2009-05-22
As you’ve read in a previous post, or perhaps on VE3XGD’s blog, I assisted in a balloon chase recently. VE3XGD and I were the chase team that recovered the first balloon. We are experienced geocachers, and that experience really helped when we were looking for the balloon on the ground.
What I learned on this chase is that there is a bit of wisdom and experience that can be passed on to other people who may wish to chase balloons in the future.
First off, it’s important to know what GPS accuracy actually means. There are a number of factors that can affect how close the position calculated by the GPS receiver actually is to the true location on the Earth’s surface. Among these factors are:
Indeed, if you have a high-end GPS receiver, with a clear sky, a view to the horizon, and good coverage of the WAAS system, you can get 2m accuracy out of your GPSr. The number displayed on your GPSr for accuracy is an estimate of the “Circular Error Probable” or CEP. That number means that there is a 50% chance you are within that distance of the position displayed on the GPSr. If you multiply that number by 2.5, you come up with an approximation of the second diamater Root Mean Square error. Basically that means there’s a 95% chance that you are within that distance of the position displayed on the GPSr.
Thus, if you are showing a position of N 45° 0.00″ and W 75° 0.00″ with an estimated error of 5m, there is a 50% chance you are, in fact, within 5m of that position, and a 95% chance that you are within 12.5m of that position.
Now consider the display itself. At best, you can expect +/- 1 in the last digit for accuracy (that’s probably being kind). If you are using dd mm.mmm (degrees and decimal minutes to 3 places), that’s accuracy of 0.001 minutes added to the CEP error. That works out to 1.8m of latitude, and at my latitude, it’s 1.3m of longitude. If you are using only 2 digits of decimal minutes (like APRS from this balloon was), the accuracy is at least 10x worse (18m and 13m respectively).
Well, if you’re looking for coordinates set by someone else, when they measured the position, they have all those accumulated errors. Then you go look for their point, and you get all YOUR accumulated errors too. Your errors add to the position errors from the person who set the waypoint you’re looking for.
The balloon we were chasing was sending position updates to 2 decimal places on the minutes. That’s 18m of accuracy, plus whatever other errors. It was a rainy, miserable day when we went out, and the balloon had landed in a field, but in a generally forested area near a big reflective building. My own GPS was reporting estimated errors of 15m or more, depending on whether I was under the trees or not. It’s fair to estimate, then, that the balloon on the ground would be in a similar situation.
Therefore the balloon position is a CEP of 15m (for the GPSr) plus 18m (accuracy of the numbers transmitted), for 33m. The 2dRMS circle is 2.5x that or 82m. And that’s just the balloon position.
When I show up, I have a 15m accuracy from the GPSr, and I can work in 3 decimal places, so there’s an extra 1.8m I have to take into account. That all gets added to the error for the balloon. So I am searching with an error of 33m (balloon) + 17m (my GPSr) or 50m. There is a 50% chance that the balloon is within 50m of what I see as “Ground Zero” on my GPSr, and a 95% chance the balloon is within 125m.
As you can see, it’s not necessarily so. In the geocaching game, you get the coordinates to 3 decimal places, but that can still leave you a huge search circle. For this balloon, we only had 2 decimal places and an effective search circle that was bigger than two football fields and mostly tree-covered. Suddenly you wish the balloon could signal with a flare or an air horn…
After searching the 50m circle quite extensively for nearly an hour, we did not find the balloon – even though there were at least three teams of searchers. In this case, our 50% chance failed… the balloon was likely in the wider, 95% circle.
I went back and talked with one of the other searchers (VE3JGL) who had seen the balloon come down. He had a direction for the balloon, so I used a geocaching technique: walk the line. I knew roughly how far, and approximately what direction. There’s a note here about estimating the distance of a falling object: If you see something fall from the sky, unless you see it actually hit the ground, it’s probably a lot farther away than you think. With no reference for size of a falling object in the sky, it is REALLY hard for a human to estimate how far away an object in the air actually is. The important thing is to accurately guage the DIRECTION in which the object fell.
VE3XGD and I determined there was a good chance that the balloon had landed on the roof of a nearby building based on this. In the process of looking for a good vantage point to see on the roof, we discovered the balloon near the building in a little field.
Here is an aerial view (click to embiggen):
[image lost to the ravages of time and database failure]
As you can see, ground zero was 107m from the APRS transmitted location… inside the estimated 95% circle, but well outside the CEP circle. I guess it wasn’t a good day to buy lottery tickets since we failed our 50% chance. Oddly enough, there was sufficient tall grass around that little spot of field that even though I personally had walked around the building once already, I did not see the balloon package.
I marked the wall and glass from VE3XGD’s blog post. The pile of broken glass was about 3 m tall. This goes to illustrate another important point:
When someone has a sign up that says “Danger, do not tresspass here” they probably mean it. The company that owns this land did have such warnings up. That’s something else to think about, especially if you’re doing this sort of thing with kids.
You can read about the Lanark Space Agency’s balloon chases on their site.
You can track the travels of this balloon here.
2016-06-08