First, I just wanted to say that I really enjoyed the behind the scenes aquarium tour Monday and that I think we did a great job with the BBQ! Thank you to everyone who put in a lot of time and energy to make it what it was. This week I have been working on annotating pulse 66, which is the last pulse I will be doing for my research. I plan on finishing pulse 66 by the end of the week and will finally be able to compare it to the other two pulses. Once I do that I can start playing with the dataset. I am excited to have my mentor teach me about a few programs, of which I have yet to be educated on, that will allow me to run stats and see patterns within the data that I was unable to before. Learning how to make use of these programs will be greatly beneficial in my future graduate studies. After I analyze the data I will be able to start focusing on writing my paper. As usual, this week has flown by! Hope everyone is having a great week.

This week was spent fabricating a fixture in the MBARI Model Shop to allow for more accurate IMU sensor calibration as well as rerunning the accelerometer sensitivity tests with the linear translation table. Before being allowed to operate the milling machine in the Model Shop, I met with Mike Parker for safety training and a proficiency test. I have operated Bridgeport milling machines with CNC control at the UC Davis engineering fabrication lab in the past and was very familiar with the equipment. Once authorized to use the mill, I designed a simple 6-sided fixture using a scrap piece of 3" x 4" 6061-T6 aluminum rectangular extruded tube that I found in the scrap bin. This simple design would allow the IMU to be mounted inside the aluminum "frame" and then be indexed on all six faces. This is required to setup the accelerometer and magnetometer calibration offsets to ensure the best performance of the Kalman filter which fusses the accelerometer, gyroscope, and magnetometer data together to produce an estimate of the sensor orientation relative to an inertial reference frame (the ground).

With the fixture fabricated, I was able to finish the IMU calibration process and read the calibration offset values from the sensor's memory registers. Unfortunately, these calibration offsets are stored in the sensor's temporary memory and are erased when the sensor loses power. Thus, the calibration process must be completed every time the sensor power is cycled on/off. To avoid this problem, it is possible to manually overwrite these calibration offsets to the memory register every during the initialization process right after a reboot. This ensures that even if the sensor temporarily loses power and reboots during ocean deployment, the correct calibration constants will be manually written to memory and the output sensor values will be as accurate as possible.

With this process complete, the rest of the week will be spent rerunning the acceleration tests that I ran two weeks ago using the linear translation table having accelerations ranging from 1 mm/s/s - 100 mm/s/s. During the previous tests, the sensor was not fully calibrated. For this next round of tests I hope to see less noise in data and compare the raw sensor outputs to that of the fusion outputs from the Kalman filter.

Bonus 01: we had the intern BBQ this week on Wednesday and it was a big success (even if we were 30 minutes late). Special thanks to BBQ master Drew for helping setup the grill and managing all the details during the prep and setup. I was really impressed with the team and how everyone pulled together to make it all happen. Kudos to all my fellow interns!

Bonus 02: I went out on the Paragon this week with engineer Brent Jones and Jared Figurski. We successfully deployed Brent's coastal profiling float on Tuesday morning. After deployment, the CPF performed basic diving tests and Brent said all systems looked good. To float was collected the following day.

Goals for next week:
a. test various signal processing schemes/filters on new data set with Matlab Signal Processing Toolbox
b. identify the lowest possible acceleration that can be reliably detected by BNO055
c. perform overnight test with sensor and produce Allan Variance plot to characterize noise profile of gyro and accel
d. switch to the GHI Lemur board and fabricate a new circuit board with integrated IMU and battery

Bridgeport Milling Machine

Coastal Profiling Float

Intern BBQ 2016 - setup

We've been having some success with mid-water, but the results haven't been as successful as the benthic observations. I noticed one of the specimens we were wanting to get images of was collected and studied in 2012 by MBARI. After checking in VARS I found him popping up a bunch around that time. Instead of trying to catch images of him as we drove by, what would they look like if we used the videos from collection of this particular specimen? Well, they look amazing! we're getting some high quality images from a lot of different depths of field and coloration. Unfortunately, we have been using tracking to get better images of him, which helps to isolate him from the marine snow as we drive by. We should have turned it off for these videos because I now have somewhere around 4,000 pictures of this one squid. Its ok, but we want to use lots of pictures of chiroteuthis for the convNet and not a bunch of pictures of just Fred here. Considering chiroteuthis is not an abundant animal in our archive, I think we'll have to use what we can get. Either way its very promising. I have a supervised training hypothesis that I'd like to try out on Peniagone. I've been reassured by my team that its a really good idea, but probably won't work... so you're saying theres a chance!?

