Matthew Petroff

With the kite described in my previous blog post, I set out to take aerial photos at the South Pole, particularly in the Dark Sector. My original goal was to primarily take a large number of photos for a photogrammetry reconstruction, using a custom-built pan–tilt camera rig built around a Raspberry Pi Zero 2 W and a Raspberry Pi High Quality camera; while I just barely completed the hardware before I left for Antarctica, I did not finish the software in time, and I never ended up flying the camera.¹

Aerial view of a two-story blue building on an ice sheet, with stairs in the front, a circular metal cone-shaped ground shield on the roof, and a large radio telescope connected to it. Continue reading →

1. I also had issues with reliably setting the focus, so the Raspberry Pi Camera Module 3, which has autofocus, probably would have been a better choice. The camera was also heavier than I would have liked. ↩

Since the U.S. Antarctic Program prohibits the use of remote-controlled aircraft, i.e., drones, how might one go about taking aerial photographs at the South Pole? The answer is kite aerial photography, which of course requires a kite. While my original plan was to buy an off-the-shelf parafoil kite, since they’re soft kites that can easily be packed in luggage. After attempting to buy an Into the Wind Parafoil 10, which I had determined would probably be a reasonable size for the camera rig I wanted to loft and was recommended in old forum posts for use in kite aerial photography, it turned out to be discontinued,¹ and the proprietor of the small business that I had attempted to buy it from suggested that a rigid-frame kite such as a Rokkaku kite might be a better fit for the relatively-low winds at the Pole, since parafoil kites can collapse if there’s a lull in the wind. I had not previously considered such a kite, since the commercially-available options have spars that do not collapse to a sufficiently-short length to fit in a carry-on suitcase and because I was worried that their carbon- and fiberglass-composite spars may become too brittle in the extreme cold. Although there were other potential parafoil options, I was convinced that a rigid-frame kite could be the better option, since during days with higher wind speeds at the South Pole there’s often blowing snow and poor visibility. Thus, I set out to build such a kite with components that could survive the cold and that could collapse to <17″ to fit in my luggage.

A flying navy and white hexagonal kite with a navy and white fuzzy tail with a clear blue sky in the background. Continue reading →

1. It was recommended as a replacement for a similar parafoil kite that had also been discontinued, two decades ago. ↩

In what seems to be an unusual configuration—as I was unable to find any information on how to do it—I had a need to make Git prompt for an SSH username. The repositories in question are on a shared computer, but the host for the remote copies requires individual user authentication using a password and a TOTP code. A crude solution was to set GIT_SSH_COMMAND='ssh -l $USER', but this requires the user to remember to run, e.g., USER=username git pull, although this can be helped by setting USER='Prefix_git_with_"USER=yourusername"!!!!!!!!' in .bashrc. After much trial and error, I arrived at the solution of setting the GIT_SSH environment variable in .bashrc to point at the following shell script:

#!/bin/bash
# Make Git prompt for SSH username
# M. Petroff, 2024-12
if [[ 1ドル = "-G" ]]
then
 # Handle Git's SSH variant test
 ssh "$@"
else
 # Prompt for username
 read -p "Username: " user </dev/tty
 ssh -l "$user" "$@"
fi

There are two peculiarities here to keep in mind. First, stdin and stdout cannot be used, so input redirection needs to be used with read to get input directly from the terminal (/dev/tty), similar to what SSH itself does; with the -p flag, read writes to stderr, which is okay. With this resolved, there were mysterious issues with either the username prompt being displayed twice or with needing to press “enter” to see the prompt, depending on the exact details of the then work-in-progress script. I eventually realized this was due to Git’s SSH variant checking, which first calls the GIT_SSH command with the -G argument to figure out which variant of SSH is in use. While this check can be avoided by setting GIT_SSH_VARIANT explicitly, it was easy enough to handle the check in the shell script.

Early Monday morning, my brother and I drove north from the Boston area to Maine for the total solar eclipse. Originally, we had planned to view the eclipse from New York, in the Adirondacks, but, a few days prior, the cloud forecast for Maine was looking much better.¹ The goal in either case was to climb above the tree line for the eclipse, and with clouds in the forecast for New York, the Vermont peaks being closed for mud season, and the New Hampshire high peaks being too far south, that left Maine. We considered both Mt. Abraham and Mt. Bigelow, before deciding on Mt. Bigelow for its easier access and longer period of totality.

Leaving around 4 am, we encountered no traffic until shortly before reaching our destination, the Bigelow Preserve, where we were stuck in stop-and-go traffic for 15–20 minutes; the traffic was trying to turn into the access road for the Sugarloaf ski resort, which was backed up. Due to a lack of winter road maintenance on Stratton Brook Pond Road, we parked on ME-27, where the Appalachian Trail crosses it; as it was a busy day on the trail, the parking lot was full, and we parked on the side of the road. We started hiking shortly after 9 am carrying our snowshoes but quickly decided to put them on. There appeared to be at least a foot of snow on the ground, and the trail was a bit icy at this point, with a satisfying crunching noise while walking, although the snow became much softer as the day went on and temperatures rose well above freezing. When the Appalachian Trail started to climb the mountain, which was quite steep at times, the several layers we were wearing became much too warm, and we both stripped down to just a t-shirt for the remainder of the climb.

Continue reading →

1. I was carefully checking both the ECMWF cloud forecast online and loading the NWS National Blend of Models into QGIS, with roads, borders, and the region of totality overlaid. ↩

This past austral summer, I was fortunate enough to be able to spend two months in Antarctica working on the BICEP / Keck telescopes under the United States Antarctic Program (USAP).¹ Due to limited deployment slots, I was originally on a list of alternates, who also completed the rigorous physical qualification (PQ) process, and found out that I would deploy shortly before Thanksgiving—roughly two weeks before I left—after a slot opened up on the primary deployment roster.

Large white "Geographic South Pole" sign installed in snow with person wearing red parka standing next to it and metal Geographic South Pole marker in front of it. Continue reading →

1. And, yes, I got back almost six month ago and am just writing this blog post now. ↩