This past Saturday, Mansfield University hosted its first night football game since 1892, back when it hosted the first ever night football game. Some 121 years later, the university finally installed lights at its (“sprint”) football stadium and had a big party with it. To kick off the balloon mapping program at Mansfield, I planned to send up a balloon over the event to capture some aerial photos of the stadium and to do some mapping.
Luckily, I had a great crew of MU Geography & Geology students who took the reins so I could sit back and supervise.
The athletics department was hoping to acquire some imagery of this special night, but had been rebuffed by the FAA in its attempts to get a flyover of an airplane or a manned hot air balloon. However, sending an unmanned balloon to an altitude less than 1000 feet avoids any interference with the FAA, so the athletics department was very gracious to cooperate when I brought this idea to them. Using balloons for aerial photography is about as old as photography itself, but the DIY context of GIS/Geoweb in the 2010s combined with a night launch brought some special considerations. After doing some looking, I couldn’t find anyone who had used the Public Labs rig in this kind of a context before.
On Friday, September 13, the evening before the football game, we were lucky to run a brief test launch. During that launch, we discovered that:
- The camera could be convinced to take night photography of the lighted football stadium while attached to a balloon.
- Conditions, especially calm wind, would need to be about perfect to get high quality imagery in this kind of lighting context.
- Imagery comes out better when there’s not dew on (and in) the lens. Oops.
With these problems solved, and with forecasts calling for calm wind, we hoped that we could get the launch in the air and get the images we wanted without any trouble.
We left the helium in the balloon from the night before to conserve this (ever more scarce) resource. This also condensed our set-up time: after meeting at 6:00 and wading through crowds to get into the stadium, we were in the air and imaging at 6:50 pm. We specifically launched at this time so that we could 1) get aerial photography during daylight; 2) capture the lighting ceremony at 7:18 pm, and 3) keep the balloon in the air long enough to get night shots as well. Given the approximate 80 minute life of our battery, this was the prime time to achieve all of these objectives.
I had hacked the camera (a Canon Powershot A4000IS) with CHDK, and had programmed it to take photos every five seconds. This choice was representative of our battery limitations: shot-action seems more draining on the battery than simple on mode, shooting using around 1% for every ten to 15 shots. If allowed to simply shoot in continuous mode without the hack, the camera would have taken a shot every second and depleted the battery within 25 minutes. By programming it to shoot only once every five seconds, we extended battery life between three- and four-fold, allowing the temporal coverage we wanted.
Students were divided into teams of three members. One student was responsible for staffing the cord and steering the balloon, while a second student was responsible for holding and adjusting the cord on the reel. A third student accompanied the group to help navigate the other two through crowds and other obstacles. Each team ran the balloon for 10 minutes before trading off to another team. After five teams (all of the students who were interested in running the balloon), the first team got to run the show again and was a bit reluctant to give it up.
We brought the balloon down at approximately 8:18pm, having been in the air for nearly 90 minutes. By some miracle, the battery was still powered on and taking photos; however, it did finally die during the walk back as we were reviewing the shots.
We ended up with 822 pictures, of which 726 were of the balloon in the air. I suspect that the scripts I hacked caused some sort of delay with shooting, or there was a delay with shooting because of data backlog from continuous shooting mode being used for longer than designed, resulting in fewer than one picture for every five seconds. I’ve posted a Google Drive folder of all of the raw shots for the students (and anyone) to peruse. Unfortunately, these pictures ended up being only two megapixels resolution (1600×1200) when I had intended for them to be 16 megapixel. Our image quality obviously suffered because of that. I don’t know if the hack, or my clumsy fingers, are the culprit.
We came up with several shots that I sent along to the athletic department that worked well enough for single-shots, but none really stunned. This was my favorite of that bunch, mostly for its clarity rather than its coverage (click on any picture for a larger one):
In those 700+ photos from the air, maybe all but two dozen were junk (unless you’re strictly looking for trippy pictures). The main issue was the slightly longer exposure needed to capture the night shot combined with the instability of the camera in even the slightest gust. Even on a relatively calm night, at 1000′ up, there is far less friction from the surface meaning a higher possibility for stronger wind. Combine that with the fact that we’re in a valley that a 1000′ flight probably emerges out of, and we’re in a place where wind is going to be unpredictable.
We did, however, get enough shots to put together two mosaics, one during dusk:
Interactive frame at dusk, via Mapknitter.org:
Interactive frame after dark, via Mapknitter.org:
While we didn’t get any real “money shot” from the camera for PR use, we did manage to get a couple of nice mosaics that I think did well to capture the evening. In a way, we engaged in a version of “impressionist” cartography, using the balloon mapping to capture one event or mood(s) of the landscape. And, the students (mostly) seemed to enjoy mapping this way.
Most of them even stuck around to watch the balloon land:
Our photos made it to the top of the Mansfield University Facebook page. Awesome.