probably should have accomplished more in that period. The 9th or 10th I moved onto fitting the data to the fits defined by Buffo et al. in his 2019 paper. After a little guidance from Chase Chivers in our last meeting before we left, I was able to use the fittype feature in MATLAB to set it up to find the right constants for the equations used to fit. I’d fit lines to data, but never before had I known how to make unique function like this, where I need multiple constants. To be clear, I did this with old data that I am now redoing. The point is, I have the code ready for when the data is processed. The 13th I had a TA meeting, and the intermediate days I was dealing with a variety of paperwork for things from when I was away. In retrospect, I don’t like not seeing anything substantial those couple days.
Monday the 16th I spent time preparing for lab this week which I am one of the lead TAs responsible for presenting the information and grading the labs. Yesterday, the 17th I TAed a lab section, and spent a while on the phone with CRS. I also spent the last couple days researching Richard Secco and other professors for my committee. I need to stop by his office as I finally got a response from him via email. I contacted Dr. Molnar about the same thing. She was open to it but questions whether she had the right skills for my project. I haven’t responded yet because I don’t know how. This entire process is a lot to handle. I don’t know if I should just say, no I think you would work well or if I ought to dig through the variety of professors in Earth Sci and on campus. This becomes a question of my project as a whole. How much will be composed by this project? Will it grow; there are ways. In which case should be talking to chemists? If so, is that really the road I should be going down? I wish I had more to show, but I suppose this isn’t such a bad thing to be thinking about. My comps is a matter of months away.
All the while, I am still monitoring my code, keeping it going. Literally as I speak, I have to stop because a couple codes on my mac did something they shouldn’t (running deeper depths on mac and shallower depths on my pc). I’m having issue with concentrations 50% and above because the code doesn’t want to work for large time stamps, so it ends up with 10s intervals, and these 10 meter sections take hundreds of years. Return to Calendar.
9/18/19 – Day of update
I’m back! You may think I wrote this continuously, but I got seriously sidetracked about the very thing I was writing about where my code really did not want to cooperate. Honestly, these problems aren’t that common. Last issue was a week ago, and the only problem I had here was transitioning to the next depth. Nothing was lost, just a bump in the road. That took up a bit of my afternoon today. I also spent ~45 minutes introducing labs to Lab 1. This is not a regular occurrence. I introduce and grade 2 labs for the semester. All the others I just have to proctor Tuesday afternoon and Wednesday evening (two sections of the 5 or 6 for the course). The rest of my time was mostly spent writing this pretty lengthy catch up. FYI, I advise you to check out the calendar at the top which I am linking to each date as I post it. This will save you from scrolling moving forward, and if it isn’t linked, assume there wasn’t an update. Return to Calendar.
9/20/19 – Grading and Distractions
The last couple days of the week I’ve spent grading ~120 labs or so. To be honest, I had a lot of struggle staying focused on the actual grading which slowed me down. I’m actually writing this the 19th because I know tomorrow will be spent grading as well. We have the Friday research talks, and any grading I don’t finish tomorrow will be done over the weekend. So end the end here, my procrastination only leads to more stressful weekend. Luckily, there will only be 3 other times this semester I have to worry about grading (2 sets of exams and one more lab). Monday, I will begin working on the python code. It has to happen because this is a magical blog where everything I write becomes reality (or is maybe that’s a curse?). Just you wait. Return to Calendar.
9/24/19 – Finish grading, wait no, theres more.
Things have not gone as I hoped. I didn’t get grading done over the weekend, but I pushed through it yesterday, up until 2am grading, at times resting my head to regain focus. Of course, I slept in a bit today in reward then came to school to log the grades before lab. After, I had intended to finally sit down with the code. Much to my despair, the labs I graded were only about 2/3 of the whole. I realized, as I sought the pile of labs for todays lab section, that I had put those in my draw here in the lab. 3/4 the sections I was grading were on Wednesday. I graded every lab for Wednesday, a task that was far more time consuming than I imagined, and now I am left with today’s section. Not only will it be unfinished for today’s lab, I will be working into the night yet again, but I swear to god I will get it done before I go to sleep. I really hate grading. The teaching isn’t so bad, fun even, but the grading is pure hell. I’d still rather get it done in binges than spread out over the semester.