This was a pretty fun week! The aquarium tour was awesome, it was great to get a glimpse of what happens behind its tanked walls. Also, the BBQ was a total hit! Many MBARIans have expressed to me their sincere appreciation of the food, so nice work team!

Apart from intern shenanigans, I'm getting light!!! Not from the doliolids (_), but from some radiolarians. This is great because I can at least start characterizing the radiolarian photoproteins. Today I am running a regeneration experiment on radiolarian and doliolid samples to 1. determine if the photoproteins from the doliolids were already conformationally changed when I was testing them and 2. regenerate "spent" photoprotein from the tested radiolarians (i.e. those that were conformationally changed and released light when Ca was added). If this works (fingers-crossed) I will have much more material to work with.

Pretty awesome week so far if I do say so maself.

Kicked it off with a very special tour of the aquarium, and even though i've been there a thousand times its always great to see something new. Getting to see the TRCC was a first for me, and definitely my favorite aquarium experience so far.

Next came the BBQ, which dominated my productivity this week, but I think it payed off with a delicious and enjoyable experience for everyone involved. It was great to see you all come together to accomplish something, and even though its just a BBQ, I think its a formative experience for intern groups every year, to work hard on something together. I challenge you now to attack your looming final presentations with the same intensity and teamwork. We are all in this together, so put together practice talks get feedback from your fellow interns, and use this week to pivot into the final month with the same spirit, and community that we developed with the BBQ this week.

Finally on to the science I've been working on this week. I have to submit an abstract for my thesis work to the Eastern Pacific Ocean Conference by tomorrow so i will be wrapped up in thesis-land, for most of the morning today. As far as my mbari project goes, i've finally been able to run the rotary analysis on my data sets, and now with the harmonic analysis, I am getting a clearer picture of the fluid environment around the rover. I am now starting to put together my final powerpoint (as I encourage you to start doing) so that I have an idea of what visuals I need to produce, and still have time to make them look nice.

So we started out strong this week, loaded out the rest of our gear on Monday and made some final pre-cruise touches to the fitting program, and everything was set to go out.

Launched on Tuesday and dove to 1500m, collecting samples every 100m or so on the way up until we had to recover after dinner, getting 6 different jellies and a few spectra from each. We were getting data consistent with what we found on the Carson, which is encouraging given how strange it is. Still don't fully have an explanation for how jellies make their jel, but it seems that depth (or actually temperature and pressure, Peter Brewer would kill me if I tried to claim depth dependence) affects how they configure their water molecules. Peter suspects they have a very clever way of manipulating the water so that it's semi-solid but still flexible at high pressure and low temperature, which would allow them to move down there.

Wednesday started early, diving at 6:30 down to 2800m. We found one jelly down there, a small one, and shot a few spectra. This time, the data was not entirely consistent with the rest of our in situ jelly spectra. It appeared that this species has (at least) one more way the water hydrogen bonds can stretch, but we suspect this is just a species variation, rather than a hole in our previous data. We also got a sample spectrum of our jello on deck, and then again down at depth. After correcting for the absurd amount of fluorescence the jello has, we discovered that bovine gelatin doesn't really do anything to the water molecules themselves, but constrains the entirety in a solid. Basically, jello looks just like liquid water when shot with a laser. Which is particularly interesting then for the jellies, since it adds to our suspicion that whatever jellies are doing is exceedingly clever, to maintain flexibility of the jel even in extreme conditions.

We then came up to about 2600m and captured another jelly, but before we could collect any data one of the pilots noticed that the cable housing our fiber optics cables had slipped. We now had bare cables supporting the weight of the housing, and decided to recover instead of keep working. On the way up, both the collection and excitation fibers broke (and we suspect the spares too, although have not actually checked). So we came in last night and grabbed the spare cable, hoping to be able to head out this morning. Instead, the epoxy holding the ceramic cover for the cables had gone bad (amazing what 13 years of sitting in a cardboard box can do!) and broke off in the connection port to the laser. Now we're hoping to get the fibers reterminated by tonight and head out tomorrow morning for a shorter dive day.

In the meantime, we have plenty of data to fit curves to, so I get to have a nice relaxing day of making small changes to fit parameters and hoping it doesn't mess everything up.

