Balloon Mapping 1

Introduction

This week the goal was to launch the helium balloon in order to obtain aerial photographs of the campus of the University of Wisconsin-Eau Claire (UWEC). Earlier in the semester we had started to plan out the event by designing the camera carage, weighing out the appropriate payload, and then testing all of the equipment as well. We planned on using our three hour class period to fill the balloon, attach the equipment, and walk around campus to get full coverage. The weather on this particular day was unseasonally cool due to fairly high winds. In the end, the wind played a very big factor in the operation and how well the images turned out. After the camera was recovered, we were to download the imagery and then georeference them to an orthoimage of campus.

Methodology

The first thing that the class did once everyone was together, was to split into different groups; each of which were assigned a different task in the hopes to get the balloon in the air as fast as possible. My group was assigned the task of manning the helium tank down the the first floor of the building and then out to the department's garage outside.

Fig. 1 - Our genius way of transporting the helium tank.

Once we had it in the garage we connected a large piece of clear plastic tubing to the nozzle on the tank and then ran the other end into the bottom of the balloon which was held secure by someone gripping the opening over the tubing while the balloon was filled.

Fig. 2 - Hooking up the plastic tubing to the helium tank.

The filling of the balloon took a good amount of time since we needed to fill it with enough helium to get the balloon's diameter to 5.5 ft. It was important not to overfill the balloon since the balloon expands the higher in the air it goes.

Fig. 3 - Balloon in early stages of being filled.

Fig. 4 - Nearing the 5.5 ft. diameter mark.

Fig. 5 - Hooking up the GPS unit and camera carriage.

Once the balloon was properly filled, we then hooked up a GPS unit to the rig as well as the camera carriage via multiple knots and carabiners. Notice the spool in Fig. 5. This acted as our kite spool, if you will, on which colored marks labeled 50 ft. increments so we knew how high the balloon was.

Fig. 6 - Balloon gaining some altitude as it ascends into the sky.

The camera carriage was a styrofoam worm bucket which was retrofitted to accomodate the digital camera. The camera rig was tied onto the main string a foot or two below the balloon in order to be able to swing freely and obtain images without a string in them.

Fig. 7 - Balloon rig taking some abuse from the wind.

As the balloon got higher in the sky, the wind was blowing it off of its vertical axis. This meant that a large amount of string was being let out and not going straight up. This made it hard to maneuver with the balloon as it was blowing horizontally across campus. The shot in Fig. 7 depicts the balloon around 100 ft. according to the laser range finder we were using. We managed to get around the majority of lower campus and then decided to go across the river via a walkbridge. This is where trouble started. The wind was beating on the balloon so bad it looked like a pancake flopping around. Eventually the string close to the balloon snapped sending the balloon higher into the sky while the camera rig came crashing down into the river. Luckily, it floated close enough to shore where our Professor could snag it with a stick. The good news was that the camera was still working!

Fig. 8 - Pulling the camera rig out of the river.

Fig. 9 - Gathering around the recovered camera rig.

A variety of software was available to us to use to georeference the images: MapKnitter, ArcMap, and Erdas Imagine. An orthoimage of campus was put into the class folder to serve as the reference image. I chose to use MapKnitter since I had never used it before and figured I would try it out since it looked like it was a quick, easy way to georeference.

Results/Discussion

After downloading all of the imagery (over 2,000 images), we soon realized that it would be a difficult task to georeference them and make a quality mosaicked image from them since the shots were very angled from the high winds. The MapKnitter software was very easy to figure out and seemed like a very easy method of georeferencing. The issues I have with the software is that since you're not collecting ground control points (gcp's), I don't think the output is very accurate. Also, I do not know how accurate the reference image is that they provide. I much rather prefer using ArcMap or Erdas for georeferencing since they are desinged to do a more thorough job of georeferencing. The output image I got from MapKnitter isn't terrible considering the quality of the images I had to work with. The bad angles and differing altitudes made it very difficult to get any consistency since the scales varied so much. There is no way that we could have mosaicked the whole campus with these images. Notice how in Fig. 10 some of the sidewalks do not properly line up. With the imagery available it is not the worst thing ever though.

Fig. 10 - Resulting mosaic from MapKnitter.

Conclusion

We learned today that wind is the enemy when trying to conduct balloon mapping. For this reason, the available imagery gathered is not very useful to do any sort of georeferencing or mosaicking. This is rather unfortunate since we put a lot of work into this. However, we were pleased with how long the balloon lasted and that our rig was not a failure. Everything would have worked perfectly had it not been for the wind. Next week we will be launching another balloon with a different rig setup to see if we can get more stability out of it. Hopefully the conditions are more favorable next time so that we can get some high quality imagery to put together.