Week 9: Balloon Mapping Part I

Introduction:

For this week in class we constructed our balloon mapping rig and put it to the test. The idea came from the Grassroots Mapping website where they have examples and tutorials of how to make a balloon mapping rig. Our activity as a class was to run a test of the Balloon Mapping by creating a rig, launching it, kiting it around campus and then georeferencing the images. We would use the images to create our own aerial image of the University of Wisconsin – Eau Claire campus. This week was used as a testing session so if anything went wrong we could correct it for the following week’s activity and suffer no setbacks.

Methods:

In preparation for this activity we had done previous planning in Week 3 of our Field Methods class. However, with more time to think about the plans we made our professor decided to make changes. It started with the rig set up for our camera. The initial design called for a cut open 2-liter soda pop bottle that housed the camera inside with the viewer looking down (see Figure 9-1) but after some thought there were changes made. The new rig consisted of a small hard foam box that had a view hole cut into the bottom of it (see Figure 9-4). This design offered more protection to the camera and we hoped that the camera would stay in its position and capture high quality photos at a perpendicular angle to the ground.

Since we started class with a rig already set up we needed to have the balloon filled with helium and the string measured out. A group was designated to measure out 400ft of string. They went out into the hallway and measured the string in 50ft increments all of the way up to 400ft. (see Figure 9-2) With marks along the string it was easy to know when to stop the balloon from rising any higher and help maintain that altitude during the photo capturing session.

Meanwhile, myself and other classmates lugged the helium tank down the elevator and outside to the storage garage alongside of Phillips Science Hall. We brought the balloon out and connected the tube to the tank and started filling up the balloon. (see Figure 9-3) The balloon was filled until it was around 5.5ft in diameter and then we tied it off. To tie off the balloon we used zip-ties and had one at the top to start the seal, then placed a rubber ring onto the end, folded it over, then placed two more zip-ties on to secure the seal. The video that was made shows the output best.

With the balloon filled up and tied off we attached the string and rig using a carabineer. The rubber ring within the tied off portion of the balloon was a handy piece to have so we could easily attach the carabineer. With the rig with the camera securely attached to the balloon we were ready to start the continuous shot mode on the camera and GPS tracker and then launch the balloon. (see Figure 9-4) With the markers on the string we released the balloon with 400ft of string and began our mapping session.

The plan was to get a good portion of campus photographed so we began to walk throughout campus mall to gather our aerial photos. Our walk took us around the campus mall, but by the time we got back to the start position we had a problem. (see Figure 9-5) The wind was strong enough to flip our camera rig around, as a result the rig ended up facing the bottom of the balloon. We soon thereafter reeled in the balloon, this wasn’t the worst thing that could happen as we also wanted to test the video camera for our HABL (High Altitude Balloon Launch) rig that we would be launching at a date yet to be determined. Once reeled in we stopped the camera and went back to add more helium to the balloon.

After more helium added we used the same method to seal off the balloon and we attached the carabineer and this time we put on our video camera rig. (see Figure 9-6) The video camera we used was a Flip Cam, which is a very versatile camera that is small, high quality and waterproof. Once again we launched in the same place and began to walk the slightly arbitrary path again. This time we extended our walk as our professor wanted to get more footage of campus. We had to dodge trees and lamp posts frequently (see Figure 9-7) and the path we chose took us on the walking bridge that crosses the Chippewa River. (see Figure 9-8) With the wind picking up and our balloon seemingly running low on helium we decided to begin drawing the balloon in, that’s when disaster struck. Our professor was a few minutes too late in telling us to reel in the balloon and suddenly the balloon tore away from the string and the rig fell to the earth. Luckily our rig landed ‘safely’ right into the river and we were able to recover it. See the Video here.

With the balloon portion done for the week we headed back inside for the analysis of our images. We had many images to sort through, and few of them were at a proper perpendicular angle to work with. Once I amassed about 15 adequate images we were required to bring them into ArcMap, Erdas Imagine or Map Knitter and mosaic the images together. Map Knitter is an online website where you can import your images and georeference them simply by overlaying them on aerial imagery, typically from Google Maps. (see Figure 9-9) The Map Knitter website is a cool way to georeference the images and works quite handily, but I preferred to use ArcMap.