This week has been a good step in the right direction! I was feeling a little lost on where to go next with my project, but I met with my mentors after the aquarium tour on Monday and figured out where to focus my efforts, so I feel like I have a place to go now. What I've most realized is that bad results are still results- I haven't succeeded in making an effective PAMGuard classifier, but I'm coming to believe that this is in large part the fault of PAMGuard's approach to click detection and classification. I came into this project knowing that there will never be a perfect classifier, just because life is messy and nothing ever fits into the perfect classifications that theoretically should be right- biologists can't really just assume everything is a perfect sphere.

I'm having a lot of issues with Risso's off-axis clicks looking like beaked whale clicks, which led to an interesting train of thought about acoustic social communication (aka humpback whale songs) versus odontocete echolocation clicks, which are what I'm looking at. With social communication there is clearly an evolutionary drive to differentiate a species' calls from other species, since the expressed goal is to communicate with members of one's own species and the individuals need to be able to tell which calls are from their own species somehow. With echolocation clicks, though, they're not trying to communicate- they're just trying to eat, and this is the most effective way for them to "see" through the murky water in their offshore habitat. There's not really incentive for them to differentiate from another species, since the signals are comparatively so short-ranged that they're not as likely to get mixed up. All this results in it being really hard for me to tell common species (Risso's dolphins, which I was fortunate enough to see whale watching this weekend! That was super cool. Now I have some context on this species I am constantly frustrated by!) from uncommon species (my dearest beaked whales). I think I'm going to try to tune my classifier to be more picky, to take only beaked whales at the risk of missing some bad beaked whale clicks, rather than be broad enough to catch all the beaked whales but also catch dolphins, rendering the data pretty useless for determining population estimates or habitat density, which is kind of the direction I'm going in. Catching every click would be more useful for behavioral information, but at this stage in the technology's capabilities I am more interested in trying to get this larger population information.

At this point I've just accepted that this method of classification that PAMGuard utilizes is never going to be very good, and I'm increasingly interested in the wavelets method and neural networking as potentially different approaches (the two seminars this week were very relevant to my train of thought!). But my project is PAMGuard, so I have to stick with it and make it as good as it can be, despite my theory that it's a little too user-friendly, aka doesn't use enough parameters, to really do a good job. A couple days ago Nick was kind enough to give me a crash-course in basic Matlab signal processing toolkit usage (and some electrical engineering, because my background is not that strong), so now I'm in business on Matlab. I was feeling a little incompetent post-crash-course just because it pointed out just how very much there is for me to learn about the technical side, so to boost my confidence I wrote a very pretty function that will take my file name and sample number and spit out the exact real-world start time of the click. Sample number to seconds could've been accomplished in three lines of code, but I really enjoyed increasing the amount of work that Matlab will do for me because I do like coding. Now I'm going to use the spectra to look at some really great treasure troves of verified Cuvier's beaked whale clicks that was sent by an MBARI consultant to see their characteristics and tune my classifier to those good examples. It's a little weird using Matlab to tune Pamguard when I can (conceivably) use Matlab to just accomplish the same thing that PAMGuard does with the tools in the signal processing toolkit, but I'm hoping it will help make PAMGuard less bad.

I have a lot to think about now, because everywhere I look there are interesting questions being posed that eventually acoustics will hopefully be able to answer. For example, there's evidence of potentially diagnostic phonemes in some of the data, which is a whole other really cool line of study, since I typically think of phonemes as something you see in social communication. Obviously there's some sort of utility in hunting or it wouldn't exist, and I'm really curious about what that could be (and how we would figure it out! Shark cams on beaked whales, anyone?). But that's out of the scope of my project. So back to Matlab and PAMGuard I go! Lots of exciting stuff this week.

I had to start over with my project (yay?) because the template that I built my story map on is still in beta version and therefore, it sucks. I literally could not customize anything and the limitations were too extreme for me to use. I started a new version, changing from the cascade template over to the journal styled story map. The visuals are slightly less appealing compared to the cascade design but in the grand scheme, it will be much better at conveying the message. When I restarted, I wanted to the content to be transferable to the CeNCOOS folks after I leave, so I had to make a separate account. However, the "free" account made it so I had to make other accounts such as youtube/picassa to store videos/photos. I wanted to avoid that as there is a chance that the story map would not work if those sites were to change/go down. MBARI has an institutional account with ESRI/ArcGIS but the online division is apparently separate. I approached Todd from IS about the problem and he informed me to call ESRI to figure it out. Guess what they told me? They can't talk to me unless Todd puts in a ticket with them... great... A phone call to Todd got me nowhere as he was too busy to help. In the end, I just went with the Youtube/Picassa system.