I finished grading the labs around 7, but I had to upload the grades which also included added up the scores. This took time, and I continued doing it once I got home for 2 of the other 3 labs. I still need to load the grades for the 3:30PM lab tomorrow (my lab), but I will do that tomorrow. They’re already alphabetized for ease of upload. I count this as being done though, so success. That makes 4 workdays dedicated to grading, for better or worse. Return to Calendar.
9/26/19 – Emails and getting into the code
I spent Wednesday going through emails and taking care of some administration stuff. In particular, I went though the own your future program and registered for a number of workshops over the next 6 months or so. Unfortunately, I was limited this semester by TA obligations on Tuesdays. The last half of my day was spent TAing and preparing. I finally sat down to get into the python code Thursday. I updated my Xcode and began going through the notes I made with Chase to adapt his 2D code to Titan.
This code takes the results of Jacobs model, and recognizes what concentrations will be stuck in any point depending on the conditions (initial salinity, thermal gradient) when it freezes. The model assumes a well mixed pond and no horizontal mass transfer. The changes are not major and are mostly to do with parameters. I’ve spent the day looking through the literature to find the right values.
9/30/19 – Researching parameters
I have not gotten deep into the code itself. Rather, I am updating the basic parameters to fit Titan conditions. I will go through my logic here for feedback and to reference later. I am using a depth of emplacement of 1cm, but this may change. I intend to set up a meeting with Chase to discuss this again because there are some complications with modeling the boundaries. I use a surface temperature of 94K (as opposed to 95 used by Neish et al. (2006), but I am not too concerned about that. The more problematic piece is the basal temperature.
I am unsure about the temperature at at the bottom of Titan’s upper ice shelf. Mitri and Showman (2008) have a plot (Figure 4) showing basal temperatures for given grain sizes and ammonia concentrations. They show a range of 0-15%, so I assume 5% (~middle) for a grain size of 0.1mm because this is what they used to test convection. This gives a value of ~220K. 15% concentration goes off the plot but eyeballing it suggests ~140K. Neish et al. (2006) use T=176K for the ammonia hydrate when modeling a cryovolcano. If one assumes this is from the ocean (as opposed to a melt lens), I could use this to assume the basal temperature is 176K. The purpose of the basal temperature is to establish a thermal gradient in the ice shelf. For now I am assuming 176K to be conservative. I could use some advice on what is best or other sources to check out. I looked over Nimmo and Bills (2008), Tobie et al. (2005), Grasset and Sotin (1996), and Grasset et al. (2000). In the end, Mitri and Showman (2008) seemed like the best source.
Most of my day today (30th) has been spent investigating Titan crater melt formation (O’Brien et al., 2005; Neish et al., 2006; Artemieva and Lunine, 2003). Right now I have a thickness of 250m set in the code, but I know I need better logic there. Artemieva and Lunine (2003) give a conservative estimate of 2-5% melt in a crater 10-15km in size. O’Brien et al. (2005) uses terrestrial knowledge about the change in crater melt with size to scale this up to 5-10% melt for a crater 150km in size. Their logic is that the value may differ, but the basic function that controls crater melt production (i.e. increasing with radius, R^3). A key part of Artemieva and Lunine (2003)’s process was accounting for melt loss at the rims and to the atmosphere. Bigger may produce more melt, but as the depths begin to decrease, more melt may be lost. Regardless, I am going to work on a code to produce a range of melt lens for 2%, 5%, and 10% to see how significantly this effects the melt width and depth. This shouldn’t be too difficult because I need to do this to get constraints anyway, and all I have to do is run it for different melt fractions. I’d also like to return to this to do a more in depth post on crater melt production because it is something I will need to discuss in my thesis later on.