The images that were imported into ArcMap have no geospatial reference so the georeferencing tool in ArcMap was used to position the images correctly in our mapping session. We had to begin with a high quality aerial image of the UWEC campus as our control image. We used the Add Control Points to find good points to georeference the image from the camera to the given control image. The process is best used when you choose locations near the border of each image as a control point. Once you have a point in each corner of the image it is then time to move the control points to other areas on the image. (see Figure 9-10) For the first of our images it was difficult to do this, as is shown on the pictures, our campus is going through some major construction. The student union building on campus is no longer there along with complete changes to a majority of our campus mall. This made it difficult to georeference for some images unless you used only the images that were already georeferenced. 

Discussion:

Though we lost the balloon at the end of the activity the day was a great success. We tested out the balloon mapping and found that we would have to find a less windy day to get the best results. The day we chose to test the winds were stronger than we anticipated. Though we reeled out 400ft of string, the balloon was always at an acute angle to the ground which gave us lower than desired resolutions for the captured images. (see Figure 9-11) To go along with the balloon not being at the designated height it was always bobbing around which made the camera rig sway back and forth fairly rapidly. The continuous shot camera generally had no issues focusing and capturing good images. Some images, though not perpendicular to the ground, were fantastic images to see such as Figure 9-12 which gives a great shot of our new student union. Later in the flight, the rig rotated within the holster and ended up looking at the bottom of the balloon since the camera rig wasn’t secured well enough for a constant downward viewing angle. The camera inside fell out of place also and ended up taking numerous shots of the inside of the hard foam capsule. (see Figure 9-13)

Creating the rig for the balloon was rather simple since we had so many groups working on different pieces. There were no issues to speak of while filling the balloon with helium, nor were there any with attaching the rig either time. The launch also went swiftly with the wind carrying the balloon out and up quickly. The group who marked up the string to show the lengths that we were releasing the balloon up into the air placed a black piece of tape at the 400ft mark so we could find the point to stop unreeling at. (seeFigure 9-14)

One problem that arose was that we believed to be a deficiency of helium in the balloon close to the end of each of our flights. The balloon wasn’t as high as we had hoped and although the wind was blowing strong we expected the buoyancy of the balloon to keep it higher than it was. When we brought it in the first time we inflated it with more helium in hopes of it flying higher than it had been.

Another problem that came up was the detachment of the balloon to the string, see the video here. As I mentioned before it was lucky that we were crossing the river so the rig had a safe place to fall to. I believe that the rig broke off because of the rubber ring we used to attach the carabineer end of the string to the balloon. It was being tugged and pulled on the entire time we were out flying the balloon and rubber isn’t usually all that durable. It certainly wasn’t the balloon ripping or popping because it took off and flew out of site in just a few minutes. (see Figure 9-15) The rig that fell to the river had the entire thing intact along with the carabineer and the rest of the string was reeled in from the river.

Once back inside we had the opportunity to view the images immediately and we soon saw what we had to work with; a few out of hundreds. As mentioned before most images were at bizarre but fantastic angles capturing images of the surrounding landscape, but few were at the angle we wanted for mosaicking. So we hoped the video footage from the Flip Cam would serve fruitful to our cause. After a short viewing session it was easy to tell there wouldn’t be much use as it was a dizzying clip of rapid movement.

After working the images in ArcMap with the georeferencing tool I was able to create a satisfactory mosaic of images to represent a small portion of campus. I did my best to georeference the sidewalks together but it was a challenge to use the full images. It was a challenge due to the fact that the edges of the images are warped due to the lack of altitude that the image was taken at. (see Figure 9-16) I attempted to remove the edges of the photos to try and obtain the most perpendicular portions of the images using a raster extraction tool. I tried extract by circle but found it to be inefficient and very time consuming. The images we were working with were not of good enough quality to spend so much time working with, but it was a good chance to practice for the major mosaicking we would be doing for the real balloon mapping activity.

Conclusion:

In conclusion this week’s activity was a fantastic precursor to what will be a great mapping opportunity. We had multiple trial and error issues that arose while conducting the test and we documented those properly to help have the best chance of success when we fly our true mapping balloon. The important things to take from the test that we conducted are that we need a better way of attaching a camera to the balloon, and we need a more suitable day weather-wise for balloon mapping. Georeferencing the images we captured was a bit painstaking but it is an awesome way to piece them together in a single viewing apparatus. I certainly hope we can find a better day to be able to balloon map as this map will be one of the first of our new campus layout at UWEC.