On a different note, I was able to get some really good photo and video content for the story from my visit to Hog Island Oyster Company and speaking with the owners. I had envisioned a picture with 1/2 of the shot underwater showing oyster lines where the farmed oysters sit and the top 1/2 of the picture showing the restaurant/farm. However, arriving at the site, it was a different story because their lines were miles away and it was high tide. So I had improvise and borrowed some oysters to do the shot. Here's a shot of what it looks like.

The great news is that I already had most of my content so I was able to finish a first draft of it yesterday. Revisions are being made today and I will be showing my work around the office to get feedback. If any of you are bored and want to look at it, let me know!

With each passing week, time seems to fly by ever faster... sigh...

Like others have expressed, Monday's tour of the aquarium was awesome and a great start to the week. In recent years, I became aware of Barbara Block's work, so visiting the tuna tanks was a real treat. I was also excited to visit the jellies husbandry room; in all my behind-the-scenes visits of the aquarium, I'd never had the chance to visit that room up until then. After the lead guy (forgive me for not remembering his name) gave the breakdown of the room and processes, I jumped on the chance to ask him a few questions that had been lingering in my mind since the Jellies Experience exhibit (e.g. the variations in sizes, numbers, and conditions of the blubber jellies was due to difficulties in breeding and captivity). Another interesting moment was watching the SORAC folks transport the surrogate otter and pup. While our guide was discussing the kelp tank, I noticed staff, dressed in the concealing ponchos and face masks, approaching the holding pen. My attention was fixated on what would happen next, so if I looked like I was staring off in la-la land during this part of the tour, now y'all know why.

As a bonus, Desmond's friend Taylor happened to be working that day, and he gave a small group of us a behind-the-scenes tour of the Tentacles exhibit. The spaces that the aquarists work in are pretty darned small, but they certainly make use of everything...the vibe is kind of like mom and pop pet store plus mad scientist. My favorite parts: peering over a holding tank of very curious juvenile stumpy cuttlefish and hi-fiving one of the giant Pacific octos!

The bulk of this week involved the Brewer cruise on the Western Flyer. We ventured out into rough waters to the jelly collecting fields. In simplest terms, jellies were captured at various depths and shot at with a Raman laser. Yep, shooting jellies with lasers, all in the name of science! Along with Matt, Peter Brewer, Ed Peltzer, and Peter Walz could probably give you a more through (and correct) explanation, but from my understanding, the laser provides chemical and molecular information of the object being targeted. Brewer's team is especially interested in the properties of water inside gelatinous species (there's some magic going on inside those jellies and ctenophores!). As the only person that wasn't coming from a strict/heavy science background, it was truly amazing to witness first hand the collaboration between the science, engineering (ROV), and ship crews and teams. I did get the opportunity to fiddle with VARS, video recording dives and making screen grabs. I also discovered that the sleepiness that I often encounter on boats is just a manifestation of seasickness. I was pretty tired for the majority of the cruise, but didn't get nauseous; I felt even sleepier when out in rough waters. ¯__/¯

Unfortunately, our cruise was cut short. During yesterday (Weds) morning's dive, a critical component of the Raman laser became detached. We returned to shore in the evening for repairs, in the hopes of heading back out today. No dice. Last I heard, the team is hoping to have repairs completed and the laser ready to go for a day trip tomorrow. I may/may not go...while it's tempting to take advantage of any opportunity to spend time on an R/V and in a control room, I still have plenty to get done with my project!

I'm looking forward to next week, as I'll get a chance to chat with an MBNMS intern who is also focused on social media and communication. Hopefully, I'll also be able to get in on the MBA's weekly communications/media/marketing meeting. I've been eager to learn how their teams collaborate in developing and executing messages and information to the public.

Busiest week of the internship so far. After a long day of prepping, last minute experimental design changes, and endless autoclaving on Monday, my partner and I arrived bright and early Tuesday morning for one of the longest days of my life. Spawning induction began at 8:00 am at which we began the protocol of temperature and chemical changes to collect our much needed gametes. Despite the assurance of our mentor on the success of the process, it was pretty nerve-wracking waiting for the abalone to spawn. But spawn they did

At this point, Aileen and I split ways. I started micropipetting sperm into eppendorf tubes with 1.5 mL of low/high pH seawater. They were exposed for 1 minute and then brought to the microscope for video. I did this one at a time for 12 samples (6 each of high/low) over the course of an hour. The protocol was a little shaky at first, as it took a few trials to get the optimal sperm concentration for video. But once I established a routine, the rest of the experiment went smoothly. I repeated the sampling process at two more time points throughout the day to also see the effect that gamete age has on sperm motility.

Now I have 72 videos of sperm that I need to analyze in ImageJ CASA, so it's time to get cracking!

Lots of fun things this week too. First awakenings, aquarium, bbq, a trip to Monterey Abalone farm (super cool), and a great seminar by John Benson on ecology of large carnivores at Hopkins.

Also, I saw a lot of interns helping out with Aileen's experiment. Big thanks to all!

Less than 4 weeks to go until we all present our work and it's definitely starting to feel like crunch time already. Although I have a pretty good idea of where my project will end up, it's still a bit stressful as my mentors continue to add new datasets to the mix. I have now added Acoustic Doppler Current Profiler (ADCP) data from the moorings to my analysis in an attempt to link the variability in the other parameters to current fluctuations since many of the PCA results for variables like temperature, salinity, and nutrients resemble each other even though conventional understanding of El Niño wouldn't necessarily predict that.
Trying to get to a place soon where I am no longer adding elements. Once I have completed all the PCA that I intend to do, the next step will be to correlate the resulting time series with different climate indices such at the Multivariate El Niño Index (MEI), the El Niño Modoki Index (EMI), North Pacific Gyre Oscillation (NPGO), and Pacific Decadal Oscillation (PDO).

In other news good work on the BBQ everyone, even people who did sign up for leadership roles really stepped up and made it a success. Thank you Drew, for helping out so much.

Also this week Alaina and I got to help fertilize some abalone eggs which was a cool (literally, it was freezing the seawater lab) break from our computer screens. Hope all the experiments go well, Aileen!

I will be first comment next week just you wait and see.

This week was actually really busy. Besides running around for the BBQ which was a success, I have been meeting with my mentor to go over more about my project. What I love about this project is that it is constantly evolving and I am always learning new ways to look at data and even new ways to write code! My mentor is super awesome too! Besides being a genius, she is super nice and she will take me through her thought processes and shows me how the data is constantly evolving when she finds a way to make her algorithms better, which is amazing to see. I so far have found some pretty interesting data and some of my figures may be in my mentor's proposal for next year! Momma I made it! Anyway, mine is always the smallest confluence, which I guess is a good thing because nothing is every really going wrong so there is never anything negative to report..Knock on wood...but I do not have any fancy pictures to post so mine never looks all that exciting but it is I promise!

Man I keep getting later and later. Next week I'm setting an alarm. I've been fairly busy this week, and not just because of the BBQ. While we weren't able to make the library that converts XML schema into scala work, so we're adjusting the scope of the project. Instead of being a more in-depth production of part of the process, I'm aiming to have a complete end-to-end prototype. Instead of turning the full XML schema into scala, I'm going to write a functional subset of the schema by hand and prove that the new language works with that limited scope. Then it is simply a matter of extending the schema to make the language more complete. Towards that end, I've been writing an end-of-the-pipeline translator that takes the abstract syntax tree (the representation of all the relevant information) and produces a string of XML code as well as rewriting sections of the abstract syntax tree and the parser itself. I'm meeting with one of the LRAUV operators later today to go over some existing mission scripts to help wrap my head around what's going on.

 In other news: BBQ was a blast. Perhaps too much watermelon? Also, despite having been to the aquarium countless times before, the tour on Monday was awesome! I've never seen any of the behind-the-scenes stuff before. 

This week was the busiest one by far! The behind the scenes at the aquarium was so cool and I feel so lucky to have opportunities like that one, also it doesn't matter how many times I go to the aquarium, I never get bored of it!
The BBQ was a total success (all the people from my lab have thanked me for the food and the great job) so congratulations to all of us for the hard work!!
About my project, this week was so intense. We made a protocol for the big experiment that we have just finished right now about measuring micromonas alive with the Accuri and fixing samples for FCM (flow-cytometry) for 24 hours. So yes, I slept here at MBARI with one of my mentors and actually it has been a great experience. I'm so so tired right now but it was worth it because I had more time with my mentor so now I know her a little bit better than before, also I realized how hard is to work on experiments like this one but how great it's when you have someone who can helps you with it, team work!!
On Wednesday we saw in the Accuri some spots that could have been micromonas eating!! But after this whole 24h experiments we're thinking that the percent of micromonas that are supposed to be eating maybe it's to low so we can't confirm that they're actually eating. We think that maybe it's just a representation of the beads sticking on micromonas, but we still need some more time to have real conclusions.
I feel so lucky to have new experiences every week!!